Segmenting Performance Data
Segmenting your performance data can be extremely enlightening, and building a culture that thrives on finding insights in segmented data can be powerful.
However, this can also go sideways when a company falls into a pattern of never-ending segmentation and, ultimately, analysis paralysis.
So how do you reap the benefits of this without it becoming a detriment to your progress?
Why it matters
Segmenting data matters because some of the most significant insights to move your business forward continuously are not found in the top two levels.
Most companies know they can't simply look at total trial signups or revenue, but many only go one level deeper — segmenting that data by channel.
When you get deeper into the details, you understand things like:
- the combination of campaign targeting and offer that converts the best
- the countries with the best CAC:LTV ratio
- the makeup of the lift in trial signups from the winning A/B test variant
When your company can routinely uncover these insights from segmented data, it can constantly improve by stopping what doesn't work and investing more into what does.
How to do it
Early on, a company should be segmenting however it can and constantly looking for insights.
Over time, you should settle on some of the core forms of segmentation that provide good insights but don't completely bog down reporting or give you information overload.
These become the standard segmentations known across the company and valuable for ongoing benchmarks and planning.
In my experience, this has looked like:
- paid vs. non-paid acquisition metrics
- revenue, average deal size, and win rate by customer type (SMB, mid-market, enterprise)
- trial signups by ad campaign (with campaigns specific to each core country)
Beyond the company-wide segmentation, each team should have several more segmentations for all the metrics they are responsible for.
This ensures that every possible stone is unturned, and questions of "why?" from the company level segmentation can be answered in more detail without the company needing hundreds of variables presented regularly.
Finding the balance
You don't manage your personal finances as one total amount spent each month or a simple segmentation like 'food' and 'not food.'
At the same time, if you know that you spend ~$100/month at coffee shops, you don't really need to dig into which ones, what sizes were ordered, and so on.
The prior points should give a good balance to the organization, but individual teams need to make sure they have a good balance.
The best way for individual teams to do this is to have their standard segmentations and then have a regular cadence (probably monthly, no less than quarterly) to dive deep and segment deeper to understand the more subtle trends.
What to do next
The point of segmenting your data isn't to have more charts.
It needs to be put into action.
One of the easiest examples would be evaluating the ad performance inside a particular ad group, turning off the one with the worst conversion metrics, and creating a new test variant off of the highest performer.
More advanced applications could include combining multiple factors like country, company size, and customer type to provide a tailored onboarding experience or more granular sales rep assignment logic.
Wrapping up
Take a look at how your team and company overall approach data segmentation.
Do you need to implement more processes around segmenting data to extract more insights?
Do you need to rein it in and establish a few standard ways to segment data across the company?
Do you need to get out of constant analysis mode and start taking action?