Change Regression Detector
Automatically links performance drops to the page changes that caused them
- agent ›Regression on the checkout's delivery step: conversion fell from 61% to 52% in the 48 hours after yesterday's change.
- agent ›Evidence: the delivery options were reordered, and 'Pick-up point' is no longer pre-selected.
- you ›We changed four things on that page yesterday. Why blame the delivery order?
- agent ›The other three are copy edits below the options, and clicks on them didn't move. Clicks on the delivery selector rose 34% — people are hunting for the option they used to get by default.
- you ›Our carrier wants home delivery listed first. Can't we keep it?
- agent ›Keep it first, but pre-select pick-up point again — that's the part that converted. Leaving it as it is costs about DKK 9,800 a day at today's traffic.
What it does
Runs after every crawl. For each page where Historian found a direct content change, it compares the 24–48 hours after the change with the period before it: click rates on the page's elements, conversion rate and revenue. It looks at the pages that changed, not the whole site, so a regression on one checkout step is not drowned out by a good day elsewhere. When the numbers drop, the insight carries the evidence: the page, the element that changed, the kind of change (copy, image, order, removed or added), and the before-and-after figures — enough to decide whether to roll back without opening a dashboard.
An example insight.
What a result from Change Regression Detector looks like. The numbers are an illustration, not a customer's.
Last deploy made the mobile checkout CTA hard to see
The latest deploy changed the primary button on the mobile checkout page from dark green to light grey. Its click rate dropped 22% in the 48 hours after the change; the rest of the page held steady.
Related features
All features →History and change detection
See your site exactly as it looked on any day since Historian started capturing it, with that day's numbers and the events around it.
Version comparison
Put two periods or two versions of a page side by side, each with its own heatmap and its own numbers.
Explain
When a number moves, Explain breaks the movement down by products, traffic sources and devices, so you see what drove it.
More Action & Recommendation agents
All Action & Recommendation agents →Bad Change Detector
Detects whether any page changes correlate with performance drops
Content Error Detector
Catches embarrassing content mistakes on key pages before customers see them
CTA Optimisation Advisor
Surfaces every underperforming call-to-action with a concrete improvement brief
Navigation Improvement Agent
Finds where users get lost and recommends structural navigation fixes
Broken Page & Redirect Advisor
Finds pages receiving traffic that no longer exist or lead nowhere useful
A/B Test Opportunity Finder
Identifies the elements most worth testing with expected uplift estimates
Product Description Enhancer
Finds products where thin content is wasting organic traffic
Campaign Freshness Monitor
Detects expired promotions left live on key pages
Mobile UX Gap Agent
Finds every UX element that works on desktop but fails on mobile
Template-Wide Fix Advisor
Identifies issues affecting entire page types — one fix, hundreds of pages improved
Newsletter Strategist
Weekly, data-driven newsletter recommendations based on trends, top products, and content.
Every category
The agent catalog →Strategy
Know what to focus on
9 agentsRevenue Opportunity
Find the money hidden in data
12 agentsAction & Recommendation
Specific things to change today
12 agentsCompetitive Intelligence
Know what the market is doing
5 agentsRedesign
Before, during and after a redesign
4 agentsPlatform Migration
Move platforms without losing revenue
4 agentsOther
More agents in the catalog
6 agentsStop guessing why revenue dropped.
Book a demo and we'll show you what Historian finds on your own store.