2018 Portfolio Knowledge Share

Table of Contents

Thanks, and Methodology 2

Traffic Trends 3
Marketing Insights 5
Marketing Providers 10
Product Insights 17
Product Providers 24
BI Insights 31
BI Providers 33
IT Insights 41
IT Providers 46

1
Thanks, and Methodology
After last years report, we got a lot of feedback about what you wanted and needed to make this report
create more value. As a result, we have created the new format you’ll find below. On top of that we took
your insights and challenges and discussed them with some leaders to try to share peers advice in a
simple easy to digest format. In each of the four sections you’ll find insights based on your responses to
this year’s greatest challenges. After the insights, you’ll find provider results broken down into 3 sections,
a summary “TL;DR” (Too Long; Didn’t Read) a “What you Told Us”, and “Our Conclusion”. This was done
via consolidating your feedback, weighting it based on teams stages and needs, and then reaching out to
individuals at a number of companies to validate the results. What you’ll see now is the direct feedback
that’s been anonymized and then a weighted and validated ranking of the tools for each category.

With this methodology we leaned on your expertise and time to create a better report. For that, we want
to thank you all for your time and participation because this report is only possible with your support and
help.

If you have any doubts, concerns, feedback, improvements or further questions, please don’t hesitate to
reach out.

On behalf of Kaszek Ventures and the Growth Team,

Thank You

Andy Young
VP Product and Growth

2
Traffic Trends
Last year we saw total organic traffic of the portfolio increase by 38% which outpaces statistics we could
find for total Latin America traffic. Additionally, the average company grew 191% Year-over-Year (YoY)
with median growth of 63%. This year saw a significant jump in organic App growth, as it was the biggest
needle mover.

When looking at total mobile traffic we saw 66% YoY growth, compared to 2016’s 75% YoY growth.
Although, this was still the largest growth driver, this deceleration in Mobile YoY growth is the first we’ve
seen, a signal that mobile is getting closer to its peak or giving way to something new (potentially voice)
which is reflected in the KPCB annual report.

App traffic growth was significant (228% YoY growth) and was driven by the companies who are
differentiating their app and mobile web traffic. This trend was key in the majority of apps who saw over
100% YoY growth. We also saw engagement time jump which reflects Brazil in the Internet Trends 2017
report showing a tripling of time in apps in 2017.

One striking behavior is desktop traffic decreased by 4%. This primarily occurred as older companies saw
their user bases shifting to mobile, while younger companies continued to see mobile growth. There
were no companies with total traffic drops, just shifts.

The mobile web share had an average of 50% through 2017, with a constant upward trend. The annual
breakdown was 49% Mobile, 48% desktop, and 3% tablet. During Q4 of 2017 the mobile web traffic
share hit 61%, with a December peak of 66% which blows past the old records. However, traditionally
this number drops back down during Q1 of the following year.

3
The largest channel of the total traffic during 2017 remained organic search, owing in part to strong SEO
efforts as well as significant brand growth. Paid search closed the gap in total traffic this year. Paid traffic
saw an increase of 52%, primarily generated via paid mobile acquisition. Display did not see similar
growth as many companies discovered the ROI was poor compared to paid search. The minimal growth is
heavily influenced by a number of companies who turned off all Display spending in Q4.

The increase on direct traffic was 35%, generated mainly by an increase in Direct App Traffic accounting
for 27% of this growth. In terms of relative growth social made the biggest relative jump. With an
increase of 66% YoY, generated by a trend of increasing investment and expertise, especially on
Facebook.

4
Marketing Insights
For marketing, we’ve broken things down first into the insights you have shared with us in person and in
the survey. We wanted to tackled the biggest problems, move to how you plan and budget, where you
spend, and then finally into the tools. If you’re just looking for ratings, skip ahead.

The biggest challenges this year have framed the order and responses for the rest of the report. We saw
3 huge issues that stood far and above everything else, scaling the business, balancing investment
(mobile/desktop), and improving attribution models. We’ve done our best to focus on tools that scale in
marketing and effective investment in the marketing section and will dive into attribution more in BI.

The other big challenges that we covered lightly are teams (we focused on IT) and offline (something we’ll
focus on in 2018).

If you’d like to dive deeper into any topic, please don’t hesitate to let us know as we try to run 6-8
monthly marketing meetings at any given time.

Marketing Budget Allocation

We have divided the budget allocation into three main buckets: pure performance, bottom up, and top
down. All three approaches are valid, but the decision tends to be driven by a few factors including:
product road map, leadership team, team makeup/seniority, business model, and attribution (often
linked to lead time).

In the first type, the performance driven approach, the main characteristic is that there is only a roughly
defined budget or no defined budget (by channel, allocation, etc. normally there is still a maximum
budget). You can invest as much as you need, as long as you meet your ROI/CAC and Revenue/Conversion
targets. As well as you can shift the investment from channel to channel freely in order to meet those
targets. This type of budget allocation is most common and successful in companies that advertise mainly
in online media and have more immediate measurable revenue. In this type of allocation, it’s critical to
have an appropriate attribution model and to understand the dynamics between the top and the bottom
of the funnel. We see more success is established companies. This has not been as successful for
companies who are adding major new products because the ROI is almost always greater for existing
products and according to the formula, little to no money is added to these new channels/products.
Additionally, this approach didn’t see great success with companies that have long funnels with
challenging attribution (multiple visits or days to complete).

The second type is the bottom up approach and the differences with the previous type are that it has a
strictly defined budget and has estimates based primarily on trends, data, and expertise. The budgets are
defined by the smallest components of the marketing team, usually advertising channels, or products to

5
collectively create a budget in order to meet your Sales/Conversion targets, using historical data. This
method is used by companies that where the dynamics within the top and the bottom of the funnel is not
clear so they can allocate a specific budget to feed the top of the funnel. It tends to work best with a
more senior marketing team who has experience with data and estimates. It is a good approach for long
funnels where the marketing team needs to have more involvement in the funnel. This has not been
successful in companies with young teams, small teams, and companies where the product is very fluid.
A marketer often feels tied to the estimates and a dynamic situation makes it hard to follow strict
budgets.

And the third type, the top down approach, where the budget is defined by estimating the cost at the
higher-level as a percentage of the total revenue from there they decide how to allocate budget
downwards. This type of budget allocation is more commonly successful when used by companies with
offline presence or with mix of online and offline presence and branding targets. It also fits teams with
strong leadership but junior teams. Finally, this can be valuable for marketplaces where the bottom up
and performance don’t balance liquidity. This has not been successful in companies with very strong and
creative marketing teams, teams maximizing for short term ROI, and companies focused on first
conversion value not LTV.

We at Kaszek do recommend a hybrid approach that takes your team and business model into account.
We’ve seen great success with top down initial estimates paired with both performance and bottom up
as secondary layer.

