Methodology

This page explains exactly how a verdict is built, so you can judge it for yourself.

A verdict is calculated, not guessed

We collect the evidence, score each part by fixed rules, and combine the scores with what you told us matters to you. The same product with the same answers always gets the same result. Where we can't find evidence, we say so and leave it out.

What goes into the score (0 to 100)

Some parts count for more than others, and this can differ by category.

• How well the product fits you: your use, priorities and budget.
• The quality of the product itself.
• What you get for the price.
• How today's price compares with what is credible.
• Whether something else would suit you meaningfully better.
• Whether there is a reason to wait.

From score to verdict

The verdict is not a direct mapping from the score. A serious blocker or a much better alternative can override a high score.

• GO: good fit, good product, acceptable price and no major blocker.
• WAIT: suitable, but the price is high, a replacement is confirmed or the timing is poor.
• SKIP: poor fit, poor value, a serious recurring problem, a clearly better alternative or a price far above your budget.

Where information comes from

Sources are not equal. A manufacturer is very reliable for specifications and less so for opinions. A single forum comment is weak evidence on its own, so we look for the same point across several sources.

• Manufacturer pages for specifications and warranty.
• Retailers for current price and availability.
• Independent reviews for measured performance.
• Owner reviews and discussions for recurring praise and complaints.

Confidence

Confidence is High, Medium or Limited. It depends on how many reliable sources we found, how closely they agree, how fresh the data is, whether we matched the exact model and whether we found a current price. We show it separately from the score because a high score with limited evidence is a different thing from a high score with strong evidence.

What we will not do

• We will not invent a price history. If we don't have one, we say 'Price history unavailable'.
• We will not announce a replacement model unless it is reliably confirmed.
• We will not give a verdict when evidence is too thin. No credit is used for that check.

Who calculates what

Scores and verdicts come from fixed rules applied to the evidence. AI is used to read and summarise sources and to write the explanation, not to set the number.