Never Start Your User at Zero

The psychology of progress bars in app onboarding — and what an audit of 47 real flows says about who's actually using it.

Two loyalty cards for the same car wash. The one on the left needs eight stamps — eight washes, then the ninth is free. The one on the right needs ten stamps, but two are already filled in. So it also needs eight washes. Identical work, identical reward. The only difference is where the starting line is drawn.

Researchers handed out three hundred of these at a real car wash and waited. The card that started at zero: 19% of people finished it. The card that started at two: 34%. Same eight washes, nearly double the completion rate. And among the people who finished, the head-start group came back about three days faster between visits.

Now the part that should bother you. When researchers pooled 32 randomized experiments on progress bars, the eighteen that tested a normal, honest, linear progress bar averaged out to an effect on completion that was statistically indistinguishable from zero. And one specific kind of progress bar raised the odds of quitting by 56%.

Both of those things are true. The gap between them is the single most useful thing you can learn about onboarding, and it's why most of the progress bars shipping in apps right now are decoration.

This piece covers four things: the two laws of motivation that make the car wash result work; why most progress bars do nothing and some actively hurt; what 47 real iOS onboarding flows are doing right now, from our own audit; and a five-rule playbook — plus the three ways all of this backfires.

The two laws

Law one: people speed up as they get close. In 1932, Clark Hull proposed the goal gradient hypothesis: an animal moving toward a reward moves faster the closer it gets. Two years later he built the apparatus and measured it — a straight runway with electrical timing contacts every six feet. The rats ran faster in the last six feet than the first. (Nearly every blog cites 1932 for the rat experiment; the experiment is 1934.)

Seventy-odd years later, researchers got the punch-card data from a real café — 949 completed cards, roughly ten thousand coffee purchases. Same curve: customers bought coffee about 20% faster at the end of the card than at the beginning. Not because they wanted coffee more. Because they could see the finish line. Effort is not constant — it's a function of perceived distance to the goal. Users who are one step from done are the easiest users you will ever have. The problem is getting them there.

Law two: you drew the starting line. If effort scales with perceived distance, there are two ways to make someone try harder. Move them closer to the goal — which is expensive, that's real work. Or move the goal posts and tell them where they already are. That's what the ten-stamp card does. It doesn't reduce the work; it redraws the coordinate system so the user is already inside it. Nunes and Drèze called this the endowed progress effect, and their explanation is worth writing down verbatim: framing the task as one "that has been undertaken and is incomplete rather than one not yet begun" makes people more committed to completing it. A task you haven't started is a decision. A task you're in the middle of is an obligation.

One under-quoted detail from that paper: the head start works better when you give a reason for it — and in a follow-up study measuring how appealing the program felt, a throwaway reason ("here's a bonus because you came in today") worked as well as a real one. People don't audit the reason. They just need the head start to feel like it belongs to them.

The synthesis: motivation doesn't track your position. It tracks your perceived rate of closing distance. And rate is what you control, with four levers that cost no engineering work: where the starting line is, what order the steps go in, what the denominator is, and which number you show.

Why most progress bars do nothing

Survey researchers have been randomizing progress bars for twenty years, because they care obsessively about mid-survey abandonment. The 2013 meta-analysis by Villar, Callegaro and Yang sorted 32 randomized experiments by how the bar behaved. Constant, honest, linear bars (18 experiments): no significant effect on drop-off. Bars that move slowly at first and fast later (7 experiments): drop-off odds 1.56×. Bars that move fast at first and slow later (7 experiments): 0.80× — apparently helpful, but that result only reaches significance after removing an outlier. The robust finding is the negative one: what matters isn't whether you show a bar, it's which half of the flow feels fast.

The asymmetry shows up elsewhere too. Conrad and colleagues found that progress running faster than expected improves how people feel without reliably improving completion — while progress running slower than expected makes people feel worse and quit. The downside is bigger than the upside.

So: a progress bar is a speedometer, not an engine. It doesn't create motion; it broadcasts the motion your flow already has. If your flow front-loads the hard stuff — account creation, email verification, permission prompts, connect-your-bank — the bar isn't helping you. The bar is telling on you.

And the honest counterexample: Monzo published their US signup funnel. Baseline: 39 screens, about 210 taps, seventeen minutes, and 9% of downloaders finishing signup. They got it to 40% — by deleting 22 screens, not by adding a progress bar. One verification check was causing 40% of total funnel drop-off. If your onboarding is genuinely too long, psychology will not save you. Fix the flow. Then frame it.

What 47 real flows do

That's the lab. To see what's actually shipping, we audited 47 real onboarding and setup flows and screens across 40 top iOS apps (August 2026, via the Mobbin design database). Four findings stand out.

Endowed progress is now the norm, not a growth hack. Of the flows where the starting state was determinable, 87% start the user above zero. It comes in two species: a small cosmetic pre-fill on quiz bars (Duolingo, Brilliant, Evernote), and substantive pre-credit on setup checklists — 81% of checklists ship with at least one item already checked. Coinbase hands you "Account created ✓" before you touch the list, then frames the rest as "2 steps remaining." Qonto goes further: it writes your first invoice number for you, then credits you for it. The bar has been raised — starting users above zero no longer differentiates you, but starting them at zero now makes you conspicuous.

The misses are instructive. The clearest violation in the sample is Strava's getting-started screen: the headline announces "You've joined the world's largest team!" directly above a meter reading 0/4. The copy endows; the counter refuses.

