What if Numbers Could Talk? How Data Segmentation Gives Numbers a Voice

by Etienne Garbugli no comments.

✉️ Sign up for Updates







The idea behind Highlights first popped up when Ludo and I were still working at LANDR.

At that time, I was in charge of analytics tasked with understanding business growth and weekly performance.

With aggressive growth targets, the business was actively trying to gain market shares internationally.

While scaling our efforts, we were signing up more users than ever, and our conversion rates had remained about the same.

At the aggregate level, things were looking great, but when drilling down looking at longer-term trends, something had clearly changed.

Using Data Segmentation to Investigate ?

Now, the challenge with uncovering data patterns is that, as an analyst, it’s hard to know what you don’t know.

Sure, there’s common data segmentation analyses you can run to figure out what’s going on:

  • Traffic Sources: Are the behaviors different by acquisition channels? Did the breakdown of acquisition channels change?
  • Location: Are the behaviors different per country or per region?
  • Devices: Do behaviors vary per devices and/or browsers?
  • Languages: Are there differences in the way native and non-native English speakers behave? Do site languages affect on-site behaviors?

It’s a good start, but when none of these analyses give you the answer you seek, you have to dig deeper: What patterns do you look for? What is noise? What is signal?

It took some time, but we eventually realized that payment gateways and payment failures had been a blindside for us. It turns out that, what instinctively looks like a good conversion rate for a market, can actually be subpar.

We learned that credit cards and payment can have major effects on the ability to do business in Germany, the Netherlands, Brazil, etc.

The Cost of Having the Wrong Customer Data Segmentation

In October 2017, I decided to dig deeper to see if I could help growing businesses identify blindsides when expanding internationally.

Over the course of a few weeks, I spoke to dozens of business and growth leaders.

One of the stories that stood out was the story of an European business that made the strategic decision to expand in Latin America.

The decision made sense on many levels: 1) they had really good acquisition, activation, and retention metrics in Spain, and 2) they were able to acquire customers for cheap in LatAm.

As the company’s footprint grew in South America, they decided to focus on the Colombian market ??, the second largest Spanish-speaking country.

In Colombia, customer acquisition was cheap and their engagement metrics looked good. What the team failed to realize while they were burning through their runway was that most credit card payments were not going through.

It turns out that, many Colombian credit cards can’t make international payments. This issue went undiscovered among the noise created by successful payments in other countries.

Hindsight is Always 20/20, Data Segmentation is Not

The problem with insights like these is that they look obvious in hindsight, but aren’t when you’re deep in the business.

With hundreds of ways to slice and dice data, and analysts and teams not knowing what they don’t know, similar patterns often go undetected.

Although we didn’t end up building those features, they’re still part of our vision for Highlights.

How to Make Numbers Talk with Data Segmentation

Too often, businesses rely on aggregate data to make decisions.

Total number of users, visitors, conversions, or emails sent show you the trajectory. The real insights come when you segment the data.

With data segmentation, you can make numbers talk and understand the context behind them. Context in the form of:

  • Timing: At what what stage of their customer journey are your visitors? Are they ready to buy?
  • Trajectory: Can it be repeated? Is it growing or slowing down? Will 100 be 85, or 120 next week?
  • Customer Lifetime Value: How valuable are the visitors or customers in this segment?
  • Demographics: What is the age or gender of your visitors or customers? What other information do you have on them?
  • Psychographics: What are the personas or profiles of your visitors?
  • Culture: To what culture or country do visitors or customers belong? What are their native tongues?
  • Engagement: How engaged are they with the offering? Is this their first visit or are they repeat customers?
  • Advocacy: How satisfied are they with your brand and product(s)? How likely are they to recommend the offering?
  • Site Performance: How well does the website perform for them? Is it fast enough?
  • Traffic Sources: How did these visitors or customers discover the offering? How likely are they to bring in other visitors?
  • Conversion Profile: Did they convert? How likely are they to buy, signup or engage?
  • Willingness to Pay: How much can they spend on your product(s)? Can they afford them?

Context can be a lot of things. Things you’re keenly aware of, and things you’re not. It’s hard to be aware of all the factors that can affect your interpretation of the numbers.

How Intelligent Data Segmentation Helps You See More Clearly

The promise of artificial intelligence is that computers can quickly analyze large quantities of data and leverage learnings. This allows them to recognize patterns more effectively than the human eye can.

We take this promise to heart at Highlights. We show it in 3 key ways:

1. Traffic Channels

We already talked about the importance of analyzing and understanding traffic sources to grow your site traffic.

Well… Highlights no longer looks at aggregate traffic data. ? Our algorithm now breaks down all traffic numbers by traffic sources:

Data Segmentation – Traffic Source Analysis
Highlights Data Segmentation – Traffic Source Analysis

One thing we discovered is that traffic channels have different potentials to generate repeatable traffic. For example, organic is more likely to drive consistent traffic than referrals. That’s one of the reasons why we ❤️ organic traffic so much.

This realization led to the creation of a custom metric we lovingly refer to as LTTR, or the Likeliness of Traffic to Repeat.

By breaking down traffic sources and factoring in each segment’s LTTR, we can determine whether a landing page or blog post will become an Iceberg… or a Burb and Fizz, impacting site traffic only for a few days.

2. Traffic Patterns

Knowing the kinds of traffic a page is getting is good, but being able to understand and project traffic acceleration or decline is better.

We spent a lot of time refining Highlights’ statuses and trends. The work we put in allowed us to project the trajectory of traffic and performance for a page. This helps us determine which pages are worth promoting.

Data Segmentation – Traffic Patterns
Data Segmentation – Traffic Patterns

3. Goals & Conversions

Highlights also factors in conversions and micro-conversions via Google Analytics Goals.

Data Segmentation – Goals & Conversions
Data Segmentation – Goals & Conversions

By assigning scores to the pages that are converting, and those that should convert, we’re able to help our users optimize performance for the emails and landing pages that matter most to their business.

Channels, patterns and goals are weighed in by Highlights. This helps give our users a clear picture of the performance (and the potential for performance!) of their content. It’s the kind of analysis that takes hours to do well, and Highlights makes it available in seconds.

Highlights – Prioritization
Highlights – Prioritized Opportunities

Go beyond aggregates. Focus on data quality. Segments are the key to understanding true performance and doubling-down on the right campaigns.

Highlights, making numbers talk since 2018. ?

Etienne Garbugli
Étienne is a three-time startup founder (Flagback, HireVoice and Highlights), and the author of Lean B2B: Build Products Businesses Want. At Highlights, Étienne is responsible for customer success.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.