AdWords
When looking at SEM optimization we can see that only 29% of the respondents are using a provider or
internal tool. Aligned with last year’s results most companies are using internal tools and only 1 company
is actually using a 3rd party. On the other hand, 54% are optimizing manually or using Google Automatic
Bidding Tool.

When looking inside that 54%, we’ve noticed that the bigger and more experienced teams start to move
towards optimizing manually and based on their responses they are more satisfied with its performance
than they were with automated bidding. Volume and marketing team size then dictate CPA Bidding
followed by a small percentage of teams using eCPC. This process makes sense as manual bidding is
easier after a company has established a large keyword database, Adrank, and has historical data to help
point them in the right direction. With that being said, it is time consuming and limits the ability to mine
a large volume of new keywords if you are still building out your AdWords account.

With that being said, we expect ROAs to be a winner in 2018. Some teams are experiment with ROAS -
Automatic Bidding with interesting results in terms of performance. If you’re using a data-driven
attribution model, take some time to consider ROAs.

6
You all also shared some very interesting information about how often each task is done and what is
considered the “norm”. Clearly there is a lot of variation based on models. The one big data point we
would suggest considering is landing page testing. The majority of respondents are only thinking about
this quarterly and there have been very interesting results for companies testing it weekly or monthly.

85% of the companies are running Display Remarketing Campaigns. Here most companies are segmenting
campaigns by main variables such as Funnel, Recency, Profit Tier and Ad Size. Regarding Ad Serving most
companies use Dynamic Remarketing Ads which is an easy to use budget tool, but when talking about
Display Retargeting, personalization and CTR are the two main pillars.

60% of e-commerce use Shopping Campaigns. The optimization of the year is to reduce the bounce rate
by creating special landings for Shopping Ads. Companies also saw success by grouping campaigns by
product’s profit margin.

70% of App owners are running App Download Campaigns. As a general insight most companies are
having good CPI even allocating low budgets.

Facebook
The behavior in Facebook is similar to SEM, where 65% of companies don’t use a 3rd party provider. The
most common tool was again Smartly. 2017 also saw a new player Adphorus, the only loved solution in
the section. The list of providers is very different on Facebook than it is for AdWords as some companies
work with bidding, others with creatives, and others with both. On top of that, Facebook has seen more
vertical specific efforts unlike AdWords. In Facebook, eCommerce, SaaS, travel, and others aren’t treated
equally and there are providers for specifics.

7
Smartly.io continues to be a solid option for bid management and retargeting, they have Spanish and
Portuguese speaking staff. Refuel4 continues to be the best, and by far most expensive, platform for
creatives. And in travel, Sojern bought Adphorus as a Facebook effort in the travel space. All are very
good options that got a lot of high ratings, but none are cheap. For Facebook, we can’t recommend any
one provider but are happy to help you find the one that works for you.

With that in mind, let’s take a look at the creative side of things. The most successful campaign as rated
by companies are:

Overall B2B B2C

1. Carousel 1. Video 1. Carousel
2. Single Image 2. Carousel 2. Single Image
3. Video 3. Single Image 3. Collection
4. Collection 4. Collection 4. Video

The companies that saw the most success tested and used multiple formats regularly, avoiding over
saturation, and testing similar ideas across multiple mediums.

On top of creatives, there were a number of insights shared on Audience creation. The most common
and what we’ve seen as the most productive internally is that you need to seed lookalikes based on
known characteristics. It is more effective to build the lookalike using information you have then try to
build an audience from scratch with those attributes. However, when you don’t have enough information
or want hyper targeting some companies have seen success creating custom audiences mixing lookalikes
with interest and locations.

As additional recommendations, some companies suggested to monitor the audience saturation on the
delivery insights report since the delivery decreases sharply when you target users with the same creative
for a period longer to 7 days.

Branding Campaigns
In 2017, 40% of the respondents are running some type of Branding Campaign. Most are running offline
traditional campaigns but some are using additional new channels like influencers on Instagram or
Youtube.

The most common offline campaign type for B2B are Conferences and Exhibitions while physical locations
such as billboards or buses was the most used for B2C. Generally, companies are using more than one
type of branding campaign as reach needs to be amplified to create sustained success

8
The companies that are running TV campaigns use Agencies do the planning and booking of media as well
as the production of the creative. For the rest of the channels, everything from booking to production is
done internally.

The major challenge for most companies when it came to branding is measuring the performance of your
offline/branding efforts. From the qualitative side, companies had some success running user surveys to
understand where the user heard about the company and what was the message they remember about.
They gather data for a period of time and they look for changes in channel’s perception share as well as
messages that are better remembered. From the quantitative side, the companies running campaigns
with a high reach such as TV or “Out of Home”, they measure changes in total visits and bookings on
channels like SEO, SEM Branded and Direct. Whenever it’s possible they define a control region and they
run regional A/B test. This has been critical for the success.

Those companies running more targeted offline campaigns such as Exhibitions and Conferences measure
changes on Direct traffic 30 days after event.

9
Marketing Providers
Email Service Providers
TL;DR
ESP Usage: 95% of the portfolio use ESP Providers
IP Certification: Grew from 33% in 2016 to 50% in 2017
Our Transactional ESP selection for 2018: Mandrill
Our Marketing ESP selection for 2018: Kahuna & Mailchimp
Our IP Certification selection for 2018: Return Path

What you told us:

Although the majority of companies are using ESP providers to send both Transactional and Marketing
emails, only 37% uses the same provider for both Transactional and Marketing Emails. Salesforce was the
most common solution if you’re using a single provider, however we would recommend HubSpot of
Watson if you’re going that route.

To better understand he providers information, this is the current percentage of companies utilizing email
campaigns for the following purposes:

As such, we ‘ve split the results into transactional and marketing ESPs.

In the case of Transactional Email providers, Mandrill was the most used ESP for the second year in a row
and tied for the most loved again.

10
What is the most interesting for us is the emergence of companies using ML and AI for CRM
management. HubSpot and Watson appear best positioned to move in this direction in the short term.

Transactional Email Providers

Moving to marketing emails, the most popular responses are around Mailchimp, a solid and known
solution, and Kahuna, a new provider in the section. Mailchimp was by far the most used solution and the
most loved provider, while Kahuna received the highest average score and highest amount of “love” per
user.

