Methodology behind Reviato insights
Learn how Reviato builds market-first reports for readers and bounded benchmarks for operators, from source data through evidence selection and publication review.
Data sources and scope
Reviato insights summarize public platform data at market scope. Published review counts describe the total visible review volume attached to listings in scope, while sampled review texts describe the subset analyzed for themes, quotes, and distributions. The two numbers can differ because not every visible review is always part of the text-analysis sample.
Rating benchmarks
Rating benchmarks use current collected Google Maps and Tripadvisor listings, not a census or matched unique businesses. Their published review counts include star-only reviews and listings without analyzed text. Article figures describe a separate, frozen sample of text-eligible reviews, so their totals may differ. Each eligible listing has equal weight. Missing ratings or review counts, disabled listings and misclassified Tripadvisor listings are excluded.
Percentiles and median intervals
The p10, p25, median, p75 and p90 thresholds use the nearest-rank method. The self-check percentile is the share of listings strictly below your value; ties are excluded. Median intervals require at least six listings and use distribution-free 95% binomial order statistics. These intervals are conditional on the collected listings and do not measure collection bias.
Benchmark thresholds and recent history
A benchmark needs at least 30 eligible listings on each platform and 1,000 analyzed article reviews to meet its data threshold. Current listing ratings remain the main comparison. The separate 24-month comparison requires 30 listings per platform and 1,000 dated analyzed reviews. It uses each listing's mean of dated, analyzed reviews, with a full stored-history comparator. Listing mix may differ between periods, so it cannot establish a change in the same businesses.
Aggregation and comparison
Locality, subdivision, and country listings aggregate multiple listings into one comparable snapshot. That pooled view is useful for pattern detection, category comparison, and market orientation, but it is not a substitute for a listing-by-listing audit when a team needs to make a property-level or restaurant-level decision.
What the percentages mean
A praise or complaint share is the fraction of distinct analyzed reviews holding at least one accepted mention of that topic and sentiment. One review counts once per topic no matter how many matching sentences it contains, and a review can praise one aspect while criticizing another, so praise and complaint shares do not sum to 100. Percentage-point gaps are calculated from exact counts before rounding, never by subtracting displayed percentages. Uncertainty intervals describe sampling uncertainty around a share; they do not prove the sample represents every business, nor do they measure classifier accuracy.
Theme and quote handling
Complaint and praise sections come from analyzed review-text samples, not from every published review. Reviato filters out vague fragments and favors complete, concrete examples from different reviews and businesses. An excerpt can illustrate a measured theme, but it cannot establish prevalence or carry the conclusion by itself.
Facts, prose, and review
Code calculates the metrics, selects approved evidence, builds charts and citations, and applies the publication gates. A writing model receives only the small evidence pack needed for one section. It can improve the prose, but it cannot create facts, set the methodology, or approve a page for publication.
Limitations
Individual businesses may differ sharply from the pooled view. When theme analysis is unavailable or platform comparability is limited, the listing states the boundary directly rather than implying confidence the data does not support.
Geography and category handling
City, subdivision, and country labels are used to group listings into a readable market view. Category labels such as restaurants, hotels, cafes, bars, and attractions keep unlike businesses from being blended into one vague hospitality average.
Platform comparability
Different platforms attract different review behavior. Google Maps, Tripadvisor, Booking.com, and Airbnb can reflect different moments in the guest journey, so platform comparisons are treated as directional signals rather than perfect apples-to-apples measurements.
Freshness and interpretation
A market snapshot is strongest when the collection date, platform, category, and sample size are clear. Older snapshots can still help with broad orientation, but recent operator decisions need fresh business-level evidence before the team changes staffing, service, or recovery routines.