Foreglow/Editorial methodology

How Foreglow researches and compares manifestation apps.

Foreglow makes a competing product. This methodology exists so a reader can see exactly how that conflict is handled, which sources support a claim, where the data stops, and how to report an error.

The rules in one minute

Primary sources first

Use the current App Store or Google Play listing, official product site, official terms and platform support documentation before summaries elsewhere.

Unknown means unknown

If a billing period, platform feature or policy is not stated publicly, leave it blank or label it not disclosed. Do not infer it from the price.

One denominator per claim

Always state whether a number describes reviews, ratings, users, events or visits. Never translate one into another.

No paid rankings

Foreglow does not accept payment or affiliate commission for a position in a comparison.

The app is live

Foreglow launched on the US App Store on 2 September 2026. Availability comes from the live store listing; performance claims still require measured evidence.

Source hierarchy

For a product's current price, publisher, rating, platform availability or in-app-purchase list, the first source is its live store page in the relevant country. A US App Store price is labelled as a US price; it is not presented as global. The product's official website and terms are used for claims about trials, renewal and product positioning. Apple or Google support pages are used for platform-managed steps such as cancellation.

When two official sources disagree, both are reported with the date checked. We do not choose the version that makes a competitor look worse. For example, Stella's landing page described a seven-day trial on 15 August 2026 while its terms described three days. The comparison page records the discrepancy and tells the reader that the in-app purchase confirmation governs the offer for their account.

Public reviews are evidence of what a reviewer wrote, not proof that the underlying event happened exactly as described. Quoted reviews keep the star rating, platform and date close to the text. Quotes are selected to illustrate coded themes rather than to estimate how common an experience is.

How prices are calculated

Recurring plans are converted to a common annual period: weekly price multiplied by 52, monthly price multiplied by 12, and annual price used as listed. Monthly equivalents divide the annual total by 12 and round to two decimal places. Calculations do not include tax, currency conversion, trial discounts or promotional eligibility unless explicitly stated.

An App Store page can expose historical, promotional and currently configured in-app-purchase tiers at the same time. Therefore a list of public price points is not described as the offer every new user will see. If the store page does not label a price as weekly, monthly, annual or one-time, we do not assign a period. The purchase sheet inside the app remains the final source for the account-specific price and renewal date.

The weekly subscription calculator accepts any value and runs locally in the browser. It exists to make unlike periods comparable, not to assert that every weekly subscription is deceptive.

Review corpus and coding

The 2026 research corpus contains 56,780 coded public reviews collected from Apple App Store and Google Play listings, with coverage ending 27 July 2026. Coverage windows and platform mixtures differ by app. Store feeds have retrieval limits. Consequently, the corpus is a sample of retrievable reviews, not a census of every rating or customer.

The pricing-friction analysis restricts its denominator to 4,539 retrievable one- and two-star reviews. Automated multi-label rules mark text associated with cancellation or refunds, paywalls before useful value, surprise trial charges, expensive subscriptions, weekly billing and unclear terms. One review can match several themes. Individual rows overlap; the union of 1,047 reviews counts each matching review once.

These rules support discovery of repeated language at corpus scale. They do not determine whether a reviewer is correct, whether a developer issued a refund, or what share of all users experienced a problem. The public research package includes an aggregate CSV, a concise methodology file, accessible chart descriptions and a denominator-safe citation. It contains no review text, reviewer names, account identifiers or review identifiers.

Where editorial judgement enters

Labels such as “best for journaling” or “best for narrated stories” are editorial judgements based on the product's public feature emphasis and available evidence. They are not scientific scores. We avoid a single best-overall winner because the products solve different jobs and because Foreglow would benefit from declaring itself the winner.

Every comparison should state a meaningful strength of the other product, distinguish verified facts from review themes, and avoid treating missing evidence as evidence of absence. Star ratings are context, not a ranking formula: audience size, rating prompts, platform and age make direct comparisons unreliable.

Foreglow's own product rules receive the same treatment. The public store listing now verifies availability and the advertised US price points; local-storage, export, purchase-sheet behaviour and free-note claims still require direct product evidence. We do not use projected retention, downloads or customer outcomes as if they had already happened.

Automation and human review

Scripts help annualise prices, count records, apply repeatable codes, generate structured data and detect contradictions. Automation is not treated as a source. Material claims are checked against the cited source, and pages are reviewed for denominator language, dates and disclosure before release. Generated structured data describes visible page content; it is not used to add claims a reader cannot see.

Dates, change logs and updates

Volatile facts receive a visible “checked” or “updated” date. A research corpus keeps its historical end date even when a comparison's live store facts change. Important discrepancies may receive a source-and-change log. Page titles and descriptions are changed deliberately rather than repeatedly, so search performance can be evaluated over a stable period.

Price and store checks are refreshed when evidence suggests a change, when a reader reports one, or during scheduled maintenance. A stale date is not evidence that a price is still current. Readers should confirm the final amount in the platform purchase sheet before buying.

Corrections and contact

Factual, arithmetic, accessibility and source-description corrections are welcome. Minor wording changes may be made without a log. A material change to a denominator, result or interpretation receives a new update date and an explanation on the affected page. Historical research is not silently rewritten to match a newer corpus.

Email support@getforeglow.com with the page URL, the disputed sentence and a primary source when available. We also provide accessible chart formats and aggregate-data clarification through that address.

Open the research packageRead the comparisons