You search for the best password manager, or the best VPN, or the best cloud storage plan, and you get ten pages of rankings that mostly disagree with each other. Somewhere near the bottom of each one, in grey text, is a sentence explaining that the site may earn a commission on purchases made through its links. At that point a reasonable person concludes the whole genre is advertising and closes the tab.
That conclusion is half right, and the half that's wrong is expensive. Affiliate funding is how independent comparison work gets paid for without a paywall — including ours. It creates real pressure on rankings, but it doesn't make every ranking worthless, and refusing to read any of them just means you buy on brand recognition instead, which is a purer form of advertising.
The key takeaway up front: stop asking "is this site biased?" — the answer is always yes, everyone has an incentive — and start asking "is this bias inspectable?" A site whose criteria, weights, and money are all visible can be corrected for. A site that shows you a rank without showing you the reasoning gives you nothing to correct, and should be treated as an ad. Use affiliate reviews to harvest criteria and build a shortlist; never to take a verdict on faith.
What the affiliate model actually pays for
To read these pages properly you need to know the mechanism, not just the disclosure line.
An affiliate program pays a publisher when a reader clicks a tracked link and then does something the vendor values — usually starting a trial or paying for a subscription. The publisher doesn't get paid for being right. There is no clawback if the product turns out to be a poor fit and you cancel after the refund window. The money rewards the click-through and the signup, and is indifferent to whether the recommendation served you. Every distortion below follows from that single fact.
Two further structural details matter:
Payouts are not uniform. Different vendors in the same category offer different commission structures — some pay a one-off bounty per signup, some a share of the first year, some a recurring share for as long as you stay a customer. A publisher's revenue per reader therefore varies enormously depending on which product wins the click, even when the products cost you the same.
Not every product has a program. Some excellent tools — open-source projects, small independent vendors, products sold only through direct enterprise contact — run no affiliate program at all. Recommending them earns the publisher nothing.
Neither fact is scandalous on its own. Both create gravity, and gravity is enough — nobody has to be corrupt for a category's coverage to lean.
The four distortions worth watching for
Outright lying is rarer than people assume, partly because it's legally risky and partly because it's unnecessary. The common distortions are subtler and mostly consist of true statements arranged usefully.
Omission
The cheapest form of bias is the product that simply isn't on the list. A shortlist containing only vendors with affiliate programs isn't false — every entry may be a legitimate option — but it's a filtered universe presented as the whole category. This is the distortion that costs readers most, and it's invisible by construction. The check: before you trust a list, find the well-known option in that category it didn't mention, and ask why.
Rank-order drift
When two products are genuinely close on the merits, and one pays materially better, the tie doesn't break randomly. It doesn't require anyone to act in bad faith either — near-ties are exactly where subjective judgement lives, and judgement drifts toward the outcome that funds the site. This is why published weights matter so much: weights fix the tiebreak in advance, in public, before anyone knows which way it points.
Manufactured testing
"We tested 40 VPNs" is a claim about work performed. Sometimes it describes real hands-on evaluation. Sometimes it describes a writer reading vendor documentation and other review sites. The tell isn't the claim, it's the residue: real testing leaves specifics behind — what the setup was, what broke, what surprised the tester, what a product does badly. A review that reports only advantages, and no friction anywhere, describes a marketing page, not an experience.
Recycled consensus
Because everyone reads everyone, an emphasis introduced by one widely-copied page propagates across dozens of others. Five review sites agreeing feels like corroboration but is often a single opinion repeated, sometimes tracing back to a vendor's own press kit. Agreement is evidence only when the sources are independent — and here they frequently aren't.
A five-minute audit for any review site
Run this before you read the ranking, not after. You're auditing the publisher, not the products.
1. Is there a published methodology, and does it have weights? You want to see the criteria and their relative importance. "We evaluate on security, ease of use, features and value" is a start; stating which of those dominates and why is the real thing. Without weights, a ranking is unfalsifiable — you cannot check it, only believe it.
2. Can you reconstruct why #1 beat #2? Take the top two entries and try to explain the gap using the site's own stated criteria. If you can, the ranking is doing work. If the difference is vibes and adjectives, the order was decided somewhere you can't see.
3. Is the business model disclosed plainly, near the content? A disclosure buried in a footer link is technically compliant and practically evasive. Look for a site willing to say how it earns in the same breath as the recommendation.
