Commitment Ladder: Move B2B Prospects Cold to Close

A commitment ladder is a sequence of asks arranged so that each one costs the prospect slightly more than the one before it. You never jump from "has never heard of you" to "sign here." You ask for a click, then an email address, then a reply, then thirty minutes, then a signature — and each yes makes the next yes cheaper, because people work hard to stay consistent with what they have already done.

That is the whole idea. The rest of this post is the evidence it rests on, where that evidence is weaker than marketers pretend, the six rungs I actually use, and a runnable model for deciding whether adding a rung helps or hurts.

The research: Freedman and Fraser, and the caveat nobody quotes

The foundational study is Freedman and Fraser's "Compliance without pressure: the foot-in-the-door technique," published in the Journal of Personality and Social Psychology in 1966 (PubMed 5969145). In their second experiment, researchers asked Californian homeowners to put a large, deliberately ugly "Drive Carefully" billboard on their front lawn. In the control group 17% agreed. In the group that had, two weeks earlier, been asked to display a three-inch window sticker — a request almost nobody refuses — 76% agreed (summary and figures).

That is the ladder in one experiment. Same ask, same house, same billboard. The only difference was a prior, trivial yes.

Jerry Burger's review in Personality and Social Psychology Review (1999) unpicked why it works, and found it is not one mechanism but several: self-perception, commitment and consistency, conformity, and attribution all contribute, and which one dominates depends on the setup (Burger, 1999). Cialdini's consistency principle — that once we commit, we behave in ways that justify the commitment — is one strand of that, not the whole thing.

Where it does not work

Here is the part that never makes it into the LinkedIn version. Beaman, Cole, Preston, Klentz and Steblay ran a meta-analysis across 120 experimental groups in Personality and Social Psychology Bulletin in 1983 and concluded the effect, while replicable, is weak — with nearly half the studies producing no effect or an effect in the wrong direction (Beaman et al., 1983).

So: a real effect, reliably demonstrated, considerably smaller than the persuasion-book framing suggests. Treat a ladder as a way to avoid a catastrophically large ask, not as a machine that manufactures yeses. It stops you losing deals. It does not conjure them.

I have written more on which of these effects survive replication in the psychological triggers that actually make people buy and cognitive biases every marketer should understand.

The six rungs

The step-through column below is an assumption set I invented to feed the model in the next section — it is not measured data, and the only measured number in it is rung 1, which is my own Search Console click-through rate. Replace all of them with your CRM's numbers before you trust any output.

Rung 4 is the one almost everyone skips, and it is the most important. A reply is the first rung where the prospect creates something attributable to themselves — which is precisely the self-perception mechanism Burger identifies. Someone who has typed a sentence to you has cast themselves as a person who talks to you.

Rung 5 is where most of the deal is won or lost. I broke that rung out separately in the discovery call framework that converts prospects.

The arithmetic of ladder design

A ladder is a product of rates, not a sum of steps. That one fact governs every design decision, and it produces two conclusions that are not obvious.

# Commitment ladder arithmetic.
# Every rate below is an ASSUMPTION except TOP and rung 1, which are my real
# Search Console figures. Swap in your own CRM numbers before trusting output.

LADDER = [
    # (rung, step-through rate, my hours per prospect who reaches this rung)
    ("1 See",       0.0048, 0.00),   # impression -> click. Measured: 0.48% CTR
    ("2 Read",      0.1800, 0.00),   # click -> reads to the end
    ("3 Subscribe", 0.0400, 0.01),   # reader -> gives me an email address
    ("4 Reply",     0.0900, 0.10),   # subscriber -> replies to something
    ("5 Call",      0.3500, 1.00),   # replier -> books a call
    ("6 Sign",      0.3000, 2.00),   # call -> signed agreement
]

TOP = 4553  # my real impressions, 85 days to 2026-09-03


def walk(ladder, top):
    """Population surviving each rung, and cumulative hours I burn getting there."""
    n, hours, rows = float(top), 0.0, []
    for name, rate, hrs in ladder:
        n *= rate
        hours += n * hrs
        rows.append((name, n, hours))
    return rows


def end_to_end(ladder):
    r = 1.0
    for _, rate, _ in ladder:
        r *= rate
    return r


def split_pays(current, rate_a, rate_b):
    """Inserting a rung splits one ask in two. It pays only if a*b > current."""
    return rate_a * rate_b > current


def set_rate(ladder, name, rate):
    return [(n, rate if n == name else r, h) for n, r, h in ladder]


