Practical framework / 5 MIN READ
Check whether the rate changed or the mix changed
Rebuild aggregate rates from their counts and inspect stable groups before explaining the headline.
AI-assisted research and analysis for the SNS Creative Fieldbook
Sources checked 2026-11-06
An overall conversion rate can fall even when every meaningful group improves, because more traffic moved into a lower-converting group. The aggregate describes the business result, while the group analysis helps explain its composition.
What you will leave with: Calculate weighted rates and a transparent fixed-mix comparison without confusing either with a causal estimate.
Use this when
- The headline trend conflicts with results inside major segments.
- An acquisition expansion changes the balance of new and returning visitors.
This is descriptive decomposition, not permission to claim that segment membership caused the outcome.
Reconstruct the total from counts
Start with the numerator and denominator for each mutually exclusive group. Add the numerators, add the denominators, then divide. Averaging group percentages gives each group equal weight regardless of size and can produce a number that does not describe the total population.
Google Search Console’s CTR definition is a concrete example of a count ratio: clicks divided by impressions. The arithmetic lesson applies to an explicitly defined custom rate too, but the relevant counts must share the same unit and scope. Do not combine account outcomes with visitor denominators.
Choose groups that explain the business question
Use a small set of stable, meaningful groups such as new and returning customers or eligible market areas. Confirm that everyone appears once and only once. Overlapping campaign audiences cannot simply be added without deduplication.
Display each group’s rate and its share of the denominator in both periods. This makes the two mechanisms visible: performance within groups and changes in group weights. Avoid creating dozens of tiny slices after seeing the result. The aim is an understandable explanation, not an exhaustive search for a flattering subgroup.
Calculate a clearly labeled fixed-mix view
To ask what the overall rate would look like under an unchanged composition, apply one declared set of group weights to both periods’ rates. State whether those weights come from the earlier period, a pooled population or a relevant planning population. Different choices answer different descriptive questions.
Keep the actual aggregate beside this standardized view. The business experienced the actual mix, so standardization should not erase the resulting workload or revenue. It is an explanatory calculation, not proof that performance would remain unchanged if the team actively altered the mix.
Connect the explanation to a proportionate action
If within-group rates improved but the aggregate fell, investigate why the lower-rate group expanded and whether its volume is strategically valuable. A lower rate can accompany more total customers. If a high-value group deteriorated, the same headline could conceal a more serious problem.
Bring costs, value and uncertainty back into the decision. Small group samples can produce unstable rates, and segment definitions can drift. Document those limits. The useful conclusion identifies what changed and what question to examine next rather than reducing performance to one flattering or alarming percentage.
Make the reasoning concrete.
In a hypothetical period A, returning visitors convert 40 of 100 and new visitors 10 of 100: 50 of 200, or 25%. In period B, returning visitors convert 45 of 100 and new visitors 60 of 400: 105 of 500, or 21%. Both group rates improved, from 40% to 45% and 10% to 15%, while the aggregate fell.
Actual mix
Opening: Use 105 conversions divided by 500 visitors.
Hypothesis: The 21% rate describes period B as experienced.
Risk: It does not isolate changes within groups.
Fixed equal mix
Opening: Weight period B’s two rates equally, as in period A.
Hypothesis: The standardized rate is 30%.
Risk: This is a hypothetical composition, not the actual business total.
Report that conversions increased from 50 to 105, the actual rate fell with a larger new-visitor share, and both group rates improved. No campaign causality follows from this arithmetic.
Take the questions into your next brief.
Use these prompts in your own notes. Nothing is submitted from this page.
Groups
Are the groups stable, exclusive and collectively complete?
Counts
What numerator and denominator belong to each group?
Weights
How did each group’s population share change?
Standardization
Which reference mix answers the explanatory question?
Action
Does the new mix add value despite a lower aggregate rate?
Watch for the wrong conclusion.
- Averaging percentages without their denominators.
- Adding overlapping audiences as if they were exclusive.
- Replacing actual outcomes with a flattering standardized figure.
- Treating a descriptive mix calculation as a causal effect.
Your next action
Rebuild one headline rate from two or three meaningful groups and show the change in both rates and population shares.
Sources & scope
Frameworks, worksheets and hypothetical examples are original editorial synthesis. Linked primary sources support only the stated concepts. Source checking is not an independent audit of effectiveness. No named human author or reviewer is claimed.
- Google Search Console Help: Performance report overview →Primary platform documentation · Supports: Definitions of clicks, impressions, CTR and average position. Supports these Google Search measures only; the editorial measurement workflow is not a Google recommendation.