TL;DR. Interviews and surveys measure intention. Intention is a genuinely good predictor in the weak sense and a genuinely poor lever in the strong sense, and the published numbers let you quantify the difference: a sample-weighted correlation of r = 0.53 across 422 studies, versus experimentally induced behaviour change of d = 0.36, which converts to a correlation near 0.18. Prediction and causation differ by about a factor of three in correlation and nine in variance explained. A pre-sale replaces the stated signal with a costly one. The protocol below runs in seven days, defines in advance what counts as a pass, and includes the delivery obligation that turns an unfulfilled pre-sale into a regulatory problem rather than a learning experience.
The student’s instinct on a new product idea is to go and learn: ask people, listen, iterate. The CEO’s instinct is to ask what a decision costs if it is wrong, and to buy the cheapest reliable evidence before committing capital. Both instincts are right, and they point at the same conclusion, which is that the standard advice to “talk to twenty potential customers” is a fine way to learn what to build and a poor way to decide whether to build it.
The number that dismantles interview-based validation
Start with the good news for interviews. Intention really does predict behaviour. Sheeran’s 2002 synthesis, which meta-analysed ten previous meta-analyses covering 422 studies in total, found a sample-weighted average correlation of r = 0.53 between intentions measured at one point and behaviour measured later. In the language of the field that is a large correlation, and intention outperformed attitudes, norms, self-efficacy, risk perception and personality factors as a predictor.
Now the part that matters for your decision. A correlation of 0.53 explains 28.1% of the variance in behaviour. Which means 71.9% of what determines whether a person actually does the thing is not captured by asking them whether they will.
It gets sharper. Predicting is not the same as causing. Webb and Sheeran’s meta-analysis of experiments that actually manipulated intention found that a medium-to-large change in intentions produced only a small-to-medium change in behaviour, d = 0.36. Convert that effect size to the same currency as the correlation above and you get roughly r = 0.18.
Original analysis: what asking predicts versus what asking causes
CEOtudent editorial framework. The two published figures are reported by Sheeran and Webb, 2016. The conversion from d to r uses the standard formula r = d divided by the square root of d squared plus 4, and the variance and ratio columns are our calculation. These two quantities are not measuring the same thing, and the point of putting them together is exactly that.
| Quantity | Published figure | As a correlation | Variance explained |
|---|---|---|---|
| Intention measured, behaviour observed later (422 studies) | r = 0.53 | 0.53 | 28.1% |
| Intention experimentally changed, behaviour change produced | d = 0.36 | about 0.18 | 3.1% |
| Gap | about 3.0x | about 9.0x |
Read that bottom row carefully, because it is the entire argument. The correlational figure is the one quoted whenever someone defends customer interviews. The experimental figure is the one that describes what happens when you actually persuade somebody that they want your product. They differ by around a factor of nine in variance explained.
Sheeran and Webb identify who is responsible for the gap, and the name is unimprovable: inclined abstainers. People who intend to act and then do not. Not people who lied to you, not people who were being polite. People who meant it at the time.
Your prospective customer telling you they would definitely buy this is, on the available evidence, a sincere statement with roughly 3% of the causal force you are implicitly assigning to it.
The signal ladder
If stated intent is weak evidence, the design question becomes: what is strong evidence, and how cheaply can you get it? The organising principle is cost to the respondent. A signal is worth roughly what it costs the person giving it.
CEOtudent Validation Signal Ladder
This ranking is our judgement, built on the evidence above rather than measured by it. No study has ranked these specific signals against each other. The reasoning is stated so you can disagree with it.
| Signal | Cost to the respondent | What it actually establishes |
|---|---|---|
| “That’s a great idea” | Zero, and socially rewarding | That they are a pleasant person |
| Survey response, would-you-buy | Seconds | That the concept is comprehensible |
| Interview enthusiasm | Minutes, plus a desire to be helpful | What language they use about the problem |
| Email signup for a waitlist | An address and mild future obligation | Mild topical interest |
| A specific stated commitment with a date | Social exposure | Meaningfully more than a survey, still stated |
| Payment | Money, irreversibly | Demand |
| Payment at full intended price | Money, at the number that determines your economics | Demand at a price that works |
Everything above the bold line measures intention. Only the bottom two measure behaviour. This is why the protocol is a pre-sale and not a landing page with a signup box: a waitlist tells you a person is curious, and curiosity is exactly the state that produces inclined abstainers.
One important caveat about the top of the ladder, because we are not arguing that interviews are worthless. They are the wrong instrument for the go or no-go decision and the right instrument for almost everything else: what problem people think they have, what words they use for it, what they have already tried, what they currently pay for. Do the interviews. Just do not let them cast the deciding vote.
