# step counter reviews by theme · open data · Angel Step > Open data: 7,859 App Store reviews of 10 iPhone step counters, counted against 21 themes by rating band. Counts only, no review text. Canonical: https://angelstep.app/data/review-themes Language: en Updated: 2026-09-30 Whole site in one file: https://angelstep.app/llms-full.txt This file is the markdown twin of the page at the canonical address above: the same words, without markup. Cite the canonical address, not this file. --- what step counter reviews are about We read 7,859 App Store reviews of 10 iPhone step counters and counted what they talk about. Not whether the apps are good — what their reviewers chose to write about, split by the rating they left. The file is public, machine-readable and free to reuse with credit. Angel StepCute Pedometer, Daily StepsFree · In-App Purchases · iPhone Scan with your iPhone camera ### What only shows up once they're counted - What splits the ratings: losing history appears 5.7× more often in 1-2★ reviews than in 5★ ones — 4.6% against 0.8%. - What everyone brings up either way: step-count accuracy, 16.4% of 1-2★ reviews and 11.3% of 5★ ones — the subject raised most often in both bands at once, by people who liked the app and by people who did not. - The single largest complaint is wearables and watches at 17.8% of 1-2★ reviews; the most common praise is motivation at 15.4% of 5★ ones. ### Every theme, by the rating people left Each cell is the share of reviews in that band that talk about the theme. One review can talk about several, so a column does not add up to a hundred. The 3-4★ band is shown but takes no part in the comparison above — that is how the method defines it. The code beside a name is its value in the theme column, shown where the two differ. Share of reviews mentioning each theme, US App Store, 19 August 2026 — 1,101 reviews at 1-2★, 1,274 at 3-4★, 5,484 at 5★ | theme | 1-2★ | 3-4★ | 5★ | | wearables and watches wearable_watch | 17.8% | 17.7% | 6.4% | | step-count accuracy step_count_accuracy | 16.4% | 16.7% | 11.3% | | ads | 15.5% | 9.8% | 3.3% | | subscription and price | 13.4% | 8.3% | 3.8% | | the daily goal daily_goal | 8.0% | 13.3% | 10.0% | | syncing with Apple Health apple_health_sync | 7.6% | 6.7% | 2.7% | | streaks streak | 5.1% | 7.3% | 8.4% | | losing history data_loss | 4.6% | 3.2% | 0.8% | | friends and competition | 4.4% | 7.1% | 9.0% | | simplicity | 4.1% | 8.8% | 14.0% | | how it looks looks_and_design | 3.9% | 6.4% | 3.6% | | clutter and complexity | 3.6% | 2.0% | 1.3% | | motivation | 3.5% | 8.2% | 15.4% | | widgets widget | 2.5% | 2.2% | 1.4% | | weight and diet | 2.5% | 7.6% | 9.2% | | notifications | 2.0% | 2.0% | 2.8% | | not counting in the background background_counting | 1.1% | 2.6% | 1.1% | | the lock screen lock_screen | 0.7% | 0.1% | 0.3% | | battery | 0.6% | 1.1% | 0.8% | | numbers not matching Health mismatch_with_health | 0.5% | 0.2% | 0.1% | | how walking makes you feel mental_wellbeing | 0.0% | 0.3% | 0.6% | ### How it was measured, and what that costs This is not a random sample, and the limit is Apple's rather than ours. Apple chooses the order — mostHelpful is its own ranking — and the store caps the feed at 500 records per app per sort order. This collection hit that cap, so the largest apps are truncated to 500 out of hundreds of thousands. Read the shares as what the reviews the store shows first talk about, not as what the market thinks. The labeling is manual. Themes are matched by a hand-written dictionary of phrasings, not by a model. What is reproducible is the procedure, not a truth about the category. 8,492 records were collected and de-duplicated down to 7,859 unique ones — people edit their reviews, so the same review can arrive twice with different endings. The ten apps were not picked at random either. Nine were taken by number of ratings; one was taken for position rather than weight — Steps — Simple Pedometer, the only one of the 23 apps across our two datasets whose first App Store screenshot leads with a lock screen widget. We checked all 23 first screenshots — on 27 August 2026, and again on 29 September 2026, when our own app joined the list — rather than say “in the category”: the category is thousands of apps, and we looked at these. ### Using it The file itself is step-counter-reviews.json — 630 rows of app × theme × band, plus the same counts rolled up across apps — or step-counter-reviews.csv, the flat form. The rolled-up figures are not lost in the CSV: every one of them is the sum of the rows that are in it. Every row carries its app_store_id, which is the same key used by the store-listing survey — so the two datasets join, and six of the ten apps here appear in both. The four that don't are not an oversight: the two datasets were selected on different grounds, and that is what makes the pair honest. What is not in the file is any review. No text, no author, no per-review record — only counts. That matters for what we can license: the reviews were written by App Store customers and belong to them. What we publish is the compilation, its structure, the theme dictionary and the method, under CC BY 4.0, with credit to Angel Step, angelstep.app/data/review-themes. How to cite it: Angel Step (2026). What reviews of iPhone step counters talk about [Data set]. Measured 19 August 2026. https://doi.org/10.5281/zenodo.22212783 The dataset also has a permanent identifier — 10.5281/zenodo.22212783 — which keeps working if this site ever moves, and always points at the most recent measurement. ### Who made it Angel Step is a step counter for iPhone built around one number, with widgets for the home and lock screen. It is not one of the ten apps counted here — it shipped on 29 September 2026 and has no reviews yet, and an app with no reviews cannot be in a dataset about reviews. That is also why nothing on this page ranks anything. A participant scoring its own market is an advertisement, and these counts are meant to hold up for someone who has never heard of us. Updated 30 September 2026 the app that counted these. one number, on the home screen and the lock screen.