Once again HubSpot and Watson were well rated in both transactional and email marketing campaign, so
we can conclude that are a good choice when looking for versatile solution in the market.
Email Marketing Providers

11
As far as IP Certification goes, no one loved a single provider, but we still needed a #1. Return Path is the
most used solution widely, having +50% of the use share followed by Mandrill and SendGrid which are
used equally. None of the 3 got amazing or terrible ratings.
Our Conclusion:

CRM
TL;DR
CRM Usage: 83% of the portfolio use CRM Systems.
Our Recommended CRM Provider for 2018: Zendesk as a standalone. HubSpot/Watson as email+CRM

What you told us:

The CRM solutions have a huge penetration in the portfolio, 83% of the companies are using a provider or
internal tool. In this category the penetration of internal tool is 17%, given in companies where knowing
the user and its history is part of the core business.
CRM Providers

The most used and loved 3rd party solutions is Zendesk. It has good reach and consistently solid
feedback. Salesforce has a high penetration and two companies love it because of it ease to integrate

12
with other platforms and tools. But on the other hand, we received more negative comments and ratings
than for any other solution. Some companies actively recommended to steer away from the it. Our
conclusion is that is a strong for complicated user journeys, which lends itself more to B2B. In fact, it
seems that companies using it in that way love it, but the rest do not. We’ll dive into that in the next
section.

What we found most interesting is the overlap between email providers to CRM with HubSpot and
Watson again. These were consistently rated great but not amazing but provide full stack solutions.

Our Conclusion:

Customer Journey
TL;DR
Customer Journey Augmentation: 48% of the portfolio use CJ Tools.
Our Recommended Customer Journey Tool for 2018: Camunda & Salesforce

What You Told Us:

48% of the respondents explicitly use customer journey tools to go along with CRM, email, and push. We
split this out because how you treat a customer and how a customer treats you aren’t always the same.
However, we understand that for many of the companies this was taken as a duplicate. With that in
mind, there are some very interesting responses.

The ranking is leaded by Camunda a solid and relatively new tool that is winning over companies quickly.
It has had a very strong year in ratings and appears poised for future success if they can stay ahead of the
curve. As mentioned in the CRM section, we also found Salesforce to be a great tool here. For
companies who use Salesforce as a customer journey tool none list it as their CRM. From some

13
discussions and looking at usage, it appears that this is where Salesforce excels, and if you’re considering
CRM and customer journey the same, maybe consider it as an option

Customer Journey Tools

Our Conclusion:

App Push Notifications

TL;DR
App Push Usage: 59% of App Owners are sending push notifications.
Internal Tool Penetration: 40% of App Owners.
Our Recommend App Push Provider for 2018: Internal if you can, otherwise Firebase

What you told us:

The most startling piece of information is that only 59% of companies with apps are sending push
notifications. Every company who is sending pushes rated them as a positive (above 5) making it a
surprise to find anyone not sending. In this category the most used solution is Internal Tool driven by the
need of customization. As for the rest of the users, the best rated provider was Firebase maintaining the

14
first position from last year. The second position was taken by Kahuna which scored the single highest
rating in this this category.

Push Notification Providers

The most common recommendation passed on by the companies is that ad copy matters, a lot! Over half
of app owners optimize its campaign by testing variations of personalized messages and testing different
Ad Copies. Many companies suggest trigger-based push notifications as a compliment to general
campaigns.

Our Conclusion:

Other Online Platforms

TL;DR
Non-Google/FB Platforms: Companies tried 6 different platforms
Satisfaction for other platforms: No platform scored in the top tier

What you told us:

15
In other platforms use, most companies are running retargeting campaigns with Criteo and RTB House.
Criteo is the only scored provider in this section and the scores reflect a poor product, as it received one
of the only 3 or below scores in the entire survey.

For those companies running both Criteo, RTB and any other retargeting campaign we suggest to validate
your attribution model. The outcome has been high frequency targeting in the first 2 days, generating
two undesired situations: 1. You could be paying a session that may happen anyway through Direct, SEO
or SEM Branded; 2. The user can click in both Criteo and RTB,
both platforms will fire each pixel and you will be paying the
same conversion twice. Companies have seen odd or no
behavior changes when deactivating one or both.

A similar situation happened with Shopback. Some

companies detected a gap between the number of leads
reported by the provider and the real amount of leads
generated. We recommend to always validate platform’s
conversions with internal conversions.

Our Conclusion:
Test diligently and be very aware of these networks. Always remove request Criteo not use Facebook if
you’re already retargeting there.

16
Product Insights
Looking back at 2017 the top challenge was to build autonomous teams inside product. This came down
to three main challenges: team structure and culture, self-prioritization, and hiring. We’ll dive into team
structure and methodology as a whole below, but on hiring the biggest piece of advice we’ve heard time
and again is that if you can find a great designer/UX, hold on to him/her! Their demand is increasing and
they should be even more challenging to find in 2018.

The second major challenge you all reported was decision making and how to become a product
company that makes the right decision. The underlying theme was “how can you be driven by data and
users effectively without slowing down”. Balancing the two with speed and effectiveness is an ongoing
theme. Hopefully the techniques and tools shared below can help better understand what’s making the
most successful teams tick and how we can all evolve.

Product Team Structure

In 2017, we saw a large push into the “Spotify-esque” squads with a focus on building multi-disciplinary
groups with a sweet spot of 5-8 team members. The majority or the portfolio describe their organization
in this way, while a handful of the larger teams have grown from squads to scale into tribes. How these
squads work and are balanced varies quite a bit. From the feedback we’ve gotten there are some near
universal truths and some that seem to be company dependent on their success.

Almost Universally Beneficial

Ensuring squads have joint technical and
Divide by Business Units or Products Having some "support" squads product leads or joint marketing and product
leads

Working well in many cases

Squads re-allocate team after Tribes as a parent of squads to Having PM/PO run multiple
Squads divided by User Journey
each mission tackle larger teams squads

Concepts that have worked in limited enviroments

Squad members choose what squad to be on Cross squad/tribe guilds OKRs at a squad level

As you can see, it’s a case by case situation, but the most important factor is finding a fit for your
business, reflective of how resources should be allotted. There isn’t a magic bullet here. One big factor

17
we’ve seen that isn’t in the chart above is hiring. The role of hiring the talent that goes into your
structure has been key. Hiring someone who is passionate about a challenge vs. being passionate about
being challenged lend themselves to very different styles of grouping squads. In the passionate person,
it’s important to surround them with data driven individuals to balance their believes, and in the case of
those who prefer to be challenged, it’s important to add UX individuals who talk to the users and ensure
the empathy is there.

Product Methodology
As for product methodology the most self-reported primary methodology is Agile, followed by Lean then
Design Thinking. These are in no way exclusive as many teams start with design thinking, use Lean
methodology for the iterations of design and UX, and then create under the banner of Agile.

What’s most interesting to us here are two things: the focus of the product teams and the comparison to
the IT.

First, Agile is known for worshipping the end user whereas Lean should worship the process. As a result,
the teams that we see focused on Agile tend to have a larger UX and CX component, and very
interestingly, some of the larger teams actually say different product areas use different methodologies.
The key here, is understanding where you stand in the process today as Agile is better for less information
and more variability (as it drives it) and Lean tends to be more effective at scale or established businesses.

Second, we saw a much larger lean towards Agile in IT than product, meaning that even in cross
functional squads (reported in both reports) there are different mentalities. We found no correlation
between the two reports, but we wanted to highlight that even in the same squads some teams feel they
can have flexible methodologies and see success. The big take away for us is that you don’t have to be
100% rigid when following a methodology.

Sharing Knowledge & Communication

An important pillar for product teams is to share the knowledge internally and across teams. Data should
be accessible for all team members, avoid centralizing knowledge on managers or key team members.

18
For sharing internal knowledge 57% of the companies use Google Drive, followed by Confluence, Slack
and Wiki in that order.

The methodologies to share test results & learnings variates a lot from company to company, but the
most used practice we’ve seen so far is:
1. A report on PowerPoint, Excel or Word (or GSuite equivalent).
2. Sharing through a Slack Channel or Email.
3. Regular meetings to discuss next steps.
4. Having them participate

We saw two big things, lack of participation and singular focus. First, not a lot of teams focused on having
non-UX specialists participate and witness interviews. Feedback has shown that teams who embrace this
practice consider it the most effective way to share learnings and make the product seem more “real”.
Second, only a few companies mentioned that they were regularly testing anything besides prototypes.
This hyper focus is great for some learnings but is another practice that disappears once teams start
embracing additional tests such as live products, competitors, or benchmarks form other regions.

To no surprise, the actual distribution of information remained focused on Slack and Email. For both
internal and cross team communication more than 75% of the companies use Slack.
Product Frameworks
Most product teams aren’t using an out-of-the-box solution for creating their creative product process.
There are a huge number of combinations that lead to effective product teams. With that being said, the
techniques that appear again and again are Design Thinking and Lean Startup. These are very
complimentary techniques that tend to feed into each other.

After some deep dives with a number of product teams, what we’d recommend is starting with Design
Thinking as a launch point, then moving into the technique for product innovation that your team loves,
whether it is Human Centric Design, Product Centric Design, Jobs to be Done, or any other technique that
fits your market and product. From there, we continue into the Lean Startup Cycle for iteration. This

19
flow of research to innovation to testing is a fantastic technique for any team and can be built out based
on your headcount, resources, and skill set.

As we see this 3-step development cycle as a common theme, we want to dive into each phase. For the
design and research phase we’ve seen a huge reliance on existing products, as Analytics, Tracking, and
Analysis. These are great techniques for iterations, but for companies that are truly trying to innovate
and launch brand altering product changes, we’ve seen more success with Ethnographic research and
User Interviews. This pair of what a user says and what a user does in the absence of our product or with
the product have led to huge gains for companies who have invested the time.

We would love to stress that teams who self-reported Ethnographic User Research all said they loved it
and it was a huge reason for their success. We suggest everyone trying to launch at least a small research
test in 2018.

We also would like to pass along one insightful comment about the team distribution. Some of the larger
more experienced teams suggested to have BI teams working directly with UX teams and designers in
order to better include quantitative data early in the process.

As we’ve seen increased reliance and usage of user focused research and tests, many companies have
reached out to see how other teams are recruiting users. As you can see below, many teams are relying
on their existing user base. This is fantastic for incremental improvements and testing but isn’t the most
effective for research or testing certain pieces of the flow. As such, companies are using a few other
primary sources.

20
We did not get agency costs but we have gotten initial feedback that paid tests vary from $10 (30 reales)
to $50 (150 reales) for recruiting and having a mix of shows and no-shows. Going higher in price has
reduced no-shows. Some teams have recommended gift cards as the optimal “reward” for participants.

Post design and research, teams move to innovation and build phase for new ideas. The self-reported
next steps are:

Based on the responses along with our experience, we recommend relying on testing, prototypes, and
wireframes only when you are looking for smaller iterations on a product, or speed. If the team is
focusing on making larger changes or trying to move a metric in a significant way via complete innovation
or redesign, mapping, especially systems, experience, and scenarios have been highly successful first
steps. Some heads of product reported that the switch to proactively understanding the real-world
implications of a product instead of relying on reactive feedback has revolutionized their teams.

21
After creating initial product, the iteration cycle relied heavily on testing. Nearly 90% of teams
immediately move to testing instead of simply moving forward to a new project. The majority of teams
use different tests for different scenarios, but the priority has been given to the flow and structure over
the details. This is significant for teams evaluating new tools as many of them do not allow for multipage
tracking. If your team believes it should be testing content look for tools that allow multipage testing.

As far as the actual tests go, there has been an evolution in techniques in the past year as companies
have started going beyond simple A|B tests. A|B are still the dominant test, but with scale and
experience, we recommend going beyond. The KV growth would recommend that every product team
schedule regular A|A tests, as we’ve seen multiple companies catch errors in their testing flow.

We would like to note that on top of these standard tests, nearly 75% of companies added a unique test
(1-2 teams using) they do for their particular product, ranging from in product surveys to click tracking.

22
Product Providers
Remote Click and Session Recording Provider
TL;DR
Remote Tracking Usage:
Our Remote Recording selection for 2018: HotJar
Our Remote Recording selection for SEO oriented Companies 2018: Server-side tracking

What you told us:

Hotjar continued another year of dominance as it expanded its product offering and has improved its
customization while remaining affordable. However, not everyone loved HotJar as it has been linked to
terrible page speed performance, for those of you concerned with SEO, it’s worth considering other
options or only using HotJar when absolutely necessary.

Remote Click and Session Recording Provider

Our Conclusion:

23
In-Person Click and Session Recording Provider
TL;DR
Remote Tracking Usage:
Our Session Recording selection for 2018: HotJar
Our Free Session Recording selection for 2018: AZ Video

What you told us:

Companies who loved HotJar for remote tracking also use it for in person recording (roughly 40% of
subscribers). In these users, there was a lot less love for the product. For everyone else, including HotJar
users who don’t love it, there has been a turn to other cheaper alternatives such as LookBack and AZ
Video. Lookback has seen a downward turn in love, but an upward turn in functionality over the last 12
months. We still great results from Lookback, but as they reducing the value to cost ratio as plans are
starting to get expensive for a lot of companies. As a result, we saw great diversity this year.
In-Person Click and Session Recording Provider

Our Conclusion:

User Survey Providers

TL;DR
User Survey Usage by Product Teams: 83%

24
Teams who took time to build their own Survey Tool: 21%
Overall Most “Loved” Product Provider in 2017: Typeform
Our Survey Provider for 2018: Typeform
Our Robust Survey & Market Research Provider for 2018: SurveyGizmo

What you told us:

Companies are starting to move away from SurveyMonkey which was a best-in-class solution. We aren’t sure what
the defining reason for its drop is, but we have seen companies moving towards Internal Tools and other solutions.
The SurveyMonkey product still gets love, but not as much as before. Unless we see improvements in the product,
we expect to see further migration in 2018. The same can be said of Google Forms, a mainstay due to its free usage
and Google Sheets integrations, which also failed to innovate in 2017.

