The persuasion principles that actually hold up in marketing come down to three forces: commitment, reciprocity and conviction. Fired separately, each moves the needle a little. Landed on the same person, in the same session, in that order, they carry each other. A video funnel is the machine that makes them converge.
What the experiments actually found
Ithaca, New York, fall of 1992. Ninety-two dinner tables, one variable: a small piece of chocolate placed on the check, or nothing at all. The average tip moved from 15.06% to 17.84%. One chocolate.
That study, published by David Strohmetz, Bruce Rind, Reed Fisher and Michael Lynn in the Journal of Applied Social Psychology in 2002, gets quoted in every persuasion article ever written. What almost nobody says is that it only covers the middle chapter. Reciprocity does not fire on its own. Something has to happen before it, and something else has to happen immediately after, or the chocolate stays a chocolate.
Commitment: the act binds, not the promise
Palo Alto, 1966. A researcher calls 156 housewives and asks whether five or six men can spend two hours in their home cataloguing everything in their cupboards. An absurd request. Roughly one in five says yes anyway. In one group, though, more than half agree, and the only difference is a phone call three days earlier in which they answered eight questions about their kitchen soap.
Jonathan Freedman and Scott Fraser ran that study at Stanford. The numbers: 52.8% compliance against 22.2%. This is foot-in-the-door, the direct ancestor of everything marketing now calls a micro-commitment.
The decisive detail hides in a condition almost nobody quotes. A third group agreed to answer the soap questions, but the researcher never asked them. Those women topped out at 33.3%. Saying yes buys eleven points. Going through with it buys thirty. Nearly two thirds of the effect comes from the act, not the agreement.
A second experiment in the same paper pushed harder. Residents were asked to plant a large, badly lettered DRIVE CAREFULLY sign on their front lawn, big enough to hide the front door. Asked cold, 16.7% agreed. Among those who had accepted a three-inch safe-driving sticker for their window two weeks earlier, 76% did.
Two decades later, two French social psychologists spent a whole book on the conditions under which this holds. Robert-Vincent Joule and Jean-Léon Beauvois published Petit traité de manipulation à l'usage des honnêtes gens in 1987, still in print in a 2024 fourth edition. Their thesis fits in one line: an act only binds when it is visible, costly, hard to walk back, and freely chosen. A click checks none of those boxes.
Reciprocity: measured by the gesture, not the gift
Back to Ithaca, because two lessons come out of it and the second one matters more. The market value of the gift does almost nothing: a few cents of chocolate moved close to three points of tip. What creates the debt is that the gesture was aimed at that table. Marketing learned the opposite lesson. It emails the same forty-page guide to forty thousand addresses, then wonders why nobody feels they owe anything.
Conviction: nobody is persuaded by what they sit through
1943. Meat is rationed in the United States, the best cuts go to the army, and the government wants American households cooking organ meats. Kurt Lewin tested two methods on comparable groups. A nutritionist's lecture, recipes handed out, nutritional and patriotic arguments: 3% served organ meats the following week. A group discussion where the women worked out for themselves how to convince women like them: 32%.
Ten times more, from the same information in the same amount of time. What Lewin isolated is the gap between receiving an argument and producing one. The listener evaluates. The speaker commits. Eighty years on, most online selling still runs on the lecture model.
Marketing fires the three forces separately, and loses all three
Open the acquisition plan of any serious brand and all three forces are in there. Scattered.
Reciprocity is the lead magnet: a guide downloaded Monday, forgotten Tuesday. Commitment is the three-step form or the questionnaire at the top of the page. Conviction is the webinar, the sales video, the demo. Three assets, often three teams, and above all three moments separated by days or weeks. None of the experiments above survives that split.
| Force | What the experiment measured | What marketing does with it | Why it breaks |
|---|---|---|---|
| Commitment | 52.8% against 22.2% after an act actually carried out (Freedman and Fraser, 1966) | a click, an email address typed into a field | the act costs nothing and is invisible, so it binds nothing |
| Reciprocity | average tip from 15.06% to 17.84% with one chocolate (Strohmetz et al., 2002) | the same guide emailed to the whole list | nothing was made for this specific person |
| Conviction | 3% after a lecture, 32% after a discussion (Lewin, 1943) | a product page, a sales video rarely watched to the end | the prospect sits through the argument instead of making it |
The table tells the same story on every row: the marketing version keeps the shape of the experiment and drops the mechanism.
The 2026 attention numbers finish the job. Contentsquare's benchmark, built on 99 billion sessions across 6,500 websites, puts the average visit at 2 minutes 20 on mobile and 4 minutes 46 on desktop, with engagement down 10% year on year. One visit in three now starts on a product detail page, and 61% of those visits leave without ever seeing a second page. That is the real budget a static page has to convince someone, on the channel carrying 70% of traffic: two minutes twenty.
Webinars, the supposed heavy artillery, do no better. Livestorm's 2026 report, built on 33,786 sessions and more than seven million registrations across 2025, measures a 47.7% show-up rate: out of 175 registrants, 83 join and 92 never do. The ones who join watch 26 minutes out of 68. Active participation sits at 15.4%. So the most expensive format in the entire marketing stack delivers an audience cut in half before it even starts, paying attention a third of the time, and silent in more than eight cases out of ten.
