Customer Data

Platforms: Build,
Buy...or Both?
1 day

ners Perform
Cart Abando ed Page Vie
wed

If True If False

Send to destinations
N
ot
es

Send to destinations
 Customer Data
Platforms: Build,
Buy...or Both?

Build or buy?

There’s an existential question in IT when it comes to

implementing a data platform. How do we want data
engineers to spend their time?

This isn’t a cut-and-dried decision to make. Teams often run

the gamut of evaluation criteria that includes cost,
maintenance, control, and scalability. And while it’s important
to consider each of these points methodically, businesses
don’t have the luxury of indecision.

In recent years, there’s been a cross-industry consensus that

“data is the new capital.” It’s a prerequisite for personalization,
product development, machine learning – the list goes on.

And as the world embraces our new digital-first era – at

breakneck speed, in part, due to Covid-19 – the sheer amount
of data being created has skyrocketed.

Customer Data Platforms: Build, Buy...or Both? 1

 Customer data growth in 2020

Nov 2020 1,008,957,892,636
Oct 2020 929,047,863,127
Sep 2020 897,947,126,671
Aug 2020 837,348,361,514
Jul 2020 801,990,610,104
June 2020 702,541,125,147
May 2020 694,499,868,063
Apr 2020 689,555,108,775
Mar 2020 631,897,344,543
Feb 2020 504,729,299,013
Jan 2020 499,462,638,211
Dec 2019 433,218,557,075

API calls tracked on the Segment platform The CDP Report 2021

From December 2019 to November 2020, API calls tracked on the

Segment platform grew by 132%.

More data should be a boon for businesses, right? Well, sort of.

Today, the average business uses well over 100 software apps, which means data
is often trapped in team-specific silos. In fact, a Forrester survey found that
fragmentation was the top challenge companies faced when trying to leverage
their customer data for sales and marketing efforts.

This has made customer data platforms the new necessity in your tech stack. As
businesses make a concerted effort to become more customer-centric, there
needs to be a scalable data infrastructure in place that can not only collect data
reliably, but stitch together data points into a complete customer profile, and
activate this downstream in marketing and sales tools.

Customer Data Platforms: Build, Buy...or Both? 2

 said a CDP will be critical to their
73% customer experience efforts.

A survey of over 4,000 decision-makers. Source

This makes the question of whether to build or buy a CDP even more high stakes.
But...why is it always framed as an either/or?

At Segment, we’ve supported businesses that have opted to purchase a CDP and
those that were looking for more of a hybrid model (a data platform partially built
in-house, and partially outsourced).

The reality is that businesses today have three viable options when putting
together a plan of action for their CDP:

Start from scratch and build a CDP yourself.

Buy a CDP off the shelf.

Opt for a hybrid built-and-bought data platform.

In this guide, we consider the nuances of each option and provide a framework for
making the right CDP decision for your business.

Customer Data Platforms: Build, Buy...or Both? 3

 Building a CDP in-house
The first option for businesses when it comes to implementing a CDP is to start
from scratch and do it themselves.

Perhaps the most notable benefit to this approach is complete control. Data
engineers have the flexibility to customize how data is stored, collected, and
consolidated into identity graphs. Engineers can prioritize building features that are
unique to the business, rather than be at the whims of a vendor’s product roadmap.

Because, if data is an asset, then how it’s used is a competitive differentiator –

which makes a bespoke customer data platform all the more compelling.

But, there’s a catch: time-to-value, total cost of ownership (TCO), and capability
tradeoffs. Building and maintaining an in-house data platform is an ongoing project
that requires a large investment of time and resources.

Starting out, many companies believe they have the capability in-house to build an
ID graph for identity resolution, or a system that scales and enables real-time data
delivery (as just two examples). But this type of work is both complex and costly.
While the first iteration of a data platform may seem easy to build, it’s just the
beginning of getting to a viable solution for the business. And many soon realize
they’re out of their depth.

Calculating the Cost of Delayed IT Projects Source

Businesses need to ask themselves:

Can we afford to wait while engineers ramp up an in-house platform?

Will ongoing maintenance distract engineers from working on our core

differentiators (i.e. what’s the opportunity cost? What happens when
something breaks?)

Customer Data Platforms: Build, Buy...or Both? 4

 Will building in-house slow down decision-making? (e.g. as systems get

larger, and more complex, so do the teams working on them, which can be
difficult to coordinate)

Remember that with complete control comes complete responsibility. Maintaining

ETL pipelines, setting up new integrations, complying with evolving privacy
regulations, operating and hosting costs – this will all fall squarely on the shoulders
of your engineering team when building a CDP in-house.

