Do trust badges work? The four that do nothing
The most-quoted trust badge numbers trace back to companies that sell badges. Here is what the three studies which published their methods actually found.
Trust badges work, but only the ones your visitors already recognise, and the published evidence for any of it is thinner than the badge vendors let on.
Ask do trust badges work and you get a wall of numbers. Forty-two percent more sales. Twenty-three percent higher conversion. Follow the citations and most end at a company selling badges. So we read the three studies that published their methods. They disagree, and that is the useful part.
Which badges are on your site right now?
Open your checkout on a phone and scroll to the card fields. Count every seal, padlock and logo visible without scrolling. Then ask one question about each: has your customer ever seen this mark anywhere outside your site? If the answer is no, it is decoration.
Not sure which of yours a stranger would recognise? Revslip checks the trust signals on your URL.
Why does a badge change anything at all?
Because almost nobody can judge whether a payment form is really secure. Baymard Institute's checkout testing finds users go on gut feeling, and that the gut feeling tracks how visually solid the card section looks rather than anything technical. A badge is a shortcut to a feeling.
Which is why one finding should end most badge arguments. In Baymard's seal study, a homemade seal, invented, issued by nobody, beat real SSL seals, Norton excepted. Comodo, an actual certificate authority, scored 2.3%.
Baymard's 2025 survey of 1,026 US adults found 19% had abandoned a checkout in the past three months because they did not trust the site with their card details.
Do trust badges work, according to the research?
Three published studies say yes, no, and it depends. Baymard and CXL both measured which seals people say they trust. A peer-reviewed experiment measured whether seals changed purchase probability and found no effect on average. The one variable that survives all three is recognition, not the badge.
Start with the one that measured behaviour rather than opinion. Sebastian van Baal, at the Cologne Institute for Economic Research, ran a 300-subject lab experiment on two seals common in European ecommerce. His 2015 paper in Electronic Commerce Research reports that neither seal was effective on average. Split them and it changes: the widely known seal lifted purchase probability, the obscure one did nothing.
CXL surveyed 2,100 Americans and found PayPal on top, chosen by 25%. Baymard puts Norton first. They disagree on the winner and agree on the mechanism: whichever brand the audience knows wins. CXL adds the limit itself, that brand recognition does not lead to user behaviour.
Then the awkward number. Asked unaided, CXL's respondents named only four seals above 4%: PayPal 17%, BBB 7%, Norton 6%, Google Trusted Store 4%.
- 19% abandoned a checkout over card trust, Baymard, 1,026 US adults, 2025
- 300 subjects in the only peer-reviewed purchase-probability test, no average effect
- 17% unaided awareness for PayPal, the best-known seal in CXL's survey
- 2.3% for Comodo in Baymard's comparative seal test
The case against taking our word for it Baymard sells UX research and counts Norton's parent among its clients. CXL sells CRO training. Revslip sells audits. The only source here selling you nothing is the Springer paper, and that is a 300-person lab study from 2015.
Which four trust badges do nothing?
Four patterns turn up on site after site with no evidence behind any of them: a seal nobody recognises, a badge parked in the footer, a text-only security promise with no icon, and any badge at all on a page with a visible layout bug.
- The seal nobody knows. Van Baal's obscure seal moved nothing. Inflow's 2018 round found several security badges actually lowered conversion rates on client sites.
- The footer badge. Baymard finds people judge security section by section, not page by page. A seal three screens below the card fields is not in the room when the doubt lands.
- The text-only promise. Inflow's 2022 test ran an SSL reassurance sentence against a plain graphic reading "You are safe". The sentence landed 0.9% below control. The graphic landed 1.2% above.
- Any badge on a broken page. Baymard has watched subjects abandon Amazon and Walmart over checkout layout quirks, believing the site was hacked. No seal survives that.
What to put beside the card fields
Pick one or two marks your specific audience would recognise, put them inside the visual block that holds the card fields, and fix the layout bugs before you add anything. Baymard's tested recommendation is one to two icons within that encapsulated area, not a row of six.
- Fix the bugs first. Check the card step at 360 pixels wide, in two browsers. A misaligned field costs more trust than a seal buys.
- Encapsulate the card fields. Give them a border or background the rest of the form lacks. Baymard found 89% of major sites failing this in 2012, still 66% in 2016.
- Add one or two marks inside that block. Recognisable beats accurate. A plain padlock outperforms a certificate authority nobody can name.
- Move before you add. Inflow lifted conversions 2.9% at 89% confidence by shifting an existing Norton badge below the checkout button.
Want the whole trust category checked and priced? Run a free audit.
Proving the badge did anything
Track checkout completion rate rather than sitewide conversion, and expect a small number. Published wins sit between 1% and 9% in relative terms, which most stores cannot detect. If you cannot reach statistical power, call the change a judgement and stop describing it as a test.
Here is the part the case studies bury. Inflow's published badge results run at 93%, 89%, 81% and 80% confidence, every one under the 95% bar the industry insists on elsewhere. That is weak evidence honestly reported, which is rare. If your traffic cannot get there, the low-traffic decision method applies here too.
checkout sessions × completion gap × average order value = monthly recovery
Across the 134 sites Revslip has audited, trust is one of five categories inside roughly 200 signals, and almost never the largest leak. Worth saying plainly: audited sites belong to owners who already suspected something, so our mix is not the open web's.
Who should skip badges entirely
If you are a household name, skip them. Baymard consistently finds that users raise almost no security concerns on Apple, Microsoft or Walmart even with minimal visual security, because the brand is already doing that work. Badges pay for the sites nobody has heard of yet.
Skip them too if your checkout has a bug or redirects to a domain nobody recognises. The ranked list of checkout leaks starts well above badges.
Questions people ask about trust badges
Three come up more than the rest: whether paid badges beat free ones, whether payment logos count, and whether a badge can backfire. Each answer below is tied to a published result rather than a vendor page, and none is universal, because the variable is your audience.
Do paid trust badges beat free ones?
Not reliably. Baymard's invented seal beat paid SSL seals, and Inflow's generic graphic beat a written SSL promise. Pay for a badge when the brand on it is one your buyers know, never for the certificate behind it.
Do Visa and Mastercard logos count as trust badges?
They were the most familiar marks in CXL's survey and not the most trusted, the one place familiarity and trust came apart. Treat payment logos as information about what you accept, not as reassurance.
Can a trust badge lower conversions?
Yes. Inflow reports tests where a trustmark appeared to raise shoppers' security concerns, and badges from lesser-known brands lowering conversion outright. A badge raises the subject of risk. If it does not settle it, you have only raised it.
Not sure which finding on your site is the expensive one? The conversion leak index ranks the common ones by cost.