if __name__ == "__main__":
    print(f"{'rung':<12}{'reached':>10}{'cum hrs':>10}")
    for name, n, hrs in walk(LADDER, TOP):
        print(f"{name:<12}{n:>10.2f}{hrs:>10.2f}")

    print(f"\nend-to-end   : {end_to_end(LADDER):.8f}")
    print(f"closes/85days: {TOP * end_to_end(LADDER):.3f}")

    print("\ndoubling one rate at a time:")
    for name, rate, _ in LADDER:
        alt = set_rate(LADDER, name, rate * 2)
        print(f"  2x {name:<12}-> {TOP * end_to_end(alt):.3f} closes")

    for a, b in [(0.35, 0.35), (0.25, 0.25), (0.50, 0.20)]:
        print(f"split rung 4 into {a:.0%}/{b:.0%}: pays = {split_pays(0.09, a, b)}"
              f" ({a * b:.4f} vs 0.0900)")

Conclusion one: every rung is worth exactly the same

Run it and the doubling block prints the same figure six times — 0.003 closes, whichever rate you double. That is not a bug. In a multiplicative chain, doubling any factor doubles the product. There is no bottleneck rung in the mathematical sense.

Which means the ranking question is not "which rung is worst?" It is "which rung is cheapest for me to double?" Doubling my click-through rate means rewriting titles, which costs an afternoon. Doubling my close rate means becoming a better salesperson, which costs a year. Identical payoff. Take the afternoon.

Conclusion two: adding a rung usually loses money

Every marketer's instinct is to add a step — a quiz, a mini-course, a "quick 15 minutes" before the real call. But inserting a rung splits one ask with rate p into two asks with rates a and b, and the ladder only improves if a * b > p. The split_pays output shows the knife edge: splitting a 9% ask into two 35% asks pays (0.1225 > 0.09), into two 25% asks does not (0.0625 < 0.09), and a 50%/20% split barely scrapes through (0.1000 > 0.09).

Two 25% asks feel easier than one 9% ask. They are not. They are 30% worse. Extra rungs are only free in the version of the story where nobody drops out.

The same multiplicative logic drives partially-complete progress bars, which I worked through in the endowed progress effect and its maths.

What my own data says about rung 1

I can be specific here because these are my numbers, from Search Console, 85 days to 3 September 2026: 4,553 impressions, 22 clicks, a 0.48% click-through rate, roughly 390 indexed URLs.

Feed those into the model and the ladder produces 0.001 closes in 85 days. Rung 1 delivers 21.85 people — against 22 actual clicks, which is the only reason I trust the top of the chain at all. Every rung below that is a rounding error.

That is the honest read: my ladder is not badly designed, it is badly fed. And rung 1 is getting harder, not easier. Pew Research analysed 68,879 Google searches from 900 US adults and found users clicked a traditional search result on 8% of visits when an AI summary was present, against 15% when it was not — and clicked a source cited inside the AI summary on just 1% of visits (Pew Research Center, July 2025).

There is a geographic version of the same lesson in my data. Nepal: 8 clicks from 74 impressions, a 10.8% click-through rate at average position 9.2. The United States: 970 impressions, average position 38.8, zero clicks. Same ladder, same author, same quarter. Position 38 is not a weak rung 1 — it is no rung 1 at all. For the metrics that actually track this, I set them out in the five SEO metrics that matter.

FAQ

What is a commitment ladder?

A commitment ladder is a planned sequence of increasingly costly asks, where each yes makes the next one more likely because people act consistently with their prior behaviour. It comes from Freedman and Fraser's 1966 foot-in-the-door research, in which a trivial prior request raised compliance with a large one from 17% to 76%.

How many rungs should a commitment ladder have?

However many pass the a * b > p test. Adding a rung is only worth it if the two resulting step-through rates multiply to more than the single rate they replaced. In practice that caps most B2B ladders around five or six rungs.

Does the foot-in-the-door technique actually work?

Yes, but weakly. The 1983 Beaman et al. meta-analysis of 120 experimental groups found the effect replicable but far less robust than commonly assumed, with almost half of studies showing no effect or a reversed one. Design your ladder to avoid an oversized ask rather than to generate demand.

What is the first rung of a B2B commitment ladder?

Being seen at all — a search impression, a feed impression, a mention. It is the rung most people skip when planning, and the one that constrains everything below it, because the whole ladder is a product of rates.

Where is your ladder actually breaking? Run the model with your own numbers, and if the answer is rung 1 the fix is content and search, not sales scripts. See my services or get in touch.