ASO RESEARCH / EVIDENCE OVER ASSUMPTION
ASO EXPERIMENTS.
Twenty studies explore app charts, ratings, reviews and local markets. Some ideas held up; others did not. Read the methods and limits before putting a finding to work.
What we tested
These studies use observational data, not randomized experiments. Several share underlying data; experiments 20 and 21 examine overlapping evidence from the same case. Experiment 9 was not recovered and is not included.
Do app listing edits follow ranking declines?
Original theory not supported in the acquired sample.
↗EXPERIMENT 2Does US-only competitor research miss local leaders?
Supported in the markets and dates studied.
↗EXPERIMENT 3Does faster rating-count growth predict a chart climb?
Neither rating measure added much predictive value after controls.
↗EXPERIMENT 4Do new top-20 arrivals fall out faster than established apps?
New arrivals exited more often in the matched sample.
↗EXPERIMENT 5Do large app-chart jumps happen across countries at the same time?
Narrow prediction supported; not proof of purely local growth.
↗EXPERIMENT 6Are country title-change counts repeating the same edit?
Audit of repeated observations of the same title transitions.
↗EXPERIMENT 7Are country-labelled rating counts really local?
Measurement audit: country labels do not guarantee local counts.
↗EXPERIMENT 8Do apps climb because competitors leave the chart?
Departing competitors rarely explained most of a large climb.
↗EXPERIMENT 10Are new top-100 arrivals actually returning apps?
Most arrivals had appeared in the same chart within the prior two weeks.
↗EXPERIMENT 11Do high-star reviews still contain questions?
High-star reviews accounted for a substantial share of labelled explanation needs.
↗EXPERIMENT 12Do developer reply counts hide repeated messages?
Five-message dominance theory failed; repeated wording was still common.
↗EXPERIMENT 13Do complaint topics change between app versions?
One app-version case passed; broader theory not supported.
↗EXPERIMENT 14Do translated app listings receive fewer bad reviews?
Report complete, main test blocked by insufficient comparison data.
↗EXPERIMENT 15Do bad-review spikes precede exits from the top 20?
Report complete, ranking-risk theory remains unproven.
↗EXPERIMENT 16Can highly rated apps have mostly bad recent written reviews?
Ten of 45 highly rated apps had mostly bad recent written reviews.
↗EXPERIMENT 17Can strong US app ratings hide lower ratings abroad?
One qualifying app with a large country gap; broad theory not supported.
↗EXPERIMENT 18Can bad-review counts double without the bad-review percentage rising?
One flat/falling-share case among 17 count spikes; repeated-pattern rule failed.
↗EXPERIMENT 19Does an elevated bad-review share persist after a spike?
Eight eligible apps; sample too small for the planned support rule.
↗EXPERIMENT 20When is a streak of bad app reviews unusually long?
One app passed the primary corrected test; broad three-app rule failed.
↗EXPERIMENT 21Does the newest review window change the ASO picture?
One app passed both requirements; follow-up to the same pattern in Experiment 20.
↗