YC posted a list of top alumni companies by valuation, as of October 2018. You can see the full list at https://ycombinator.com/topcompanies.

Here’s a Q&A with Matin Movassate and Ravi Parikh, the founders of Heap, one of the companies featured on the list.


What does Heap do?

Heap is a customer analytics tool that automatically captures every web, mobile, and cloud interaction: clicks, submits, taps, swipes, emails, payments, and more.

Unlike other analytics products, you can retroactively analyze your customer behavior without writing tracking code or waiting for data to collect.

How many employees does Heap have?

100

How many founders?

2

What is your most impressive recent product milestone?

Literally today. We just shipped Heap Connect, which is a big evolution of our data-out capabilities.

With Heap Connect, our customers can now pipe their Heap event data downstream to more warehouses, including Redshift, BigQuery, Snowflake, and S3. This lets them power new workflows or run more custom analysis.

The best part is that if you make any changes to your Heap events, then those changes get synced downstream to your warehouse, automatically and retroactively. No tracking code or ETL required.

What is the larger impact / societal impact of your product in the space you work within?

Our mission is to power business decisions with truth.

Data is the best lever for making this happen. Unfortunately, getting value out of data still looks more like tedious labor than data science. It too often requires heavy engineering resources. We want to automate away all the technical bottlenecks, so that more people can leverage data to make smarter decisions.

What’s an interesting element of Heap’s company culture?

One of our company values is “Hire for Slope Over Y-Intercept”. (It’s inspired by a talk by Stanford professor John Ousterhout.) We consider hiring our #1 most important job, and we go to great lengths to ensure we do it well.

The reality: assessing potential (“slope”) is much tougher than assessing experience (“y-intercept”). Yet most interviews are designed to measure the latter. At Heap, we wanted to flip this on its head. We make our interviews resemble the actual day-to-day as much as is possible in an interview setting.

For our engineering interviews, we have engineers spend a day designing and building an end-to-end feature. For our sales interviews, we have reps iterate on mock customer calls with us. Even for our management interviews, we have candidates talk to their potential team, find patterns, and produce a plan for improving the team’s execution.

Looking back, what motivated you to start Heap?

Heap was born out of my experience with analytics as a PM at Facebook. I had access to the best analytics infrastructure in the world. Yet it was all completely useless for me.

Why? Because getting the right data took way too much time.

Each time an analytics question came up — or our existing logging broke — I had to:
* Bother an engineer to write tracking code
* Wait for their code to go live (which could take up to a month for mobile apps)
* Wait for data to trickle in
* Bother a data scientist to synthesize the data into a report

This sort of feedback loop made it extremely difficult to actually use data meaningfully. I started wondering: if Facebook couldn’t even get this right at their level of sophistication, what hope did any other online business have?

We knew this wasn’t the future of analytics. With Heap, we decided to capture as much data as possible upfront and eliminate this bottleneck.

Is what you’re working on now the original idea or did you pivot?

It’s still the original vision of automating the annoying parts of customer analytics.

Were there moments where you thought the company might die? Describe one of those and anything you learned from it.

There have been many, many tough periods of time. Yet we’re grateful that we never thought that the company might actually die.

What was a particularly important insight you had about your market that made your product work?

Analytics is a problem rooted in data collection, not visualizations.

There’s lots of exciting work being done in synthesizing data: predictive analytics, nicer visualizations, faster processing speeds, etc. But none of that work matters unless you have a complete, trustworthy dataset in the first place.

And the status quo for collecting data on the internet hasn’t changed since the advent of the web in the late 90’s. You still need to sprinkle your codebase with a “track(‘event’)” API call for each interaction you want to measure. Given how much the internet evolved since then (with cloud infrastructure, machine learning, mobile), it seemed logical that data collection should evolve too.

What’s one piece of advice you’d share with a young founder?

Ask customers for money earlier.

Silicon Valley is a small world. People are afraid to burn bridges. So when you ask someone for feedback on your startup idea, you’re more likely to receive positive feedback than negative feedback. This can lull you into a false sense of product/market fit. Asking for money will surface the real concerns.

You might even get users who enjoy using your product! But you can never truly understand the magnitude of the problem you’re solving until you ask your users to commit money.