How we evaluate software

Our recommendations come from deterministic rules applied to facts from official vendor sources. Those facts are checked automatically against the vendor's pages on the dates shown; a fact counts as editor-verified only after a named editor reviews it, and we label the two differently. No AI model and no commission data is involved in scoring. Engine version: engine-1.0.0; active scoring configuration: scoring-2026.10.18-1.

1. Evidence

Every product fact (a capability, integration or price) is stored as a claim with its source URL, the date it was checked, the evidence type, plan conditions and a verification status. We prefer official pricing pages, product pages and help-center documentation. We never treat anonymous reviews as specifications.

• Available — the vendor documents it (on all plans, or the product has no tiers).
• Conditional — documented on some plans or with conditions, or the plan is not confirmed.
• Add-on — sold separately.
• Not available — only when a vendor source explicitly says so.
• Unknown — we could not verify it. Unknown is never treated as unsupported.

2. Requirements

Your answers become a list of must-have and nice-to-have requirements using documented rules — for example, three or more technicians adds dispatching as a must-have, and your accounting system becomes a must-have integration.

3. Fit score

Six dimensions with configurable weights (current: operational 30, company size 20, integrations 20, budget 15, implementation 10, industry 5). These starting weights come from our product brief and are not scientifically validated.

• Each requirement scores 1 if available, 0.75 if conditional, 0.6 if an add-on and 0 if the vendor documents it as unavailable. Must-haves count 2× a nice-to-have.
• Unknown requirements are excluded from the score and reduce coverage.
• Score = weighted average over known evidence. Confidence = share of the weighted evidence that is known.
• If confidence is below 60%, we show “Potential Match — More Information Needed” instead of a number.
• A requirement you marked as a must-have that the vendor documents as unavailable caps the score at 49 and labels the product “Does Not Meet a Required Capability”.
• Labels: Strong Potential Fit (80+), Good Potential Fit (65+), Partial Fit (50+), Weak Fit (0+).

The same inputs always produce the same score. Every saved report stores the engine version, scoring configuration version and catalog version used.

4. Budget and cost

For vendors with published pricing, we select the lowest plan that includes your must-haves and seats (technicians + office staff). Month-to-month price within budget scores full credit; within budget only with annual billing scores 0.75; up to 25% over budget scores 0.4. Quote-only vendors are not penalized — the budget dimension is simply unknown for them.

Cost totals include only known costs and are labeled partial when anything is unknown. We never invent implementation fees.

5. Keep, Optimize or Switch

• Keep when you report no must-have gaps and satisfaction of 3/5 or higher.
• Optimize when every gap is something your current product documents (often on another plan), or when satisfaction is low without a specific gap.
• Switch only when an alternative has documented coverage of all remaining gaps without failing any must-have you marked — with timing caveats for contracts and migration complexity.
• Otherwise Optimize: no documented better option yet.

6. Migration complexity

A transparent points heuristic: data categories to move (job and equipment history, maintenance contracts and accounting data weigh most), team size, integrations and locations. 0–3 Low, 4–7 Moderate, 8+ High, or Unknown when you have software but did not tell us what data must move. It is a planning aid, not a quote.

7. Independence

Partner programs, commission terms and referral eligibility live in a separate part of our system. The scoring engine's inputs have no commission fields, and an automated test verifies that changing commercial data cannot change a score. See our affiliate disclosure and editorial policy.