A repeatable method for judging any shopping tool, and the answers we give for ourselves.
Shopping tools are usually reviewed on whether they found a discount during a brief test. That measures luck. A more durable approach is to examine what the tool does structurally, because the mechanism determines the outcome across thousands of purchases rather than one afternoon.
Four questions get you most of the way, and all four can be answered without installing anything.
Some tools replace the referral credit on a purchase at the moment of checkout, including credit that belonged to whoever actually recommended the product to you. This is invisible to the shopper and costs them nothing directly, which is precisely why it is worth knowing about: the cost lands on the creator whose recommendation you acted on.
For our part: this site earns affiliate commission on outbound retailer links, we disclose that on every results page, and commission does not affect placement or ranking. Our browser extension injects no affiliate links into pages you visit and does not alter existing ones.
A tool that always produces a confident answer is not more capable than one that sometimes declines. It is less willing to be measured. The tools worth trusting are the ones whose claims shrink when the evidence does.
We publish our own answer to that question directly on the site: we do not promise to identify a winning retailer in advance. What we do instead is show which specific facts about an offer are confirmed and which are not, and leave the decision with the reader.
When a search returns only one retailer, we mark the result as limited rather than presenting a single row as a comparison. That happens more often than we would like, and saying so is the point.
Ask which retailers a tool saw for the search in front of you, not how many it claims in total across all searches. Coverage is a property of a query, not of a company, and it varies enormously by category.
Ours varies by search. Some categories return several recognised retailers; some return a single marketplace. A tool describing its coverage as a single large number is describing its ambition.
Ask when the tool last checked. A price observed hours ago and presented without a timestamp is a claim about the present built from unknown data.
This is the question fewest tools answer, and the easiest one to verify: if the interface cannot tell you when it looked, it does not know.
Run the four questions against whatever you are considering. Who is paid and for what. Does the claim narrow with the evidence. What did it cover for this search. When did it last look.
A tool that answers all four plainly is not necessarily the one that finds the largest discount. It is the one whose answers you can act on without checking its work.