How Wikipedia builds trust

Hey, I’m Isaac 👋 I founded Pistachio, a B2B content studio helping brands like Atono and Clay to build trust, relationships and loyalty with their current and future customers.

If you want to chat about your own strategy, book a free call with me here.

Read time: 4 mins 22 secs

Interested in sponsoring these emails? See our partnership options here.

Anyone on earth can edit any page on Wikipedia right now. It is also one of the most trusted sources on the internet, both for people and for AI citations.

Those two facts seem completely at odds. Wikipedia co-founder Jimmy Wales spent last year working out exactly how and why it works in his book, The Seven Rules of Trust.

Rachel Botsman, Trust Fellow at the University of Oxford’s Business School, defines trust as a confident relationship with the unknown. She says if you already knew how something would turn out, you wouldn’t need trust at all. Trust only exists where risk does. It is the thing that lets someone act before they have proof.

Wales arrives at the same conclusion, saying "trust is not a statistical thing, it's something in the individual mind".

Putting those together and we get a definition of trust as a risk someone else decides to take on you. You can be trustworthy on your own. Whether you are trusted is decided outside of your control, inside other people’s heads. Which makes "build trust" a pretty strange directive.

Charles Green and David Maister offer an equation for trust in their book, The Trusted Advisor. It puts self-orientation in the denominator, so the more visibly you are in it for yourself, the smaller the trust. No matter how credible and reliable you seem on the top.

These definitions help to understand the complexities in borrowing or acquiring trust. When OpenAI paid nine figures for TBPN, a YouTube channel with less than 70,000 subscribers, they bought distribution and the credibility of the two hosts. But they couldn’t buy the verdict of each individual audience member. That part does not transfer. You earn your way towards trust for years, but someone else holds the decision on whether you’ve arrived.

Wales offers seven rules for getting there:

I see them falling into three buckets, each one answering a question someone subconsciously asks before deciding to take the risk of trusting you. Do you see me? Can I check? And will you hold when it costs you?

Make it personal, be positive about people, and be trusting. These lean on Frances Frei's trust triangle from Harvard, her map of the three things a person needs before they will trust you.

The brand version of this looks like a grieving customer emailing Chewy to return an unopened bag of food after her dog died, and Chewy refunding her, telling her to donate the food to a shelter, and sending flowers signed by the rep who took the call. That’s self-orientation at zero, in the trust equation's terms.

Honestly, I think this whole group is table stakes. Missing any of it ends the conversation, and clearing all of it earns you nothing more than consideration. Showing your customers you understand them is marketing 101.

Create a clear purpose, and be transparent. A purpose is the promise, and without one there’s nothing to hold you to and no test you could ever fail. Transparency is then the receipt. It lets people see you keep the promise without having to take your word for it.

Wikipedia does both at once. Their promise of free access to the worlds information is backed up by their non-profit status. Then every page edit and argument between editors is publicly available, permanently, for anyone to inspect for themselves. The openness works precisely because of the scrutiny. You trust the article because you can watch it being fought over.

Even imperfection reads as proof. Northwestern research on the impact of customer reviews found that purchase likelihood peaks somewhere between 4.0 and 4.7 stars and then falls as ratings approach a flawless 5.0. A visible flaw that’s verifiable beats a perfect score that’s not. Buffer understood this power of transparency years ago when they started publishing every employee's salary.

Be civil, and be independent. The other five rules get you considered. These two get you believed, because these are the only two that can really cost you.

Start with independence. In early 2002, the Spanish Wikipedia community walked out over a rumour that Wales might put ads on the site. By that August he had committed, permanently, to running none. Some estimates say that decision costs Wikipedia around $2.3 Billion a year in forgone revenue. Twenty-four years of paying that opportunity cost is the reason "anyone can edit it" never turned into "nobody can trust it".

Economists call this “costly signalling”. A signal only carries value when faking it would cost a liar more than the lie could return. So every cheap claim a brand makes is evidence of nothing. The only thing a mission statement on your website proves is that you can type. I’ve talked before about Anthropic consistently paying the bill for their beliefs and how that has established their position against competitors like OpenAI in the wary world of AI.

The independence rule has a trap, where independent is interpreted as a synonym for neutral. Wales's own qualifier is that you stay out of public fights unless they are directly relevant to what you do. Wise is the perfect example of clearing that bar. Their whole business is built on the promise of fairer foreign exchange fees, so screaming from the rooftops that banks were stealing from customers through exchange rates was just naming what Wise existed to fix.

But calling someone else out is itself a promise about your own conduct. It turns the scrutiny straight back onto you. Wise could afford that because it stood on the customer's side, and have a product that follows through on their promise. Pick the wrong fight and the maths flips quickly. Edelman's 2026 Trust Barometer found 7/10 people are now unwilling or hesitant to trust someone who holds different values from their own. Faced with a divisive issue, 35% said a business earns their trust by helping people cooperate without taking a side, but 28% said they wanted it to back the position true to its values. A brand that stands for nothing has offered nothing, but a brand that picks the wrong fight has also lost.

Build is the wrong verb for trust. You make a promise specific enough to be tested, then pay the bill when it lands.

Most brands never get there. They never get trusted because they are never truly tested, and they are never tested because they never made a promise that could cost them anything to keep. They stayed safe, stayed liked, and stayed forgettable.

Nine figures bought OpenAI the TBPN microphone. Whether anyone believes what comes out of it is still, as it has always been, a verdict reached by each individual viewer. Wikipedia have spent twenty-four years earning theirs the only way it can be earned.

The bill always arrives. Trust is the reward once you’ve paid it.

If you enjoyed this post or know someone who may find it useful, please share it with them and encourage them to subscribe: brandchemistry.co/p/how-wikipedia-builds-trust