How to compare marketing performance before and after a change
A before-and-after comparison should use compatible definitions, time windows and audiences while recording other changes that could affect the result.
AI-assisted practical guide · Editorial approach
The useful starting point
A before-and-after comparison should use compatible definitions, time windows and audiences while recording other changes that could affect the result. It can show an association, but rarely proves that one change alone caused the outcome. Interpret the evidence at the strength it supports.
Understand the decision
Document the release date and meaningful changes in offer, traffic source, season or receiving capacity. Check whether tracking remained consistent across the period. Include absolute counts and receiving quality, not only percentage differences. If the sample is small or several factors changed, describe the result as preliminary rather than a definitive performance claim.
A practical approach
Define each event and business state before implementing tracking. Distinguish page visits, contact clicks, form attempts, accepted enquiries and qualified opportunities. Match reports to these definitions. Test whether an interruption or retry creates duplicate events. Collect only the information needed for the agreed measurement purpose; contact details and free-text enquiry content should not be copied into general analytics events.
- Record the change and concurrent factors.
- Keep metric definitions consistent.
- Use comparable observation periods.
- Include counts and alternative explanations.
Illustrative example
A hotel updating its enquiry page during a seasonal promotion might see more requests. The page change and the promotion are competing explanations. The team would inspect request relevance, task errors and follow-up evidence instead of assigning the whole difference to the redesign.
A mistake to avoid
Do not publish a causal case-study headline from a simple time comparison without supporting evidence. The limits belong next to the result.
What to review
Review consistency across observations and the specific journey improvements the change was intended to deliver.
Turn the guide into a working brief
Write a measurement dictionary containing event names, trigger conditions, exclusions, owners and known gaps. Compare the analytics report with the receiving system for the same period and document differences rather than forcing the numbers to match. Use small experiments with a written hypothesis and one primary decision. Report uncertainty, missing consent and unavailable data plainly so the team does not mistake a partial view for complete attribution.
Common questions
What should I prepare before asking for help?
Start with this checklist: Record the change and concurrent factors. Keep metric definitions consistent. Use comparable observation periods. Include counts and alternative explanations. Add your business context, existing materials and the decision you need to make. A useful initial brief can include uncertainty; you do not need to invent answers before discussing scope.
How should I judge whether the work helped?
Review consistency across observations and the specific journey improvements the change was intended to deliver. Keep the observation period and definitions visible. Use the responsible team's evidence alongside the website journey; do not attribute every change in results to one asset or article.
Sources & scope
AI-assisted practical guide. Examples are illustrative; business-specific facts and sector claims need the responsible owner’s approval.
- W3C WAI: Forms Tutorial
Reference for accessible labels, instructions and feedback. Original business examples are illustrative, not research findings or verified client results.
Use this guide to start your brief.
Share your business context, existing materials and what is still unclear. The guide title will be included so the conversation starts with the right problem.
Prefer email? Send a brief with this guide’s context.Related guides
Clicks, enquiries and qualified opportunities are different metrics
A click shows an interaction, an accepted enquiry shows confirmed receipt and a qualified opportunity shows assessed commercial fit.
Read the guide →A measurement plan before redesigning a website
A measurement plan defines the problem, intended improvement, observable events and commercial review before a redesign begins.
Read the guide →Campaign attribution when the data is incomplete
Attribution estimates which touchpoints are associated with outcomes under a particular model and available data.
Read the guide →