The rule nobody writes down: same person, same session, same order
Articles about persuasion principles all look alike. Six principles, one example each, a checklist at the bottom. None of them says the thing that matters: these forces do not add up, they condition each other.
Order is not decorative. Reciprocity creates a debt only if the prospect gave something of themselves first; without their answers, whatever you send is just more content. Conviction holds only if they were given a reason to stay; without the debt, they close the tab. And the first act counts only if something comes back fast enough to prove it was worth doing.
The three conditions that follow are logistical before they are psychological. One person. One session. No gap in between. That is exactly what a classic funnel cannot do, because it was built to distribute content to segments.
Put numbers on it. A thousand paid mobile visitors, at the median conversion rate Contentsquare reports for 2026, produce twenty sales. BodyTime, a fitness programme brand running a fourteen-question video funnel, multiplied its conversion rate by 2.5 at equal traffic. On the same thousand visitors, that would be fifty. The number belongs to BodyTime and describes no average, but it gives the order of magnitude of what convergence changes.
What a video funnel does with all three at once
A video funnel runs in two beats. The visitor answers a quiz in which the seller asks every question on camera. They then receive a personalized analysis video, assembled automatically from their answers out of segments filmed in advance by a real person. Full mechanics here: What is a video funnel? Definition, mechanics and 2026 data. What matters for this argument is what the sequence does to the three forces.
The quiz meets the Joule and Beauvois conditions one by one. Five to fifteen questions about your own situation, your blockers, your budget: the act costs time, it is explicit, it is about the person answering, and nobody forced it. The prospect has just stated something about themselves.
The video that follows fires reciprocity for the exact reason the chocolate did in 1992: it was made for this person. Ten to fifteen minutes assembled from their answers, delivered within a couple of hours. Marginal production cost stays low; perceived value has nothing in common with a PDF. Same gap as in Ithaca, moved online.
Conviction happens while the video plays. The prospect watches ten minutes of argument end to end, voluntarily, because they asked for the analysis themselves by answering. Lewin called that the difference between a lecture and a discussion. A video funnel manufactures the second one, at scale.
On the production side, the whole investment fits in a morning. Two to four hours of filming on a phone to capture the segments, around thirty minutes of setup, and a five-question funnel with three options each then yields hundreds of combinations from roughly twenty clips. Filming is the real cost of entry, and it is precisely what your competitors will not do.
The results we can quote belong to our clients. BodyTime holds thirteen minutes of attention before its offer even appears on screen. Anna Velazia, a jewellery and crystal brand, moved 70% of its ad budget onto its video funnel, now its first acquisition channel, at three times the ROAS of its other campaigns. Neither number is a promise: both are attributed measurements, on offers where advice weighs on the decision.
What this data does not prove
A word of honesty about the sources, since it is missing everywhere on this topic. The three experiments quoted here date from 1943, 1966 and 2002. They were run on the phone, on doorsteps and in a restaurant dining room, never on a product page. As far as we know, nobody has replicated foot-in-the-door at the scale of an e-commerce funnel with a publishable protocol.
What can be measured, though, really is measured. Attention time. Webinar show-up rate. Conversion at equal traffic, before and after. Psychological mechanisms explain the gaps that show up; they do not guarantee them. When an article hands you a precise percentage because it activated reciprocity, it made the number up, and no source will back it.
Three questions to ask your own funnel
Take your current funnel and ask three questions, in this order. Does the visitor do something that costs them, before anyone sells them anything? Do they then receive something that would not exist without their answers? And does that thing last long enough for them to convince themselves?
One no is enough to explain the result, and it points at neither your traffic nor your offer. Your setup fires the three forces separately, and none of them holds long enough to carry the next. The reverse build exists, described end to end in the mechanics of a video funnel.
Three yeses cost one morning of filming, once. Three noes cost you, on every single visitor, the gap between two minutes twenty of attention and thirteen. What you sell this year depends less on how much traffic you buy than on what happens between the first question asked and the last minute of video watched.
Frequently asked questions
What are the six principles of persuasion?
Reciprocity, commitment and consistency, social proof, authority, liking and scarcity. Robert Cialdini laid them out in his 1984 book Influence: The Psychology of Persuasion, and added a seventh, unity, in Pre-Suasion in 2016. Inside a sales funnel, the first three carry most of the measurable effect, and they only work in sequence.
Does the foot-in-the-door technique still work online?
Yes, but only when the first act costs something. Freedman and Fraser (1966) measured 33.3% compliance among people who had merely agreed, against 52.8% among those who completed the small task. A click, or an email address dropped into a field, does not belong in the second category. Seven considered answers about your own situation do.
Persuasion or manipulation: where is the line?
The line runs through whether what you show is true. A quiz that asks real questions and produces a genuinely different analysis depending on the answers informs the prospect as much as it convinces them. A quiz whose branches all lead to the same pitch burns trust inside two minutes, and people notice.
Which force should you build first?
Commitment, every time. It is the only one that conditions the other two: without a first act from the prospect, reciprocity has nothing to lean on and conviction has no time to work with. When traffic is scarce, the job is not to attract more people but to do far more with each visitor.
