More sales from the traffic you already have
One in five abandoned checkouts dies of distrust. The published research on what actually repairs that trust is surprisingly specific, and most badge rows I see in audits are aimed at the wrong target.
There is a particular page state I have started photographing during audits, purely for my own collection: the footer with seven trust badges in a row, greyscale, twelve pixels tall, on a store whose checkout still asks for your card in an unstyled form. It is the CRO equivalent of hanging diplomas in the garage. The intent is right, the placement and the choice of signals are wrong, and the research on what actually works has been public for years. This entry sorts the signals into two piles: the ones with evidence of movement, and the decoration.
Start with the size of the wound. In Baymard’s survey data on why shoppers abandon checkouts, 19% said they did not trust the site with their credit card information. That puts distrust in the same weight class as forced account creation, and ahead of the famous complaint about checkouts being too long, which sits at 18%. Roughly one abandoned checkout in five dies of fear, not friction.
The strange and useful fact underneath: on any HTTPS page, every form field is equally encrypted. The card field is technically as safe as the newsletter field. Users do not know this and do not behave as if it were true. Baymard’s usability sessions show people judging the safety of a page by how the payment area looks and feels, checking, as one of their test subjects put it, a hundred times before purchasing. Trust, in other words, is a design property. Which means it can be designed badly.
Familiar brands near the money. Baymard surveyed 2,510 people on which badge gives the best sense of trust when paying online. Norton won with about 36%, McAfee took around 23%, TRUSTe and BBB landed near 13% each. The technically respectable SSL seals, Thawte, GeoTrust, Comodo, scraped 3 to 5%. The punchline is that several of those seals belong to the same certificate company, and the antivirus brands are not certificate authorities at all. People trust the logo they have seen on a shelf, not the cryptography they have never heard of.
A visually protected card section. The same Baymard research found that simply making the payment fields look different, a bordered box, a shaded background, a padlock icon and a short line about encryption, raises perceived security of the whole page. Even what they call a homemade padlock works. No third party seal required, no subscription. Of every finding in this entry, this is the cheapest one to act on.
Reviews, in volume and slightly imperfect. Spiegel Research Center at Northwestern published the numbers that made review widgets a category: showing five reviews raised purchase likelihood by 270% versus showing none. For higher priced products the lift reached 380%, against 190% for cheaper ones, so trust matters most where the customer risks most. And purchase likelihood peaked at ratings between 4.0 and 4.7, then declined toward 5.0. A flawless score reads as curated. A visible flaw reads as true.
A returns policy that shows up early. In the abandonment data, 15% of shoppers quit because the returns policy was not satisfactory. The signal version of this fix is not the policy document, it is surfacing the promise, plain wording near the buy button, before doubt has to go looking for it.
Now the signals I keep finding that carry no published evidence of movement, and often some evidence of harm.
The obscure badge row is the classic. Seals from vendors the shopper has never encountered score in the low single digits in Baymard’s survey, and stacking five of them does not add their percentages together. It adds visual noise to the exact screen you want calm. Baymard’s later work on perceived security even found that an invented, completely fake seal outperformed most real SSL brands short of Norton, which tells you everything about what these images are actually communicating: familiarity theater, not verification.
Self-awarded confidence is the second pile. Badges that say 100% Satisfaction Guaranteed or Best Quality, generated in a graphics editor by the store itself, ask the customer to trust the store because the store says so. I have never seen one carry a test. The same pixels spent on a concrete promise, 30 day returns, ships tomorrow, do describable work instead.
And the padlock wallpaper. Once the payment section is visually protected, repeating security icons across the header, product pages and footer stops reassuring and starts suggesting the store is nervous. Trust language has the same dosage curve as perfume.
A trust signal works when it answers the exact doubt the customer has, at the exact moment they have it. Everything else is decoration with good intentions.
Trust experiments are quieter than layout experiments, so the measurement has to listen closely. The metric that responds is not overall conversion first, it is the step where fear lives: the transition from payment page reached to order completed. When I set up a test, whether it is a protected card section, a recognizable seal by the payment fields, or reviews pushed higher on the product page, that step conversion is the primary read, with revenue per visitor as the referee, same as any experiment in this diary.
My honest ranking after years of these tests matches the published record. Reviews first, because the Spiegel numbers are the largest in the entire trust literature. Visual protection of the payment area second, because it is free and Baymard’s evidence is direct. One familiar badge third, placed where the card fields are, not where the footer is. And the seven-badge greyscale parade goes to the same place the diplomas in the garage go: it is not hurting much, but nobody ever hired the garage because of them.
They can, but the mechanism is perceived security, not real security. Baymard's research shows users judge safety by visual cues and brand familiarity, so a recognized badge near the payment fields can calm the 19% who abandon out of distrust. A row of obscure seals in the footer mostly cannot.
In Baymard's survey of 2,510 people, Norton gave the best sense of trust at roughly 36%, followed by McAfee near 23%, with TRUSTe and BBB around 13% each. Lesser known SSL brands like Thawte, GeoTrust and Comodo all landed near 3 to 5%, despite being equally legitimate technically.
Closest to the moment of doubt. Baymard found that visually reinforcing the card fields themselves, with a distinct bordered or shaded section and a security note, raises perceived security even though every field on an HTTPS page is equally encrypted. Badges buried in footers decorate; badges beside the card number reassure.
The strongest one with published data behind it. Spiegel Research Center found purchase likelihood 270% higher with five reviews versus none, rising to 380% for higher priced items, and peaking at ratings between 4.0 and 4.7 rather than a perfect 5.0. Slightly imperfect ratings read as more honest.