This post is written by Jonah Leshin, Chief Data Scientist at Highland Math.

Over the past year and a half at Highland Math, we’ve had the privilege of working closely with a variety of data companies. We’ve analyzed user and campaign level data, off of which we’ve built insights and audience segments that help companies grow their data businesses. In many cases, our deliverables have synced with client needs, and the value creation is evident; at times, however, we’ve also been off the mark. With this experience in mind, I wanted to offer a view into the challenges and opportunities that we’ve seen presented to data companies. I’ll focus on three facets of our business that have driven growth, and three that haven’t worked as well.

Driving Growth

1) Delivering insights from the perspective of the data company

This may sound obvious, but taking the perspective of the data company is a mindset we have worked to develop. We have a view into the entire data marketplace, so a natural inclination is to generate insights through the lens of a bird’s eye view. By putting ourselves in the data company’s shoes, we have helped companies generate growth. For example:

  • Creating individually tailored peer groups for each data provider we work work with.
  • Slicing the data market by specific categories and sources– b2b, location data, etc– that are highly relevant to certain clients.

2) Repurposing data

Data companies often have a wealth of data, but may only be monetizing it along a select few dimensions. And given the complexity of the data ecosystem and the overhead of managing big data workflows, it makes sense that a data company would focus on shaping its data to fit the channels and categories that are proven winners.

We’ve been able to create value by inserting ourselves into these data workflows.  We use our market level data to extract new signals from a client’s existing data and deliver to new endpoints, which in turn generates incremental revenue.

3) Connecting the dots

We track metrics across a variety of dimensions, and the relationships between these metrics are not always evident. For example, how, if at all, might Trade Desk growth be related to a shift in CPM distribution across certain segments? We cannot always draw a direct line from point A to point B, but being able to provide plausible explanations around causality that are backed by data has enabled us to help clients develop their data strategy.

Lessons Learned

1) Going deep or abstract off the bat

We strive to succinctly communicate insights that span multiple cross sections of the data economy. We have missed the mark at times by trying to combine several pieces of nuanced information into a single chart or graph, especially if we have yet to build visualizations of the individual components of the more complex visualization. Along similar lines, it can be tempting to abstract multiple distinct metrics into a single measurement; however, when this measurement is not easily explainable in terms of a concrete metric such as revenue, confusion can arise.

2) Short-term workflows

Not surprisingly, data companies move data around a lot, and we’re able to lend a hand building some of these pipelines. In cases when we’ve helped ship a few heavy pieces of data without building sustainable infrastructure, the client receives a short-term boost, but is unable to carry value from us moving forward. We’ve had more success with short-term projects when they’ve been centered around insights we can generate from our market level view of the data space.

3) Working with small scale data

As a data science company ourselves, we take pride in being able to extract signals from messy data sets. A natural extension of this ethos is the ability to extract revenue from data by exploiting these signals while also optimizing channels and workflows. We have been reminded, however, that volume of total available raw data is an essential component of any calculation, for which there is no true substitute.

Moving Forward

With the ever-increasing volume of data and accompanying new technologies springing up left and right, there is no shortage of exploration opportunities for data companies. At Highland Math, we’ll draw on our experience to help data companies elucidate, generate value from, and ensure compliance with an expanding data landscape.