Published 2026-10-29
Checked 2026-10-29
DATA SHEET · 2026-10-29
How do we assess reading apps, and what are the limits?
Reading apps are scored on the same eight criteria as every app here, with two adaptations: capture friction becomes the steps needed to tell the app where you are in a book (or to start text playing), and forgiveness focuses on what happens when you come back after a gap. New reading entries are provisional desk reviews: we score only the criteria their official pages give evidence for and show the result as points out of the points available, such as 55/70, with no total out of 100 until we run the hands-on protocol.
Our rubric was written for productivity apps: planners, timers, blockers. Reading tools do a different job, so this page sets out how each criterion translates, what a desk review can and cannot score, and why new reading entries show partial scores.
How does each criterion translate for reading tools?
The rubric, applied to reading and memory tools
| Criterion | Weight | For productivity apps | For reading tools |
|---|---|---|---|
| Setup friction | 20 | Minutes and decisions from install to first real value | Same, including the decision “is my book supported?” |
| Capture friction | 15 | Taps from a locked phone to a saved task | Steps to tell the app your place (a photo, a page number, a phrase) or to start text playing |
| Forgiveness | 15 | What happens after a missed day | What happens when you come back after a gap: expired loans, lost places, lapsed credits |
| Reward design | 10 | Motivating without manipulating | Same; streaks and badges count here |
| Notification quality | 10 | Helpful re-engagement against noise | Same |
| Price honesty | 10 | Trial terms as written against as experienced | Same, plus whether the model charges you through months you do not read |
| Accessibility | 10 | Text scaling, screen reader labels, reduced motion, contrast | Same, plus speed control, highlighting and format support where documented |
| Staying power | 10 | Update cadence over 12 months, team and funding signals | Same |
What is a provisional desk review?
Every score on this site is currently a provisional desk score: it comes from official listings, websites and documentation, and timings and tap counts show as Not yet measured until we run the hands-on protocol. That is flagged on every entry as DESK REVIEW · PROVISIONAL.
For reading entries added from 29 October 2026 we go one step stricter. We score a criterion only when an official page gives evidence for it, and we mark the rest “not assessable”. The total is shown as the points scored against the points those criteria are worth. Magic Bookmark, for example, is 55 of 70: setup, capture, forgiveness, price honesty and staying power could be assessed from its listing; reward design, notifications and accessibility could not.
Why not just give every criterion a number?
Because a number for something the listing does not mention would be a guess presented as a measurement. A partial score is less convenient for tables, but it says exactly what we know. When we install the app and run the protocol, the gaps fill in, the total out of 100 appears, and a dated changelog line records the change.
One consequence: partial scores cannot be compared with totals out of 100, or with each other. A tool at 16/20 is not “better” than one at 55/70. Read the criteria.
What sources count as evidence?
- The app’s own store listing (App Store or Google Play): features, version history, privacy label, ratings.
- The developer’s website, pricing page and help pages.
- Platform documentation for built-in tools (Apple Support, Android Help, Audible Help).
- Dated announcements from the company, such as Amazon’s Kindle Recaps announcement.
Not evidence: user reviews quoted as if they were our tests, forum posts, other review sites, or anything we cannot link. Every page in this category lists its sources with dates.
What have we not done?
- Installed and timed the new reading tools. No setup times or step counts on reading entries are measured.
- Bought credits or subscriptions for them, so price honesty is scored on terms as written, never as experienced.
- Checked how well any recap tool avoids spoilers in practice. We describe what the developers say.
- Assessed reading comprehension or memory outcomes. We score software mechanics, not effects on ADHD.
How do corrections work?
If a fact here is wrong, email the corrections address in the footer with the page and the official source. We fix it with a dated changelog line on the page it affects, as described in our editorial policy. Prices are re-checked and dated on the price tracker.
Questions people actually ask
Did you test Magic Bookmark?
Not hands-on. It is a provisional desk review from its App Store listing and website. Five criteria could be assessed from those sources, worth 55 of 70 points; the rest are marked not assessable.
Why do some reading apps show 55/70 instead of a score out of 100?
Because we score only the criteria an official page gives evidence for, and show those points against the points available. A total out of 100 appears after a hands-on test.
Can I compare a partial score with a full score?
No. Partial scores cover different criteria from app to app. Compare the criteria rows instead.
Do you measure whether reading apps improve comprehension?
No. We score software mechanics such as setup, forgiveness and price honesty, not effects on reading or on ADHD.
Sources
Related
We may earn a commission when you buy through links here. Commissions never change a score or a verdict. How we make money →