How we test

Inclusion criteria

A tool is eligible for the catalog only if all of these hold:

• Active product. Documentation last updated in the past 12 months, status page green for the last 30 days.
• Paid plan exists. A free-only tool with no commercial path is an open-source project, not a SaaS. Both belong in the catalog — but only when the maintainer signals staying power (commercial license, contributor base, or a funded foundation).
• Real integrations. A public integration list with at least 3 documented third parties. Tools that don't integrate aren't usable in a stack.
• Transparent pricing. Public pricing page OR "contact sales" with a known starting band. Pure quote-only vendors are flagged but not rejected.
• Operates in a SaaSpark category. See the 30-category taxonomy. Tools outside it are filed under "other" and reviewed for new-category potential quarterly.

Grounded extraction

Every field on a tool fiche carries a source_quote — a literal substring of the official site that the value was extracted from. The quote is verified programmatically against the normalized HTML at extraction time. If a value has no grounded quote, the field is empty. We do not let the model invent.

When we publish "Resend's free tier includes 3,000 emails/month", we can point at the exact line on resend.com/pricing that produced that claim — and the timestamp we read it.

Pricing freshness

Every published fiche is re-extracted weekly (Monday 02:00 UTC). Three outcomes:

• No change. The fiche is left untouched, the last verified timestamp is bumped.
• Small change (price delta ≤ 25 %, same number of tiers, name unchanged). Auto-publishes; recorded in the audit log.
• Material change (new tier, removed tier, name change, model change from freemium → paid). Drops into the review queue for an editor to verify before publishing.

Ranking alternatives

Our "best X alternatives" rankings combine three signals:

1. Category match. The alternative must serve the same primary problem as the target brand.
2. Adoption. Tools used in at least one of our published stack templates rank above tools that aren't.
3. Differentiation. An alternative ranks higher when it solves a distinct sub-problem (e.g. "Plausible vs Google Analytics" rewards Plausible for the GDPR-clean angle).

Rankings are re-reviewed quarterly. Recent changes are visible in the per-page "last verified" byline.

Comparisons

Head-to-head comparisons follow a fixed structure: TL;DR with a persona-aware winner, "when A wins", "when B wins", verdict. The first draft is written by Gemini Pro from the verified tool data, then an editor passes it before publishing. We do not publish un-reviewed AI-generated comparisons.

What we never do

• Pay-to-rank. Editorial rankings are not for sale. Sponsored slots, when they exist, will be labelled clearly and placed outside the ranked list.
• Affiliate-driven recommendations. A tool's affiliate program has zero weight in its ranking. The fact that we earn from Lemon Squeezy clicks does not move it up the payments list.
• Republish AI output without review. The only exception is Mode-C weekly refresh, which only flips already- reviewed tools' pricing when the diff is small enough to be mechanical.

Disagree with a ranking?

Email hello@saaspark.dev with the page URL and the specific ranking concern. Editorial appeals are reviewed weekly. We publish corrections with the date, the change, and the reason.