The seven-day protocol
The design constraint is that the whole thing has to fit in a week, cost close to nothing, and end in a decision you have committed to in advance.
CEOtudent 7-Day Pre-Sale Protocol
A framework, not a measured procedure. The evidence supports its mechanism, not its specific day count or thresholds.
Day 1: write the offer, not the product. One page. What it is, who it is for, what it costs, when it arrives. If you cannot write the delivery date, you cannot run a pre-sale, because the delivery date is the promise you are collecting money against. This is the step people skip, and skipping it is how a validation exercise turns into an obligation you have not thought through.
Day 2: set the threshold before you can see the result. Decide now what number of sales means yes, what number means no, and what you will do at each. Write it down where you can see it later. The evidence on this is indirect but real: Harkin and colleagues meta-analysed 138 interventions and found that progress monitoring worked better when progress was physically recorded or made public. A threshold you set after seeing the data is not a threshold.
Day 3: build the smallest thing that can take money. A payment link and the one page from Day 1. Not a website. The temptation to build a real landing page is the same temptation as building the real product, arriving one step earlier.
Day 4: define the failure condition in writing. What refund do buyers get if you do not ship? Say it on the page. A pre-sale where the buyer carries all the risk is not a validation instrument, it is a loan you have not disclosed.
Days 5 to 6: put it in front of people who already know you have a problem worth solving. The pre-sale goes to the audience you have, not an audience you buy. If you have no audience at all, the honest answer is that a pre-sale will not work yet, and the cold start problem is the prior question. Our piece on the minimum viable audience covers how many people you realistically need before a pre-sale can produce a readable signal.
Day 7: read the result against Day 2’s threshold and act on it. Hit the number, build it. Miss the number, refund everyone the same day and keep the lesson. The refund is not a failure condition, it is the thing that makes the whole exercise ethical and repeatable.
Make each step an if-then plan
There is direct evidence for this specific move. Gollwitzer and Sheeran’s meta-analysis of 94 studies found that turning an intention into an if-then implementation plan improved goal attainment by d = 0.65 compared with merely forming the intention. Put next to the d = 0.36 that experimentally changing intention produces, the plan is roughly 1.8 times the lever that the intent is.
So write each protocol step in the form the evidence supports. Not “I’ll launch the pre-sale this week” but “If it is Wednesday morning, then I send the offer to my list.” The irony is worth noticing: the same inclined-abstainer problem that makes your customers’ intentions unreliable makes yours unreliable too. You are not exempt from the finding you are exploiting.
The rule most pre-sale advice never mentions
Taking money for something that does not exist yet is a regulated activity, and the regulation is older and stricter than most founders expect.
Under the United States Federal Trade Commission’s Mail, Internet, or Telephone Order Merchandise Rule, when you advertise merchandise you must have a reasonable basis for stating or implying that you can ship within a given time. If you make no shipment statement at all, you must have a reasonable basis for believing you can ship within 30 days, which is why it is commonly called the 30-day Rule. If you then learn you cannot ship in the time you stated or within 30 days, you must seek the customer’s consent to the delay.
Verified: what the FTC’s 30-day Rule does and does not reach
All rows taken from the FTC’s own business guidance on the Rule.
| Point | What the FTC states |
|---|---|
| Default shipping window | 30 days, where no shipment time is stated |
| Requirement | A reasonable basis for the shipping claim, before you advertise |
| If you cannot ship in time | You must seek the customer’s consent to the delay |
| Coverage | Most goods ordered by mail, telephone, fax or internet, regardless of how advertised or paid for |
| Does not cover | Services; magazine subscriptions except the first shipment; seeds and growing plants; collect-on-delivery orders; transactions under the Negative Option Rule |
| Not a prepaid or credit sale | Merchandise shipped with an invoice payable on receipt falls outside the Rule |
| Enforcement | The FTC can sue for injunctive relief and monetary civil penalties |
Two things follow, and the second is the one that catches people.
First, a great deal of what gets pre-sold as a “digital product” is arguably a service, and the Rule explicitly does not cover services. So the 30-day clock may not apply to your cohort-based course in the way it applies to a physical good.
Second, and this is the part the exemption list can lull you into missing, the FTC’s own guidance notes that even where the Rule does not apply, being unreasonably slow to ship or failing to ship when promised can violate the FTC Act’s general prohibition against unfair or deceptive practices. Falling outside the specific rule is not the same as being unregulated.
