If your conversion rate falls, the buying journey has not necessarily deteriorated. The proportion of different visitor groups may have changed instead. Comparing orders and visits by source helps you distinguish those possibilities, provided visits and orders are defined consistently across both periods.
Worked example: more orders, a lower rate
The figures below are hypothetical assumptions. They illustrate a mix effect, not a typical conversion rate for any channel.
| Visit group | Before | After | Purchase rate in both periods |
|---|---|---|---|
| Group A | 1,000 visits, 40 orders | 1,000 visits, 40 orders | 4% |
| Group B | 1,000 visits, 10 orders | 3,000 visits, 30 orders | 1% |
| Total | 2,000 visits, 50 orders | 4,000 visits, 70 orders | 2.5% → 1.75% |
Group B grows from half of all visits to three quarters. Its lower rate therefore carries more weight in the total. Whether those additional visits are commercially worthwhile also depends on acquisition costs, later purchases and contribution margin.
Use dimensions that belong together
GA4 distinguishes information about the first user contact from information about the current session. Google’s traffic-source guide explains these scopes. A comparison of session groups needs appropriately defined dimensions; mixing “First user” with “Session” fields may answer a different question.
To compare the same thing, keep the period, timezone, visit definition, counted outcome and source dimension together in the analysis. Changes to consent, filters or event triggers matter too: if analytics captures only some visits, dividing all shop orders by that partial denominator would produce a different rate.
Examine the group behind the decline
A few distinctions relevant to your offer are enough to begin. Perhaps more first visits arrive through a guide, a promotion advertises a different product type or mobile visitors encounter a broken selector. Those connections are more useful for diagnosis than an arbitrarily detailed division of all visits.
A channel name alone cannot tell you whether someone is ready to buy or still comparing. A search for a replacement part may be close to a purchase, while another serves initial research. Including the landing page, advertised promise and subsequent journey helps you interpret that difference without assigning a fixed buyer quality to the whole channel.
In small groups, a few purchases can move the rate substantially. Show the numerator and denominator beside it. Comparing 2 out of 50 with 3 out of 50 needs a different interpretation from a consistent difference across large, comparable groups.
Turn the comparison into a task
If the same journey deteriorates within comparable groups, investigate the relevant point. If group rates remain stable but their proportions change, examine acquisition and visitor intent first. Both situations can occur together.
Record the finding alongside the absolute figures before turning it into a redesign brief. For mismatched counts, use the guide to reconciling analytics with your shop system. If the comparison itself remains unclear, an impact analysis can be commissioned as a defined part of our CRO work.