As our previous top 2 providers have slid, two new providers have stepped in. Typeform, with its clean UI and easy
to use product emerged as the top provider in surveys and actually received the most “love” of any provider in the
product survey. The combination of the look and feel with the variety of pricing plans made it the clear winner.
With that being said, we also saw SurveyGizmo get some major love and upon deeper inspection we found that it is
a truly robust solution that offers huge functionality for surveys plus additional marketing insights, internal survey
tools, and curated insights. One thing we loved was the partition of data, a great tool for companies worried about
internal sharing or even compliance issues.
User Survey Providers

Our Conclusion:

25
NPS Providers
TL;DR
NPS Tracking: 83% of Companies
Our NPS Provider for 2018: Typeform
Where we would start shifting to Internal Tools in 2018: NPS

What you told us:

There has been a large shift away from the use of providers for NPS. This single survey question is: easy
to track and used in a wide variety of formats, making it ideal for custom solutions. As such, 44% of
companies moved to internal tools, one of the highest volumes in the entire survey. Elsewhere, the
simple and clean look and feel combined with the amazingly simple setup put Survey Monkey and
TypeForm on top in this category again this year.

NPS Providers

Our Conclusion:
Go internal if you can, but if you need an out of the box solution, we recommend the following:

26
Mental Mapping Providers
TL;DR
Mental Model Mapping: Less than half of companies are doing it
Our Stand Alone Modelling tool for 2018: Mindmeister

What you told us:

The majority of you are not using any tool for this, either because the team feels the product is not complex enough,
or it’s just not a priority. What we’ve seen is that for products of even medium complexity, the modeling tools can
be highly effective. The other big feedback is that they can help with institutional knowledge and onboarding of
product team members when done effectively.

Our Conclusion:
First and foremost, more companies should engage in the basics of model mapping and if you’re
beginning, we suggest the following:

Design Process Providers

TL;DR
Design Process Provider for 2018: Sketch
Our Stand Alone Prototyping tool for 2018: Invision

What you told us:

In 2017, Sketch was king. It grew and improved over the last year to firmly entrench itself as a top option from
wireframing to design to prototypes. Furthermore, the great integrations being done with Marvel and Invision have
made it even more valuable. The only downside that gets reported is that often there are key team members who
cannot use it because they are on the business side of things and using a Windows device or are a developer on
Linux. Outside of that, there was little fault to be found. In each category, there are other tools that really do well,
including the basics of pen and paper, but if you want to keep things inside one team, Sketch wins.

The biggest area of weakness that we saw was prototyping, as Invision’s full suite still easily outpaces Sketch.
However, there is an additional cost and some design areas that are lacking. In 2018, it will be interesting to see
how they come together and who wins as they move into the same spaces.

27
 Wireframing Providers Visual Design Providers Prototyping Providers

Our Conclusion:
On a smaller team that follows the process from start to finish, on a larger team that brings in more
teams and has the bandwidth, consider a standalone prototyping tool.

Experimenting Providers
TL;DR
Fastest Riser in Product: Google Optimize
Our Stand Alone Experimenting tool on an unlimited budget for 2018: Amplitude
Our Stand Alone Experimenting tool for 2018: Google Optimize

What you told us:

Companies are taking two main approaches to experiments, either using a free or internally developed
tool or using a robust platform that integrates core tracking and experimentation in the same place. Self-
created tools took a larger role, hitting 42% in 2017. Combine that with the free versions of Google
Optimize and it’s the majority of the portfolio. On the other hand, we have some companies committed
to Amplitude and Mixpanel, which offers testing and much more in the enterprise platform. This, plus the
lack of a speed/SEO tradeoff seen in VWO and Optimizely has shown them to be better options, if you
make it a priority.

It is worth noting that we’re not sure how long Google Optimize will keep its free version as robust before
locking top features into 360 as they have in analytics.

28
 Experiment Providers

Our Conclusion:
If you can afford it, get Amplitude and pair it with the events, but if you want a great, easy to integrate
solution, we love Google Optimize. If you have the resources, consider and internal tool, but test it
regularly with A|A tests!

Customer Onboarding
We weren’t able to draw any conclusions here as only 25% of the respondents said they are focused on
customer onboarding tools. Our thought is that many companies are using in-house solutions that they
consider part of the core product. If you’re looking for a solution, the two recommended by companies
are AppCues or Compass.

29
BI Insights
BI and reporting has grown from a question to a core piece of this survey over the last two years. It’s the
area that is evolving the fastest in the Portfolio, so we’ve tried to tackle a lot of information, but a lot of
this might pivot in the next 12 months. The most glaring number is MySQL’s fall from 75%+ usage in 2015
to 55% in 2016 to 41% in 2017 as PostgreSQL became the first non-MySQL top usage in the history of this
report.

With that in mind there are 3 main challenges that came up that BI teams are facing, plus one hold over
which is finally starting to become less of an issue. The problem we’re happy to see fading more and
more is the evangelization of data. This is the first time this isn’t the top reported problem, and we’re
happy to see this. Some companies are still reporting it, but it’s a downward trend as the entire
ecosystem is seeing data as a huge asset. BI is finally sexy!

With that tackled, let’s dive into the 3 new challenges of BI teams. First and foremost, teams are
concerned about putting the right data in place so it is used by decision makers. This theme of using the
data roots at getting the right data, and one of you said quite eloquently:
“The challenge is to make sure we measure the right stuff for decision making. Not more, not less.
Tracking information that does not add entropy actually worsens the analysis and confuses decision
makers.”
After speaking to a number of you, the key here is to make sure the data isn’t overwhelming and its
visualized and focused. We suggest taking some time to consider how complex you make your query and
visualization tools. Avoiding the need for queries is key for this. You can see more in the providers.

Second, many of you talked about including external data, which tends to come in the following forms.

30
As you can see, this is a very wide variety of sources which requires a large number of solutions. The top
piece of advice given by you all is that the focus needs to be on getting structured data. Take time to
properly format forms and processes and it saves tons of time. Always focus on clean and structured
data, not on fast but sloppy.

On top of the online, many companies have started importing their own data from offline sources. This
ends up being a very manual process but everyone we spoke to said it was worth it, so take some time,
make a process, and incorporate.

Finally, the third major challenge you all passed on was building out the right system. Hopefully the
full list of providers can push us all in the right direction, so let’s dive in.

31
BI Providers
Data Collection Tool for User Tracking
TL;DR
Our Data Collection selection for 2018: Amplitude
Our Data Collection selection on a budget for 2018: Snowplow

What you told us:

With 2 companies leading the Snowplow meetups in 2017 it has become a portfolio favorite. Companies
love the solution because gives you the ownership of the data, you only pay for the infrastructure and
you can integrate it with other sources of information. The downside is that you need mature IT and BI
teams to set it up, build the data models and visualize data.

Teams also loved Google Analytics for the same price savings, but much easier set up and usability for
non-BI experts.

And of course, Amplitude continued to receive fantastic ratings as an extremely powerful well thought
out solution, but we once again received feedback that it’s too expensive unless you’re using the
numbers for business solutions.
Data Collection Tool for User Tracking

When it comes to web specific, things shift even more heavily towards Google Analytics. Many
companies are using it with others as they’ve had it up consistently for years. The easy integrations with
AdWords, Google Optimize, and the other Google platforms make it a must. People love the price, ease
of use, familiarity, and view of their site. The biggest strike against it is that it only gives you the view of
your site, not your users and the macro view can sometimes be wrong with sampling. Companies who

32
have the non-sampling enterprise 360 edition are the most satisfied followed closely by companies
pairing it with an event-based tracking.

Website Specific Provider

When talking about App Analytics the situation shifts dramatically as Google Analytics just isn’t that great
at In-App events. As a result, the usage falls from 74% to 24%. Amplitude and Firebase emerged as
favorites, however Firebase isn’t universal, leaving Amplitude as a clear winner. However, it’s worth
noting that only 35% of the app owners track app engagement at a granular level, most just have an
overview. On top of that, only 37% of the App owners track Installs and Uninstalls.

It’s also worth considering that there are some other platforms like Adjust and AppsFlyer that aren’t
traditional systems but do help their marketing teams.

App Specific Provider

Our Conclusion:
If you can afford it, implement Amplitude. However, the prices went up again this year so we see it as
major barrier for early stage companies. If you’re looking for solid input on a budget we recommend
using SnowPlow AND Google Analytics as SnowPlow shows your users and analytics shows your site/app.

33
 Overall Tracking Web Tracking App Tracking

Data Collection Tool for User Tracking

TL;DR
Most Popular Database selection for 2017: PostgreSQL
Most Common Database Construction for 2017: Bottom-Up
Our Relational DBMS selection for 2018 (AWS): Datomic
Our Relational DBMS selection for 2018: PostgreSQL
Our Data Warehouse for 2018: BigQuery
Most Common Data Warehouse for 2017: Amazon Redshift
Volume of companies who are using a Data Warehouse for 2017: 78%

What you told us:

With data as such a central theme in 2017, many companies started moving from a simple set of
Databases to constructing warehouses and lakes where they consolidated more information than before.
As such, many companies are thinking long and hard about the construction and setup. The biggest
change is the move away from MySQL to PostgreSQL and other options. Datomic has emerged as a
powerful solution for companies on AWS, whereas PostgreSQL has emerged as a popular and accessible
solution replacing MySQL. (It’s worth noting that PostgreSQL has passed in our portfolio but still lags
dramatically in the community as a whole). The other big decision companies are making before deciding
on providers is the construction of the database.

There wasn’t much consistency among who chose what, its more about getting the right data for the
right situation. The only correlation we could see is that the companies that are deeper into the Machine
Learning process are more commonly using Bottom-Up approaches.

34
As far as the actual hosting of the warehouse and lakes, Amazon is still the most used tool and despite
having more than 50% of use share none of the companies using it loved the solution. The direct
feedback we got was that it presents significant performance problems when scaling.

Data Warehouse Provider

Our Conclusion:
This is a problem that needs to be tackled individually and our advice is to talk to the other CTOs and Data Scientists
to go deeper, if you’d like to hear more about a topic, please don’t hesitate to reach out. The biggest conclusion we
have is that Big Query has left more customers satisfied than Amazon, even with 20% of the user base, so it’s worth
looking at!

Data Processing Providers and Frameworks

TL;DR
Internal Data Processing in 2017: 25%
Our Data Processing Provider selections for 2018: Databricks and Looker
Our Framework for Big Data Processing for 2018: Spark

What you Told Us:

Talk about a category that got some love! Data Processing Providers was easily the category that had the
most winners in the survey. Databricks got a lot of love, especially in the ML crowd. Looker was loved
for its integrations customization. On top of that people loved Stitch and Splunk which we’re not familiar
with. On the other end of the spectrum, PowerBI saw its fall downwards. It’s still very useful for finance

35
and those companies using complex excel plugins, but elsewhere, its fallen all the way to 5th after being a
favorite in years past.

Data Processing Providers

As far as an accompanying framework Spark was the runaway favorite for the slice of the portfolio doing
big data processing. It was the most used and the highest rated grabbing nearly ¾ of the portfolio who
needed a framework. We also had some positive but less amazing feedback about Hadoop and MAPR.

Our Conclusion:
Test, test, test. More than any other category this seems to have a lot of great winners. We will say that
for ML we’d lean towards Databricks, but really, we couldn’t pick a winner.

Data Visualization Providers

TL;DR
Our Most Loved Category in 2017: Data Visualization
Our Data Visualization Provider for 2018: Tableau
Our most important feature for Data Visualization in 2018: Usability

36
What you Told Us:
This was a category where the majority of companies found something they loved and stuck to it. The
most loved was Tableau by a wide margin, but you’ll notice, no one loved the price! What stuck out
about it as a tool was its versatility and ease of use. For teams that have non-data scientists working deep
in the numbers, Tableau is great and really solves the problem of:
“Tracking information that does not add entropy actually worsens the analysis and confuses
decision makers.”
Companies really appreciated being able to hand the tool off and let the decision makes play a little in the
data themselves when needed. The big downside of Tableau is that it doesn’t work great for real time
integration. We haven’t seen a truly satisfactory solution on this front. We also love both Looker and
Metabase because of how well they integrate up the funnel in the BI chain. The surprise for us was
Domo. Its robust ability to visualize, integrate well into decks, and provide dashboards made it a favorite,
especially for real time reporting!
Data Visualization Providers

Our Conclusion:
We loved a lot of these, but the top 2 for purely visualization are Tableau for power and Domo for its solid
platform combined with great dashboards.

37
Internal Dashboard Providers
TL;DR
Our Dashboard Provider for 2018: Domo
Our most important tested category in 2018: Internal Dashboard Providers
Most request Dashboard Feature in 2017: Real Time Updating

What you Told Us:

Dashboards were by far the most tested and diverse category in 2017. When we asked why, teams said
they needed to find robust plus real time plus easy integration. As stacks varied so much, there wasn’t an
easy winner. What surfaced was that once again Tableau, Domo, Looker, and Metabase were favorites.
However, the biggest shift happened in Tableau with a little less love, and Domo with an improved rating.