The scope of building a CDP in-house is often underestimated. Source

Pat Ottum, an Enterprise Account Executive at Segment notes, “It’s amazing how
many companies think they’ve built a data platform,” but then realize key elements
are lacking (like the ability to comply with GDPR or handle data deletion requests
from users).

But, if your business has the resources in terms of time, budget, and engineering
talent, this ongoing maintenance may not be an issue. In that case, the question of,
“Can our business wait?” isn’t as applicable as, “Can a vendor match the pace of
our innovation? (To paraphrase David Marble, Product Manager at Facebook and
former Group Product Manager at Intuit.)

Customer Data Platforms: Build, Buy...or Both? 5

 Buy
Rather than build in-house, businesses have the option to buy a cloud-based
customer data platform from a vendor.

This sidesteps the months-long process of building a platform internally, and

offloads the manual work of maintaining ETL pipelines and setting up integrations.
This results in more bandwidth for engineers, and faster data democratization
(giving non-technical teams easier access to customer data). With low-to-no-code
integrations and features, non-technical teams are able to leverage customer
insights (often in real-time), without distracting engineers from the product
roadmap.

A study published in the Harvard Business Review, which

analyzed 1,471 IT projects, found one in six projects had a
cost overrun of 200%, on average, and a schedule overrun
of almost 70%. (Source)

Fast implementation, strategically reallocating engineering resources, a more agile

tech stack – these are all huge selling points for buying a CDP.

But there are potential challenges to consider. If you do purchase a CDP from a
vendor, be sure to choose an open, flexible platform that can easily coexist with
in-house build. There are vendors that will specify the way data is held and
maintained in their platform, which makes it difficult to switch to a new vendor
should your business ever decide to do so.

Evaluating different CDP vendors?

Check out our CDP Buyer’s Guide

Customer Data Platforms: Build, Buy...or Both? 6

 Why did ClearScore replace their homegrown platform with

Segment?

ClearScore is a leading financial technology business in the UK focused on helping

people achieve greater financial well being, by offering free access to credit reports

and scores.

A large part of their success is rooted in their data strategy. Their home-grown

system extracted event data from a number of different sources and pushed it into

a centralized database – allowing their team to analyze and track user behavior and

drive rich customer analytics.

So, why did they switch to Segment?

In 2016, ClearScore was ready to start expanding globally, prompting the question:

what would it take to maintain and scale our home-grown solution to support new

markets?

ClearScore knew that successful global expansion depended on:

1. Seamless data segregation.

2. The ability for each country manager to adopt the tools necessary to analyze

local data.

3. Company-wide insights delivered in real time.

ClearScore realized that working with Segment could lessen the burden on

engineers during this high-growth stage.

Segment gave ClearScore the ability to quickly set up separate workspaces and

flexibly integrate with marketing and analytics tools – which helped fast-track their

expansion. By offering a consistent method of collecting data across back-end

systems, websites, and mobile apps, getting data into ClearScore’s analytics

platform became simpler and more cost-effective.

Klaus Thorup

Chief Technology Officer

Rather than extracting data from multiple platforms, we now take an

‘integrate once, extract to anywhere’ approach. This reduces the

maintenance overhead of gathering data and ensures that quality exists

further up the data stream.

Customer Data Platforms: Build, Buy...or Both? 7

 ClearScore tested their first international market with Segment in June 2017 when
they launched in South Africa. Following the success of this pilot, ClearScore
continued on with this global expansion.

3x cost savings
by leveraging Segment’s platform instead of building
in-house.

25% of engineering resources

can now be preserved for building and innovating
the ClearScore product instead of the ongoing
upkeep required of a home-grown data system.

Bruce Wood
Data Principal, Analytics

We didn’t really have real-time data available until Segment. We used to have
to wait 24 hours to be able to get our hands on the data and analyze it.
Segment surfaces any issues in the customer experience that might hurt
monetization and has allowed us to make better business decisions faster.

Customer Data Platforms: Build, Buy...or Both? 8

 Deciding what to build, and what

to buy

The pros and cons of building vs. buying a data platform are enough to send any

business into analysis paralysis. But, what if this whole conversation was just

perpetuating a false dichotomy?

There’s been a paradigm shift in how businesses are starting to think about the

customer data infrastructure. Why can’t you have the benefits of building a data

platform and working with a vendor?

Across industries, differentiation is necessary to survival – businesses need to keep

building to stay competitive. But that doesn’t mean that it’s necessary to build

everything.

A few years ago, Jeff Bezos offered up a new catchphrase

for Amazon’s web-scale services: “We build muck so you

don’t have to.”

The “muck” that Bezos described has also been referred to as “undifferentiated

heavy lifting” by Amazon’s CTO. This is the necessary but arduous IT work that

needs to be done, but doesn’t actually contribute to the business’s competitive

differentiators.

