Latest · Interpreted · Published 11 Sep 2026 · GA Applications editorial
Trust is a sequence. Most websites skip the middle.
Visitors do not decide “do I trust this company?” in one step. They pass through a sequence: recognise the situation, orient themselves, see how things work, find evidence, check who is accountable, and keep agency over what happens next. A trustworthy website serves all six stages. Trust badges and stock photography serve none of them.
Grounded in Stanford’s credibility guidelines, 6sense buyer research and HBS analysis; the six-stage sequence is GA’s model, labelled inference.
Recognition
Trust is decided before you hear from anyone
The 6sense B2B Buyer Experience Report 2025 found that 94% of buying groups had ranked their shortlist before engaging any seller, and the pre-contact favourite went on to win roughly four of five deals. By the time someone calls you, the trust decision has largely been made — on your website, without you in the room.
And it is not only the named contact deciding. The Edelman–LinkedIn 2025 B2B Thought Leadership Impact Report documents how hidden decision-makers — internal stakeholders with veto power who never meet the vendor — form their judgements from published material. Your site is being read by people you will never know about, for approval you will never see requested.
None of this is new in principle. The Stanford Guidelines for Web Credibility have said for years what still holds: make it easy to verify your claims, show there is a real organisation behind the site, make contact easy. What has changed is the cost of skipping those stages — the decision now completes without you.
The model
The six stages of the trust sequence
Each stage either keeps the visitor or loses them. The order matters: evidence shown before orientation reads as noise; accountability offered before mechanism reads as a sales trick.
Recognition
The visitor sees their own situation in your words within seconds: the problem named plainly, the place named truthfully. Not your greatness — their situation.
Orientation
They can tell what you do, for whom and where, in one or two taps. Navigation matches how customers think, not how the company is organised.
Mechanism
You show how things actually work: the process, the cost model, what happens after they call. This is the stage most sites skip — and where vagueness reads as concealment.
Evidence
Claims carry proof: real work, real photos, named sources, dated observations. Evidence is specific or it is decoration.
Accountability
A real organisation visibly stands behind the page: terms in plain language, privacy stated, a human who replies and owns the reply.
Agency
The visitor keeps control: no forced funnels, no fake urgency, no form demanding more than the conversation needs. Trust includes the freedom to leave with something useful.
Observable criteria
The trust-sequence audit
Every item is observable on the page, not a vibe. Run it on your own site and tick only what a stranger could verify in one visit.
0 of 9 complete
Six or fewer ticks and the problem is sequence, not style. Criteria adapted from the Stanford guidelines and GA fieldwork; copy or print the list and mark it up.
The new failure mode
Where the sequence now breaks: accountability in the AI era
Harvard Business School’s Working Knowledge, analysing AI-era trust crises on 8 April 2026, lands on a blunt rule: “AI made me do it” is no excuse. When an organisation’s automated output harms someone, blaming the model destroys trust faster than the original error, because it reveals that nobody was minding the mechanism.
On business websites this failure now has a recognisable shape: an AI chat widget that cannot say who is responsible for its answers; generated copy nobody at the company has read, let alone stands behind; automated follow-ups with no reachable human. Each one breaks the accountability stage, and a sequence broken at stage five makes the earlier stages worthless — the visitor now doubts the evidence they already read.
The fix is not avoiding automation. It is keeping a named human visibly answerable for every automated act, and showing the visitor exactly where the machine’s authority ends.
The badge fallacy
A trustmark image, a row of association logos, “trusted by hundreds” — these are claims, not mechanism. Stanford’s first guideline is to make verification easy; a badge that cannot be clicked through to its verifier does the opposite, inviting a check it will fail. If you show an affiliation, link it to the source. If you claim a result, show its evidence state.
Evidence honesty
Sources, method and what would change this
- Measured result
Pre-contact shortlists
94% of B2B buying groups had ranked their shortlist before engaging sellers, and the pre-contact favourite won roughly four of five deals.
6sense, “B2B Buyer Experience Report 2025”
- Verified fact
Hidden decision-makers
The Edelman–LinkedIn study reports that hidden decision-makers — stakeholders with influence who rarely meet vendors — actively consume thought leadership when forming judgements.
Edelman–LinkedIn, “2025 B2B Thought Leadership Impact Report”
- Verified fact
Web credibility guidelines
Stanford’s ten guidelines centre on easy verification, showing a real organisation, honest design and easy contact — published as research-based guidance for web credibility.
Stanford Persuasive Technology Lab, “Stanford Guidelines for Web Credibility”
- Verified fact
Accountability cannot be delegated
HBS’s analysis of AI-era trust crises holds that “AI made me do it” is no excuse; accountability stays with the people who deployed the system.
Harvard Business School Working Knowledge, 8 April 2026
- Inference
The six-stage sequence
The naming and ordering of the sequence is GA’s working model, synthesised from the sources above. It is a model for auditing, not a measured law of behaviour.
GA Applications editorial
- Proposal
What would change this conclusion
If buyer research began showing decisions formed mainly after seller contact, or if verification behaviour shifted materially, the audit criteria would be revised with the change dated here.
GA Applications editorial
Have your trust sequence audited
Send your URL. GA replies with the stage where your site loses people, the observable reason, and the smallest change that would repair it.
The audit names its criteria. You can re-run every check yourself before spending anything.