We are describing the United States position because that is the one we could verify against the regulator’s own published guidance. Consumer protection for distance selling differs by country, and several jurisdictions add a cancellation right on top of the shipping obligation. If you are pre-selling into a market outside the United States, check your own regulator before you take the first payment rather than after.
What a pre-sale cannot tell you
Honesty about the instrument’s limits is part of using it well.
A pre-sale measures demand from the specific people who saw it, at the specific price you set, at the specific moment you asked. It does not measure market size, it does not tell you the product will retain anyone, and it does not establish that the price is optimal, only that it is not disqualifying. A pre-sale that fails may mean the idea is wrong, or that the offer was badly written, or that you asked the wrong hundred people. The instrument is much better at producing a green light you can trust than a red light you can interpret.
It also cannot rescue a decision you were always going to make anyway. If you know you will build the thing regardless of the number, do not run a pre-sale. Running one and then overriding it is worse than not running one, because it converts real evidence into a story you tell yourself about having been rigorous.
Frequently asked questions
How many sales count as validation?
There is no researched number, and anyone quoting one is inventing it. What the evidence supports is that the threshold must be set before you see the result, and that it should be tied to a decision you have actually committed to.
Is a waitlist signup good enough?
On the ladder above it sits well below payment, and the reason is the inclined abstainer finding. A waitlist collects people who meant it. Meaning it is precisely the state that fails to convert.
What if I feel dishonest selling something that does not exist?
Then say it exists in the form it actually exists in. A pre-sale that is described as a pre-sale, with a delivery date and a refund policy on the page, is a normal commercial transaction. A pre-sale disguised as an available product is not, and that is the distinction the FTC guidance is built around.
Does the 30-day Rule apply to my online course?
Possibly not, since the Rule does not cover services, but the general prohibition on unfair or deceptive practices still applies, and rules outside the United States may differ. Treat the delivery promise as binding regardless of which specific rule reaches it.
Can I run this without any audience?
Not usefully. A pre-sale with nobody to show it to produces a zero that means nothing. Build the smallest audience first.
Should I discount the pre-sale price?
A discount buys you sales at a price that is not the price your business needs, which weakens exactly the signal you came for. If you discount, treat the result as evidence about the discounted price and nothing more.
Sources
- Sheeran and Webb, The Intention-Behavior Gap, Social and Personality Psychology Compass, 2016, volume 10, issue 9, pages 503 to 518. The sample-weighted average intention to behaviour correlation of r = 0.53 from Sheeran’s 2002 meta-analysis of ten meta-analyses covering 422 studies; the Webb and Sheeran 2006 experimental finding that a medium-to-large change in intentions produced only a small-to-medium change in behaviour at d = 0.36; the identification of inclined abstainers as the group mainly responsible for the gap; the Gollwitzer and Sheeran 2006 meta-analysis of 94 studies reporting d = 0.65 for if-then implementation plans over merely forming intentions; the Harkin and colleagues 2016 meta-analysis of 138 progress-monitoring interventions and the finding that effects were larger when progress was physically recorded or made public.
- Webb and Sheeran, Does Changing Behavioral Intentions Engender Behavior Change? A Meta-Analysis of the Experimental Evidence, Psychological Bulletin, 2006, volume 132, pages 249 to 268. The experimental meta-analysis underlying the d = 0.36 figure, as reported in Sheeran and Webb, 2016.
- Gollwitzer and Sheeran, Implementation Intentions and Goal Achievement: A Meta-Analysis of Effects and Processes, 2006. The 94-study meta-analysis underlying the d = 0.65 implementation intention figure, as reported in Sheeran and Webb, 2016.
- United States Federal Trade Commission, A Business Guide to the FTC’s Mail, Internet, or Telephone Order Merchandise Rule. The 30-day default shipping window and the reasonable basis requirement; the obligation to seek the customer’s consent to a delay; the scope of coverage; the exemptions for services, magazine subscriptions after the first shipment, seeds and growing plants, collect-on-delivery orders and Negative Option Rule transactions; the treatment of invoice-on-receipt shipments; the note that unreasonably slow shipment may violate the FTC Act’s general prohibition against unfair or deceptive practices; and the available enforcement remedies.
- Orbell and Sheeran, Inclined Abstainers: A Problem for Predicting Health-Related Behaviour, British Journal of Social Psychology, 1998. The origin of the inclined abstainer decomposition of the intention to behaviour relation, as cited in Sheeran and Webb, 2016.
This content was compiled with the support of AI following in-depth research, then written and prepared for publication by the CEOtudent editorial team.
This post is also available in:
