Why You Should Have a Build AND Buy Product Analytics Strategy

Large-scale enterprises and those with advanced data teams have specialized product analytics needs. But neither an in-house system nor a third-party vendor alone can meet them all. You need both.

Best Practices
July 31, 2020
Image of Nate Franklin
Nate Franklin
Director, Product Marketing, Amplitude
Why You Should Have a Build AND Buy Product Analytics Strategy

When it comes to choosing a product analytics strategy, large-scale businesses looking to increase their agility and turn their data into a competitive advantage should be asking, “How can we build AND buy?” Too often the decision gets framed as “build versus buy,” which creates two choices that do not account for the needs of the teams they support.

Many enterprises build an analytics system on their own because they have highly complex data and needs, and a team of talented engineers and data scientists to work on them. But relying exclusively on an in-house system has a number of pitfalls.

For most organizations, building and using a custom analytics solution is time- and cost-intensive. Maintaining it can be even more so. More often than not, custom systems are accessible only to the most technical users. Those technical users then spend more time fielding one-off queries from other teams instead of working on complex data projects that impact the strategic direction of the company. Meanwhile, product teams struggle to get the insights they need to make data-informed decisions—waiting days if not weeks for the technical user to work through a backlog of similar requests.

The challenges above should not surprise anyone. Few businesses specialize in building product analytics. So by their nature, in-house systems will face technical and resource constraints. That’s why Amplitude exists. It’s the problem we were built to solve, and we are very good at it.

Amplitude is powered by our proprietary columnar-based storage & analytics engine, Nova. Nova was purpose-built to rapidly ingest, process, and query product event data. It regularly serves queries running on tens of billions of events, with a median query time of <1 second. Nova, combined with our out-of-the-box analyses and exploration tools, is how Amplitude delivers self-serve analytics that nontechnical teams actually adopt.

We believe you can have the best of both worlds—an in-house solution for the highly specialized needs of your data teams and a self-serve platform for the business. Here are a few key reasons why building and buying might be the right path for you.

1. You Have Many Data-Hungry Teams That Need to Move Fast

All of your employees need access to insights from customer data in order to drive growth. How quickly they can get those insights determines how quickly you can grow.

Your data teams are likely absorbed (or desire to be absorbed) with complex problems that could drive the future of your business. At the same time, your product, marketing, growth, and design teams have critical needs involving your customer data. But their questions are less complicated; the teams might need to know, “How many users of this cohort churned last quarter?” “Where are our leads coming in from?” “How does this home-page A/B test design compare to the other design?”

If you only have a custom-built analytics system, nontechnical teams will inevitably end up relying on their technical coworkers to answer those questions. Your analysts will resent the time and energy that answering takes away from their regular work, and your product managers and marketers will resent having to fly blind without data when they can’t get those answers quickly.

The result: nobody is happy, and nobody is getting what they need out of your built product analytics system.

A third-party analytics platform, like Amplitude, helps you address the needs of your entire organization. Data teams focus on the large-scale problems that matter to the business. The rest of your organization has self-serve access to powerful data insights that drive their work forward. Critically, they can explore the data freely—asking and answering their own questions with helpful visualizations in near real time. And in both cases the analytics organization is credited with delivering more insights to the business.

2. You Need to Maintain Data Integrity

We often hear from analytics leaders that they are concerned about losing control over data management, quality, and security by adopting a third-party analytics solution. While there are certainly some compromises that have to be made, like routing data to a third party, with Amplitude there are far fewer compromises than leaders expect. And let’s not forget that in-house systems run into their own data integrity issues.

Just ask HubSpot. Despite the fact that their in-house solution was built by smart and talented people, they were plagued with system outages and reliability problems. “It got to the point where people were increasingly skeptical of the data that came out of the system and not using it very much,” according to HubSpot Head of Product Analytics Dan Wolchonok.

Amplitude is an expert in the technical challenges of processing event data and querying it at scale. For example, during ingestion Nova automatically dedupes events and runs an identity resolution service that retroactively merges users’ authenticated and unauthenticated activity into a single profile.

After seeing Amplitude, HubSpot scrapped this system and built a smaller one that was closely tailored to their engineering team’s needs, and therefore easier to maintain and control for quality. Nontechnical users don’t need access to that system, because HubSpot implemented Amplitude, which is used for 90% of their analytical work across all of their teams. Amplitude’s raw SQL access via Redshift is an important feature for HubSpot because it gives analysts the ability to query the raw data as well.

Amplitude takes security and data management very seriously. In fact, we go beyond many of peers when it comes to security and privacy. And we have the most comprehensive data governance with built-in instrumentation planning, codeless transformations, and tools to quarantine or reject bad data. In addition to our pre-built charts, we offer Templates for admins to set up best practices and provide guardrails for measuring KPIs, releases, experiments, and more.

A third-party analytics solution like Amplitude can help maintain your data integrity while providing broad data accessibility, and a targeted in-house solution can meet the tightly controlled needs of your data science or engineering team.

3. You Want to Promote a Data-Informed Culture

A data-informed culture is the aspiration of most modern businesses. But without easy self-serve analytics, you can’t normalize data-informed decision-making across your entire company. As product and growth leader Fareed Mosavat said, “If you say you’re data-driven but everything has to go through an analyst, you’re not actually data-driven.”

When the entire organization relies on a custom-built analytics solution, the results usually fall short for the users outside of the data science team. Custom solutions are often complex for the end user (especially given that they may require SQL knowledge) and therefore don’t empower the rest of the team to explore even moderately complex metrics like retention, engagement, and conversion.

Amplitude bridges this gap with out-of-the-box analyses for core metrics and special tools for exploration and uncovering unknowns. This accommodates the spectrum of business teams’ needs and analytics savviness.

A custom-built system alone cannot empower the data exploration and accessibility needed to create a data-informed culture. You need to supplement your technical-but-inaccessible system with a solution like Amplitude so that everyone at your company can use data to solve problems.

Case Study: How HubSpot Got the Best of Both Worlds

Marketing giant HubSpot implemented and uninstalled Mixpanel because it wasn’t meeting their needs. Then they built an in-house system, but they struggled with the enormous cost of maintaining it when they tried to scale its use across the company. Combined with the outages and reliability problems discussed above, it entered a “death spiral,” in Wolchonok’s words.

Wolchonok ultimately accepted that, while there’s no such thing as a perfect vendor, an in-house system alone was untenable. HubSpot scrapped the majority of their system and implemented Amplitude.

Today, engineers and others in marketing, design, analysis, research, and business development use Amplitude to answer the majority of their analytics questions and drive their work forward. Their revised custom tool takes care of the specialized needs of their analysts. This dual system is more cost-effective than solely using their first internal tool, thanks to fewer maintenance needs.

Have Your Cake and Eat It Too

When it comes to product analytics, it doesn’t have to be either build or buy. You can create a blended system that allows all of your teams to do the work that fuels your business. Your custom system for data scientists will empower them to do the complex, challenging work that they love. Your third-party analytics platform will allow technical and nontechnical users alike to get fast answers to important questions about your customers. You’ll save time and resources. It’s a win-win.

About the Author
Image of Nate Franklin
Nate Franklin
Director, Product Marketing, Amplitude
Nate Franklin is a director of product marketing at Amplitude. As a former product manager and self-declared product nerd, he's often asking, "ok, so what is our goal?"
More Best Practices