As you can see below, there are a lot of 1-2 companies are using it options, so we’ll wait and see in 2018.
Internal Dashboard Providers

Our Conclusion:
With no obvious winner, we’re playing it safe and going for the tools we believe are here to stay at the
top of our list:

38
Other BI Providers
A few tidbits you passed along:
1. Grafana is good to diagnose system health and pair it next to business metrics. It helps diagnose
partial downtimes.
2. The Pentaho Suite is a great tool with a lot of amazing open source options. Definitely a tool
worth checking out.
3. Although Kibana doesn’t make for easy work for the non-BI focused team, the integrations and
ease of use with Elastic Search make it a great tool of have in your toolbox.

39
IT Insights
We had 3 main concerns coming from IT teams that popped up a few times, there is the always a
question, “How do we best structure our growing team?”, the recent “Automated testing and QA best
practices,” and the new “Are people actually applying Machine Learning effectively, and if so, how?”.

Let’s dive into the classic first, team set ups. First and foremost, when we did follow up, most companies
said they are using frameworks, very few of you are actually taking these methodologies by the book.
Teams are using hybrid and custom philosophies. With that being said, a root in Agile remained the
most popular for another year.

As always, we ask for self-reporting productivity and satisfaction. Once again, lean teams reported higher
scores, but being a smaller volume, we went deeper into the numbers and followed up.

We actually found 2 factors that were much stronger correlations. Satisfaction is tied to two things,
1. Smaller development teams with a co-founder CTO
2. Teams doing more complex projects at the moment

40
The first is something you can’t control after founding, but the takeaway is to have a strong technical lead
who is involved in the whole business so the engineering and product team feel involved. The second is
tied to teams who are actively doing machine learning and complex BI projects. Challenged teams are
happier!

Productivity on the other hand was tied closer to the team structure, however, the more important factor
was a VP engineering to go along with a CTO. Teams that have multiple tech leaders at a high level were
able to be more productive. The recent discussions about hiring a VP Engineering make a lot of sense.

The methodologies were less differentiated. We found one major insights, an increase in planning
sessions leads to greater self-reported productivity. There is a small jump at each increase time length.
We’ve given you the number below to draw your own conclusions.

Develpoment Methodology Stats

Looking at QA and automated testing, this still came up as it did last year, but to a much lesser degree.
The questions were focused around 2 main points, “what automated structures are working best?” and

41
“do teams have dedicated QA squads?”. We don’t have an answer to the dedicated QA, but we know
that most teams don’t have a dedicated squad, but many do have a team member or rotation.

Regarding QA practices, the main focuses were on Automated testing and CI/CD tools, in most cases devs
run local tests and submit pull requests to be code reviewed, after this CI/CD tools run automated unit
and integration tests before deploying. Once again here the portfolio is keeping up with the global trends
and focusing on continuous integration and delivery using automatic testing tools and pipelines.

A secondary focus was code review, including code style, stress, function and exploratory testing of the
product. One company mentioned doing focus groups for important features before their release.

Stress and Performance testing are starting to make an impact as more advanced companies, especially
fintech’s are testing and testing under extreme circumstances. On our end, the biggest yellow flag is that
only 15% of the portfolio companies are doing security testing. We strongly encourage starting this
practice as some companies have dealt with attacks over the last 18 months.

Finally, the big “new” topic of choice, Machine Learning. We have a handful of companies using more
advanced techniques and getting to online learning and similar and want to pass on what we can without
handing out any identifying details. At a high level, roughly 1/3 of companies are using some form of
Machine Learning. We wanted to split these into 3 buckets to help digest what they shared.
1. Group 1: Using a provider
2. Group 2: No provider, just started
3. Group 3: No provider, advanced

Group 1: Using a provider

42
Providers are a great starting point for classifications, both binary and multiclass as well as regression
testing. Be aware that you will likely be using more Bayesian and Decision Tree techniques for learning
across the board, but each provider offers a few other easy to implement learning techniques. The
biggest actionable advice is to start simple and prove that it works, especially because scaling in
production can be a big challenge. The 3 providers being used are DeltaBrain, IBM Watson, and Amazon
Machine Learning. At this stage, none of them stood above the others in your ratings.

The best successes in this group have been around forecasting.

Group 2: No provider, just started

Companies are diving in on their own, many after using a provider first. The starting points have been
again classification and regression testing with a focus on supervised learning as well as some logistic
regression and reliance on decision trees. Here companies are diving into more personal issues.
Companies pointed out that once they moved to internal tools that you really have to focus on the
models. One was kind enough to pass on the follow:
“Models lose about 50% of their discriminatory power over 6 months, which requires recalibration
using all the new data points and business adjustments”
Therefore, stay on top of this as you migrate from providers to internals.

Internally, the first big successes were again in forecasting, but with a close second on personalized
recommendations.

Group 3: No provider, advanced

Companies who have started out on their own and evolved have continued using supervised learning,
regressions, and decision trees, but also started on reinforcement and deep learning. The results have
been solid, but there have been many learnings here. A few are:
• Scaling continues to be a challenge,
• Watch out for different code paths between training models and production,
• Enable intelligent code reviews, and
• Make sure you recalibrate early and often.

The best success here has been around risk prediction alongside customer value.

Main Takeaway
With that being said, the CTOs I spoke to said that Machine Learning isn’t as simple as finding a provider,
the most important takeaway was that you need proper infrastructure and clean data as a prerequisite
for any machine learning.

43
Who’s Building What?
The last piece of insight we always like to share is “who’s building what?” which is essentially an overview
of what’s being used in stacks around Latam. There was a lot of movement in the last year. Predictably,
PHP fell from its top spot, all the way to #4 in the portfolio.

First, Ruby (particularly, on Rails) continued its ascent as a simple and easy framework with lots of ready
developers.
Further down the line, 2017 saw companies looking for more robust solutions to mature web applications
and REST APIs as well as a shift towards services and micro-service oriented architectures give a rise to
Spring and other Java frameworks.

Finally, JS continued to grab portfolio share in both front and backend teams.

Full info below:

Who’s Building What?

44
IT Providers
Hosting Providers
TL;DR
Our Hosting Provider for 2018: AWS
“Best Value” Host in 2017: Google Cloud

What you Told Us:

Comparing the whole IT section with last year, once again, our teams are very AWS driven. As we can see
AWS still dominates usage and “love”. It still sits as the best in class in the region, but GCS made moves
last year. Most importantly, the majority of the users “loved” the price. The lower costs, especially
without relying on reserved instances is seen a huge plus. However, even with these gains it’s still
second.
Hosting Providers

Our Conclusion:
2018 will probably see the gap between #1 and #2 shrink, especially as companies migrate off of free
credits and on to their own budgets, but not enough for us to doubt AWS, yet.

