Why can two correct percentages tell different stories?
One summary of public onboarding discussions may report confusion among all reviewed threads, while another reports it only among threads describing a completed attempt. The resulting proportions can differ even with identical coding. Reconcile the eligible populations before debating which result is correct, and keep the meaning of every denominator attached to its reported quantity.
Reconcile the supplied onboarding summaries by reconstructing eligible record sets. Label every denominator and avoid population rates from incomplete result pages.Write each denominator in words
State exactly which records could enter the calculation, including topic, time and required evidence conditions. Include unavailable or ambiguous cases in a separate inventory. Exclude hidden denominator changes such as removing neutral cases only in one summary; every omission needs a reason that a reader can inspect and reproduce.
Reconstruct the sets
Use stable identifiers to compare the eligible records for each view. Separate records shared by both from those unique to either. If the underlying result is partial, do not compute an exact population-wide rate from the visible page. Use exact reviewed-subset counts or leave the broader quantity unresolved until the required evidence is available.
Test whether comparison is meaningful
Where a common eligible set exists, recalculate only the quantities supported by that set and label them accordingly. Keep the original views if they answer different legitimate questions. Do not force a common denominator that removes the very distinction the owner needs, such as attempted versus completed onboarding tasks.
Deliver labeled alternatives
Present each result with its population and a plain explanation of the difference. Recommend the view that matches the decision without declaring the other invalid merely because its number differs. If counts cannot be reconstructed, withdraw the precise comparison and provide a qualitative finding rather than estimating the missing denominators from narrative text.
Keep the boundaries clear.
- A proportion is meaningful only with its eligible population attached.
- No example on this page represents a live test or complete platform coverage.
Start with a read-only check: ask your agent to report actual source coverage before collecting records or exporting an audience.
Primary references
Documentation reviewed 2026-11-13. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.