Refer App Documentation

Insights

Understand the quality of your attribution data, not just the volume.

Pro

Insights answers a different question from every other tab. Rather than "how much traffic did we get?" or "where did it come from?", it asks: "how well does our attribution data actually reflect reality?" A site can have thousands of sessions but poor attribution if most of them are landing as Direct because UTM parameters aren't being used.

Health Score

The Health Score is a single number from 0 to 100 that summarises the overall quality of your attribution setup. It's the average of four equally-weighted components:

ComponentWhat it measuresMaximum
Attribution Coverage What percentage of sessions have a classified channel (not Direct or unknown) 25 pts
UTM Completeness What percentage of sessions have at least one UTM parameter — a signal that someone is deliberately tracking those links 25 pts
AI Taxonomy Activity Are AI Referrer sessions being classified? A non-zero rate suggests the taxonomy is working 25 pts
Conversion Tracking Has at least one conversion been recorded? Binary: either wired up or not 25 pts

Example scores

 Score 85/100  ================================== ·····

 Attribution Coverage  100%  ############################  25/25
 UTM Completeness       72%  #####################         18/25
 AI Taxonomy Activity   85%  ########################      21/25
 Conversion Tracking   Yes   ############################  25/25

A score below 50 usually means either UTM parameters aren't being added to marketing links (low UTM Completeness) or conversions haven't been set up yet (Conversion Tracking at zero). Both are actionable gaps, not permanent fixtures.

Attribution Coverage

The percentage of sessions that the classifier was able to assign to a specific channel with a known source. Sessions classified as Direct without any UTM parameters are technically attributed (Direct is a valid channel), but they carry no information about why that visit happened — they're a gap in your data, not an explanation.

To improve Attribution Coverage:

  • Add UTM parameters to all links in your email campaigns, social posts, and paid ads
  • Check that your ad platforms are passing utm_medium=cpc or utm_medium=paid_social (many do this automatically, but some need manual setup)
  • Review the AI Taxonomy to ensure new AI tools are listed

How People Convert

Once conversions are recorded, this panel shows two behavioural patterns:

  • Average page depth before converting — how many pages did a converting visitor view from their first touch to their first conversion? A depth of 1 means they converted on the landing page. A depth of 4 means they browsed three more pages first.
  • Entry page = conversion page — what percentage of converting sessions converted on the exact same page they landed on? A high percentage suggests your landing pages are doing conversion work directly; a low percentage suggests visitors need to navigate before they're ready to act.

These two numbers together describe your conversion funnel in a way that raw counts don't. A high average depth with a low entry-equals-conversion rate suggests visitors are researching before converting — optimise the pages they're visiting along the way, not just the landing page.