CDN Providers
TL;DR
Our CDN Provider for 2018: CloudFlare

45
CDN Usage in 2017: 96%

What you Told Us:

AWS CloudFront remains the most common tool with similar penetration as past years, but still sits
behind Cloudflare in usability. Even with a much smaller usage, Cloudflare continued to generate more
“love”. CloudFlare has won this year again on the strength of its usability and simple interface.

CDN Providers

Our Conclusion:
Even with the lesser integration and usage, we still love Cloudflare especially in light of the improved DNS
offering.

DNS Providers
TL;DR
Our Standalone DNS Provider for 2018: AWS Route 53.
Our Combine DNS/CDN Provider for simple usage in 2018: CloudFlare

What you Told Us:

AWS is still king in DNS. Route 53 is a great service that gets a lot of love. If you’re on the AWS stack it
seems to make a lot of sense especially with the low cost, ease of use and customization. However, if
you’re looking for simple, CloudFlare is right there too.

46
 DNS Providers

Our Conclusion:
For a robust system, Route 53, for simplicity CloudFlare.

Automated Testing Tools

TL;DR
Our Automated Testing Tools Providers for 2018: Circle CI & Codeship

What you Told Us:

This was a crowded category, especially in light of the fact that many companies are actively using
multiple providers. There wasn’t a lot of love in the category, but there were a lot of satisfied customers;
as none of the tools were really considered bad as this had one of the higher average scores.
Automated Testing Tools

Our Conclusion:

47
There isn’t a runaway best solution, so we suggest testing or potentially integrating multiple solutions
with a little redundancy

Site Monitoring Tool

TL;DR
Our Site Monitoring Tool for 2018: New Relic
Tool with the Most Love on Product and Most Dislike on Cost in 2017: New Relic

What you Told Us:

New Relic is hard to beat, even at its cost. This is one of the “you get what you pay for” and it’s such a
critical category, that over half the portfolio still pays for it.

Site Monitoring Tools

Our Conclusion:
You can’t cut costs here, so we continue to recommend New Relic

48
Security Vulnerabilities
TL;DR
Our Security Vulnerabilities Provider for 2018: Tempest
Security Vulnerabilities Usage in 2017: ~30%

What you Told Us:

With only about 30% of companies using providers for Security Vulnerability testing, this was an area that
could use major improvement. Companies who use it have been happy that they have and again, with
the attacks we’ve seen in the last 18 months, this is something we do recommend. There wasn’t any
“love” in this category, but Tempest got solid ratings across the board and has heavy usage.
Security Vulnerabilities Providers

Our Conclusion:
More companies need to dive into these areas, whether its internal tools or the providers below:

49
Search Provider
TL;DR
Our Search Provider for 2018: ElasticSearch
Search Internal Tool Share: 19%

What you Told Us:

This is a field that has seen continued shift over the last 2 years. In 2015 and 2016, SOLR and Neemu
were on the top of this category, but with Neemu being acquired and Elastic Search passing SOLR, we’ve
seen both slowly dropping down. Companies have been migrating to Elastic Search and internal tools
heavily in 2017.
Search Providers

Our Conclusion:
Migrating to a tool you can truly control and optimize for Search like Elastic Search is the way to go.

Containers
TL;DR
Our Containers Provider for 2018: Kubernetes
Container Usage in the Portfolio in 2017: 68%

What you Told Us:

50
Almost 70% of the companies are using a containers solution, huge growth from the start of 2016.
Following the global industry trends, Kubernetes was the best ranked and most loved and despite AWS
Docker being the most used, its score was clearly below Kubernetes.
Containers Solutions

Our Conclusion:
We’re falling on the side of the global industry plus the love in the portfolio.

PAAS
TL;DR
Our PAAS Provider for 2018: AWS if you’re on AWS; GCS if you’re on GCS
PAAS Usage in 2017: 56%

What you Told Us:

PAAS is a pretty straight forward solution that means using what you’re primarily hosted on. Similar to
hosting, 80% using AWS as their main PAAS provider, receiving was quality votes. GCS was used in 2018
for the first time and was strongly recommended.
PAAS

51
Our Conclusion:
Integrate with your stack

Anti-Fraud Solutions
TL;DR
Our Anti-Fraud Solution for 2018: ClearSale.
Anti-fraud Solution Usage in 2017: 37%
Anti-fraud Internal Tool Share: 20%

What you Told Us:

Anti-fraud was a bit of a niche category in the portfolio and as such, no one found a winner. This is one of
the few categories with no “love” while in parallel receiving growth in internal tools.
Anti-Fraud Solutions

Our Conclusion:

52
Payment Gateway
TL;DR
Our Payment Gateway Provider for 2018: Mundipagg & Adyen.
Payment Gateway Usage in 2017: 74%

What you Told Us:

Never an easy category, people were hesitant to hand out 9s and 10s but we had a lot of 7s and 8s.
MundiPagg was the closest to emerging from the pack, but Adyen has proven staying power, having been
high on the list in years past.
Payment Gateway Providers

Our Conclusion:
Some of these providers will start to slip as the year progresses and by 2019, there will theoretically be
some consolidation in this field!

53
Customer Support Providers
TL;DR
Our Full Stack Customer Support Provider for 2018: Intercom
Our Chat Only Customer Support Provider for 2018: Zendesk
Our Customer Support Provider for voice heavy interactions 2018: Teravoz

What you Told Us:

Customer support is an ongoing struggle for many companies. The balance of online and voice has been
a big point of emphasis. When looking at full stack customer support, Intercom emerged as a clear
winner. However, looking deeper into companies using a lot of Voice, Teravoz emerged as the favorite in
that subset.
Customer Support Providers

Our Conclusion:
Seeing the success of the roughly 10% of companies using internal tools, we’d suggest looking into a
combination of internal tools and providers if you have the IT bandwidth. Otherwise, we suggest
Intercom for full service and Teravoz when focusing on VOIP.
Full Stack Providers Chat Tools

54
Team Management Tools
TL;DR
Our Project Management Tool for 2018 in Lean Dev Teams: Trello
Our Project Management Tool for 2018 in Agile Dev Teams: JIRA
Our Internal Knowledge Tool for 2018: GitHub
Our IT Communication Tool for 2018: Slack

What you Told Us:

There were no surprises in the team management tools in 2017. Trello and JIRA each claimed their
categories again as did GitHub and Slack. The tools are built for devs by devs so they continue to evolve
with their users. For project management we did split tools based on Agile/Lean as it makes the most
sense to us.

Project Management Tool

Internal Knowledge Tool

55
ISP
We have low volume of data points at the provider level but iPlan seems to be the provider to consider in
Argentina, Mundivox in Brazil and Telmex in Mexico. All of the mentioned providers were highlighted for
Reliability. As the information is limited we’ve attached the raw data below:

Other Tools

56