A brand pays a creator $40,000 for three posts. The posts perform. Views land, comments run warm, the audience shows up exactly as promised. Six weeks later the brand runs a recall study and finds that most of the people who watched cannot name the brand that paid.
Nothing malfunctioned. The targeting was correct, the creator was the right fit, the content was good. What failed was an assumption buried so deep in influencer strategy that almost nobody states it out loud — the assumption that affection can be pointed at a third party.
Cialdini's liking principle is straightforward: we say yes more readily to people we like. Similarity, familiarity, praise, and cooperation toward shared goals all raise compliance. The research is durable and the mechanism is well understood.
The misreading happens at the next step. Marketers took people comply with those they like and quietly converted it into liking can be purchased and aimed elsewhere. Those are not the same claim. The first describes a relationship. The second assumes that relationship is transitive — that if an audience likes a creator, and the creator names a brand, some fraction of that regard will travel down the chain and land on the brand.
Liking is not transitive. It is the only one of the seven principles a brand cannot manufacture on its own behalf. Authority is built through credentials and track record. Scarcity is engineered through genuine constraint. Commitment can be invited. Social proof can be earned and displayed. Liking exists only inside a relationship, and the relationship in question belongs to somebody else.
What a brand buys in a creator deal is not affection. It is adjacency to affection. The two are priced identically and perform nothing alike.
Research from UK firm MHP Mischief found that more than half of consumers surveyed could not recall a brand's name after seeing it in an influencer post. Only about a quarter had ever purchased a product an influencer recommended. Forty-one percent clicked a branded link. A third saved a collaboration post to come back to.
Read that carefully, because the failure is specific. The audience showed up. They watched. They liked the creator exactly as much afterward as before. The relationship survived intact. The brand simply never entered it.
The academic literature explains why. Parasocial research consistently finds that expertise and trustworthiness predict parasocial attachment, while attractiveness does not. Audiences bond with creators over demonstrated judgment — this person knows things, this person has been right before. That bond is built on an accumulated record. A record is not a fungible asset. It is a reputation, and reputations do not split.
A recent framework on parasocial trust transfer makes the failure mode explicit: influencer-brand partnerships succeed or fail according to whether they sustain or disrupt the conditions for connection, with the sharpest friction appearing when authenticity collapses. The brand is never neutral cargo. It is either consistent with the creator's demonstrated judgment or it is evidence against it.
And here is the number that settles the argument. Creator-led branded content built around narrative achieved 74 percent unaided brand recall, against 43 percent for brand-produced content in traditional ad formats. The same creators. The same audiences. The variable is not who is talking. It is whether the brand is inside the story or beside it.
Finance departments noticed before marketing departments admitted it. Performance-based compensation has moved from 23 percent of brand partnerships two years ago to 53 percent in 2026. Flat fees for reach are becoming the exception rather than the default. Target dropped standard creator commissions in favor of performance tiers this April.
Confidence in the model shifted alongside it. In 2024, 84.4 percent of marketers said they planned to run more long-term creator partnerships. Asked the same question about 2026, 27.9 percent fewer agreed. Budgets spread across more creators on shorter commitments.
This looks like discipline. It is closer to a hedge. Performance pay does not repair the transfer failure — it relocates the cost of that failure onto the creator. The brand still receives adjacency. It has merely stopped paying full price for adjacency that produces nothing.
The hedge also makes the underlying problem harder to fix. Shorter commitments and wider spread are precisely the conditions under which liking is least likely to carry. A creator on a one-off performance deal has no reason to weave a brand into their own narrative and every reason to run the cleanest available conversion mechanic and move on. The brand optimizes for measurable response and structurally guarantees that nothing durable gets built. The dashboard improves. The asset does not.
Transfer is not impossible. It is conditional, and the conditions are narrower than the market has been willing to accept.
The 74-versus-43 recall gap is the whole argument. Recall follows narrative position, not exposure count. If the brand can be lifted out of the content without changing what the content is about, the audience will lift it out too — not deliberately, simply as a function of how memory encodes. Sponsorship slots are structurally forgettable because they are structurally detachable.
The difference is easy to see in practice. A creator who spends four minutes explaining why they abandoned a routine they had publicly recommended for two years, and what replaced it, has made the replacement inseparable from the story. Remove the product and there is no video. A creator who walks through the same routine and includes a sponsored product in position three has made it a removable component. Remove it and nothing changes. Same creator, same audience, same product, same budget. Only one of them is still in anyone's memory six weeks later.
The practical test is unglamorous. Cut the brand out of the draft. If the piece still works, the brand was never in it.
Parasocial attachment tracks demonstrated judgment. A creator who endorses everything has demonstrated nothing, and audiences price that in faster than brands expect. The endorsement that moves people is the one carrying visible risk to the person making it — a category they have publicly avoided, a competitor they once praised, a claim they can be held to later.
This is why aggressive creator diversification is self-defeating. The more creators a brand appears beside, the less any single appearance signals. Reach and signal move in opposite directions past a certain volume, and most programs are well past it.
A single post asks an audience to extend trust to a stranger on one person's word. Repeated association across quarters asks them to observe a relationship instead. The second is a far weaker request, and it is the only one that compounds. Audiences do not transfer liking on the strength of an introduction. They transfer it after watching a partnership hold up under conditions where it could have broken.
The market is moving away from exactly this structure. That is an opportunity, not a trend to follow.
Most creator strategy is a workaround for a problem nobody wants to name. If a brand were interesting enough to be worth mentioning, it would be mentioned. The budget exists to purchase mentions that would not otherwise occur — which makes the budget a fairly precise measurement of the gap between what a brand is and what it needs someone else to say about it.
That gap does not close by widening the roster or tightening the contract. It closes by building the thing a creator would have talked about anyway.
Worth saying plainly: creator budgets have grown into one of the largest line items in modern marketing partly because they let a brand postpone the harder question indefinitely. Spending is easier than becoming interesting. On the recall evidence it is also considerably less effective, and the gap between the two widens every quarter the harder question goes unasked.
The creators worth partnering with understand this better than the brands courting them. It is why the good ones are difficult, why they decline more than they accept, and why their rates look unreasonable relative to audience size. They are not selling reach. They are protecting the only asset they have, which is the reason anyone listens to them at all. A brand that treats that asset as inventory will get inventory pricing and inventory results.
Stop measuring creator programs on reach, engagement, or even attributed conversion. Measure them on unaided brand recall among exposed audiences, four to six weeks out. If the number is low, the problem is not the creator and not the targeting. The brand was adjacent to a story instead of inside one.
Then decide which side of the gap to fix. Both are legitimate. Rewriting the brief so the brand becomes load-bearing is faster and available immediately. Building something a creator would mention unpaid is slower and permanent. The first buys a quarter. The second buys the category.
What does not work is the middle — paying for proximity, calling it affection, and reporting the impressions.
Liking is the one principle you cannot build for yourself. You can only become worth extending it to.
Be subtle, but seen.