4. Are sponsored placements labelled and separated from earned ranks? Paid placement is legitimate when it's visibly marked and kept out of the ranking. A fee that quietly buys a rank position — pay-to-win — is the cardinal sin of the category, and it's the specific thing a disclosure line does not rule out.
5. Does the ranking ever contradict the money? This is the strongest single signal available to you. Look for a site that puts a free tool, an open-source option, or a no-commission product at the top of something, or that tells you not to buy at all in a common situation. Evidence that the publisher will act against its own revenue is worth more than any amount of trust-badge decoration.
6. Is there update discipline? Categories move. Look for dated revisions and for the site correcting itself when it was wrong. A page left silently identical while the market changed is inventory, not journalism.
If a site fails 1, 2, and 4, close it. It isn't a review site; it's a storefront with editorial styling.
How to use affiliate reviews without being steered
Once you've audited, these pages become genuinely useful — as long as you take the right thing from them.
Harvest criteria, not conclusions. The most valuable content on a good comparison page is the list of things worth caring about in a category you don't know well. If four independent-looking sources all treat a factor as decisive — deliverability for email platforms, the logging policy for VPNs, the recovery mechanism for password vaults — that factor is probably real, even if their rankings are compromised. Take the axes; leave the verdict.
Use lists for discovery, your own weights for the decision. Broad "best of" lists are efficient at producing candidates and terrible at producing decisions, because they're weighted for an average reader who doesn't exist. Build your shortlist from them, then score it against your own weighted criteria — the process in our criteria-first comparison framework.
Read adversarially as well as favourably. Reviews are written by people paid when you buy. Support forums, issue trackers, cancellation threads, and migration guides are written by people already paying. Search for the failure modes — "how to export from X", "X billing problem", "leaving X" — and you'll find the friction no review page has any reason to feature.
Verify every specific at the primary source. Prices, tier limits, storage caps, jurisdiction, refund terms: check them on the vendor's own current pages before you commit. A number copied from a review that copied another review can be years out of date.
Our model, plainly
The rule we apply to others applies to us. Top Fully's comparison pages earn through disclosed affiliate links and ads, and through sponsored placements that are labelled as sponsored and never counted as an earned rank. Rankings are computed from published, weighted criteria — the same criteria this blog teaches — so you can check our reasoning against our conclusions. We don't print prices, scores, or specs we can't trace to a source, and the blog carries no rankings at all: the education lives here, the scored verdicts live on the hub. Audit that with the six questions above; it's built to survive them.
FAQ
Are review sites trustworthy if they use affiliate links?
They can be, but the affiliate link isn't what decides it. Trustworthiness comes from inspectability: published criteria with weights, a disclosed business model, sponsored content labelled and separated from earned rankings, and a visible willingness to recommend things that don't pay. A site with all four and affiliate links is more useful than a site with none and no links, because you can check the first one's work.
How do review sites make money if the content is free?
Mostly three ways: affiliate commissions when a reader signs up through a tracked link, display advertising, and sponsored placements paid for directly by vendors. The first two are compatible with honest rankings if disclosed. The third is only legitimate when it's clearly labelled and kept out of the earned ranking — the problem isn't sponsorship, it's sponsorship disguised as merit.
How can I tell if a review was actually written by someone who used the product?
Look for friction. Genuine hands-on use produces specifics: the setting that was hard to find, the import that mangled a field, the thing that works well only after configuration. Reviews assembled from documentation are uniformly positive and structurally identical to the vendor's own feature page, because that's their source. Absence of any complaint is the strongest evidence of absence of use.
What's the difference between a sponsored review and an affiliate review?
In a sponsored placement the vendor pays for the coverage or the position itself, usually in advance. In an affiliate arrangement the publisher earns only if a reader clicks and converts, and the vendor typically has no say in what's written. Affiliate is the weaker conflict of the two, but sponsorship is the more honest one when properly labelled — you know exactly what you're looking at. The dangerous case is either one going undisclosed.
If every list is compromised, how many should I read?
Two or three well-audited ones, then stop. Additional lists mostly add recycled consensus rather than new information. Spend the saved time on your own weighting and a scripted trial of your top two candidates — that produces far more signal than a fourth ranking.
You don't need a review site you can trust blindly. You need one whose reasoning you can check — and your own weights applied to a shortlist that's already been narrowed. Compare criteria-scored shortlists across password managers, VPNs, email marketing and cloud storage at top-fully.com, then judge our ranking by the six questions above.
