Methodology: how we collect mentions

Sources

Where mentions come from

We read open affiliate chats, industry forums and public review sites. A message qualifies only if it is visible without registration and without an invite to a closed community.

Selection

What makes it into a report

A message has to name the tool and describe a concrete experience: what the author searched for, what worked and what did not. One-word verdicts are dropped. Texts are abridged and handles are changed at the authors' request.

Sentiment

How we label

Each mention gets one of three labels — positive, neutral, negative — based on the author's overall conclusion rather than a single sharp phrase. Ambiguous cases go to neutral.

Limits

What we do not do

We do not verify claims about charged amounts: they cannot be confirmed from public sources. Competitor ratings come from public Trustpilot profiles and always carry the date they were read.

Scale

Why there is no number

We do not award tools a numeric score. Sentiment tagging yields labels, not a quantity: an open corpus has no full population behind it, so no share can be computed honestly from it — and with no share there is no arithmetic that turns labels into a score. Until August 2026 the SpyTrend report did carry such a figure in its header, with no procedure a reader could work through. We removed the figure rather than inventing the procedure after the fact. Competitor ratings stay, because those are other people's public numbers with a stated reading date: we did not compute them, and you can check them through the link.