Customer Data Platforms: Build, Buy...or Both? 9

 Crucially, it takes up a lot of time: Bezos noted that their developers were spending
roughly ~70% of their time on this type of work (i.e., server hosting, bandwidth
management, scaling and managing physical growth. etc.)

So, Amazon started packaging and selling this undifferentiated heavy lifting (i.e.
Amazon S3, EC2, etc.) to offer businesses the, “ability to compete based on the
quality of their ideas rather than on their ability to create their own muck.”

Artem Ovsyannikov
Senior Account Executive

Companies like Netflix and the Amazon’s of the world have the privilege of
earning and making billions of dollars a year and are able to hire data
engineers to focus on core data problems. What we’ve found is that building
and maintaining data pipelines isn’t core to 99% of businesses today.

Investing in API-first platforms (like Twilio Segment) allows businesses to build on

top of an out-of-the-box infrastructure. It becomes more of a partnership rather
than a full on replacement, and allows the vendor to tackle some of the harder
infrastructure problems that can become a million-dollar problem (or a ten-million
dollar problem) if left unsolved.

This includes handling rising cloud costs as you scale, and guaranteeing
at-least-once delivery to ensure there’s never any lost data.

Instead of giving developers a complete solution – which, as Twilio CEO Jeff

Lawson notes, “makes a lot of assumptions about what problem your customer
needs solved” – they’re given building blocks to more quickly create their own,
customized product.

Jeff Lawson
CEO

There’s a Darwinian evolution going on right now, which is if one company in

a competitive sector hires software developers, listens to their customers,
and starts building a better customer experience, they’re going to win. Once
one starts doing it, they all need to start doing it. So it’s not really build

versus buy anymore, it’s build versus die.

Customer Data Platforms: Build, Buy...or Both? 10

 Intuit’s framework for a hybrid data platform

Intuit specializes in financial software, offering platforms and products like

TurboTax, QuickBooks, and Mint to approximately 50 million customers worldwide.

A few years back, Intuit was evaluating how to construct their ideal data platform.
They had a list of the problems they wanted to solve, both for internal teams and
their customers. This ranged from establishing consistent data tracking, to
ultimately giving marketing and sales teams access to self-serve analytics, and of
course, personalizing the end-to-end customer journey.

David Marble, who was a Group Product Manager at Intuit during this CDP
evaluation process, offered insight into their framework for decision-making. The
team knew that they wanted their data platform to be a hybrid of build-and-buy –
but separating what should be done internally vs. what should be outsourced was
more nuanced.

Intuit started with the five overarching requirements they needed from a CDP:

Source

Then, their team borrowed Geoffrey Moore’s model for managing innovation
(tweaking the framework slightly). They split criteria between two axes, the first
being “Core” vs. “Context.” Whereas “core” refers to the products and platforms
that are essential to the business (and serve as a competitive advantage), “context”
refers to everything outside that description.

As for the second axis, Intuit drew a line between what they absolutely needed to
control internally versus what could be done by a vendor.

Customer Data Platforms: Build, Buy...or Both? 11

 Source

What followed was conversations across the company to determine what Intuit
wanted to build or prefer to build, versus buy or prefer to buy.

Source

Customer Data Platforms: Build, Buy...or Both? 12

 Ultimately, Intuit settled on buying a CDP to help with data collection, distribution,
and orchestration. But when it came to their identity graphs, adding dynamic
attributes (e.g. average invoice, “last of,” etc.), and segmentation, that internal
financial data was too sensitive – and core to the business – to put into an external
platform.

Here’s the infrastructure they ultimately settled on:

Source

You can watch David Marble’s full Synapse talk here.

Customer Data Platforms: Build, Buy...or Both? 13

 Wrapping up

To understand what would be most beneficial to build, and what makes sense to
buy, consider your starting point, available resources, and both short- and
long-term goals. We’ve compiled a list of the most salient criteria to consider:

What is “core” to your business? What features, services, assets, etc. are your
competitive differentiators that you want complete control over?

What are you comfortable outsourcing? What work is fundamental to your

business, but doesn’t require complete oversight?

Do you have the engineering resources to handle ongoing maintenance and

operational costs of building in-house? Do you have the skills internally to
reach CDP maturity (machine learning algorithms, ID graphs, etc.)?

Do you have the budget to handle operational costs?

Are you in a position to scale successfully?

Are you ready to handle evolving privacy regulations (especially in different

regions, for global companies)?

These types of questions will help you filter out the components of your CDP that
absolutely need to be built in-house vs. those that actually make more sense to
outsource. Because, there isn’t a line in the sand between the two options. To stay
competitive, it makes sense to harness the benefits of both.

Want to learn more about the right blend of build-and-buy

for your business? Schedule a demo today.

Customer Data Platforms: Build, Buy...or Both? 14