The cold-start penalty is still shipping. 14% of flows reset the bar to zero at a stage boundary, and the resets cluster at exactly one place: identity verification. "Account activated!" — celebration — then a fresh, empty bar for KYC.

And nobody flips the counter. Not one app in the sample switches a single counter from done-count ("3 of 5 done") to remaining-count ("2 left") mid-flow — the clearest prescription in the research, as we'll see below. The best flows approximate it in copy (Monzo's "You're halfway there…" → "Almost there, last section now…"), but the counter itself never flips. Which means the cheapest edge in this entire literature is sitting unclaimed.

To make all of this concrete, we rebuilt the setup flow of a fictional hiking tracker, Summit, both ways — the audit's anti-patterns on one side (zero start, slow-early bar, a mid-flow reset, a badge) and the playbook on the other. Explore the interactive before/after: demo.html.

The playbook: five rules

Rule 1 — Never start at zero, and count real work. Your user didn't arrive out of nowhere. They found you, clicked, typed an email, picked a password, maybe confirmed a link. That's work — count it. Don't render "Step 1 of 4: tell us about yourself." Render "Step 2 of 5 — account created ✓, email confirmed ✓." Nothing is fabricated; you just stopped throwing away credit the user already earned. It's usually a one-constant change in your state initializer.

Rule 2 — Front-load your cheapest step. The meta-analysis says the first half of the flow is where the indicator does its damage — and while nobody has run that exact experiment on app onboarding, a study of 25,080 real web surveys puts a number on early friction: every 50 words anywhere costs about half a percentage point of completion, but 50 words in the first question costs 1.6 points — more than triple. So your hardest, most invasive step — bank connection, calendar permission, the twelve-field profile — should not be first. Put a one-tap question first. The bar jumps, the user is in motion, and the goal gradient starts working for you.

Rule 3 — Flip the framing at the halfway point. The small-area hypothesis: people are most motivated by whichever number is smaller — progress done or progress remaining. Koo and Fishbach field-tested it at a sushi restaurant with 907 customers and replicated it in the lab: low-progress people respond to what they've done; high-progress people respond to what's left. Notice what this implies: a standard filled bar only ever shows accumulated progress, which makes it the wrong display for the back half of your own flow. "Two steps left" beats "80% complete" for the same state. The implementation is a ternary — show done of total before the midpoint, remaining left after. Per the audit: zero of 47 flows do this. It's yours.

Rule 4 — Count fields, not screens. Baymard Institute — the checkout-usability research group, working from qualitative testing rather than A/B experiments — finds the number of form fields hurts usability far more than the number of steps. So splitting a long form into more, lighter screens redistributes friction rather than adding it, and buys you more progress events to show. The Obama 2012 campaign split one long donation form into four sequential steps, ordered by which fields people were failing on, for an over-5% conversion lift. (A practitioner report, not a controlled publication — and note the test was about step-splitting, not progress bars.)

Rule 5 — The endowment has to be real. The test is clean: would the design survive disclosure? If a tooltip saying "you start at 20% because we're counting the account you already created" leaves the effect intact, that's framing — legitimate, nothing false asserted. If the tooltip would have to say "you start at 20% because we found it makes you finish," that's deception. The car wash card passes easily: two of ten stamps, genuinely eight washes to go. Reframe freely. Fabricate never.

The three backfires

The post-reward trough. In the café dataset, customers who finished one card and started a second slowed from a 2.1-day purchase gap at the end of card one back to 3.1 days at the start of card two. Motivation didn't decay — it reset to cold start. Every "Setup complete! Now let's set up your team…" screen with a fresh 0% bar charges that penalty at every stage boundary. Don't stack sequential zero-to-a-hundred bars: run one continuous gradient, or endow each new stage so it never renders at zero. (Recall: 14% of audited flows still do exactly this, mostly at identity verification.)

A goal they miss is worse than no goal. The largest field experiment in the area: a hotel chain gave 95,532 loyalty customers a progress goal. Eighty percent missed it. Those who hit it bought more afterward; those who missed it bought less than matched customers who were never given a goal at all — and the damage landed hardest on the highest-tier, most loyal customers. The promotion was still a net win overall, but it won by taking from the most invested users. If you show a bar, the target must be genuinely reachable and must genuinely close.

Getting caught. The "labor illusion" research found people preferred a transparent sixty-second wait over instant results — but the authors warn that suspicion of manipulation erodes the effect, and it fully reverses when the visible effort ends in a bad result. Show the effort, deliver the goods. And on completion rewards: a meta-analysis of 128 studies found completion-contingent rewards ("finish setup, get a badge") reduce later voluntary engagement (d = −0.36), while informational feedback ("your workspace is set up correctly") increases it (d = +0.33). Don't dangle the badge. Confirm the competence.

Close

The bar isn't the motivator. The perceived rate is. Effort scales with perceived closeness to the goal, and while you don't own how much work there is, you do own where the starting line sits, what order the steps come in, what the denominator is, and which number you put on screen. Four levers, zero units of extra work, none of them requiring a lie — and the audit says one of them (the framing flip) is currently implemented by nobody.

Homework, ten minutes: open your own product, sign up as a brand-new user, and start a timer. What does a person see in the first fifteen seconds — and is that the fastest-moving part of your entire flow? If not, you don't have a motivation problem. You have an ordering problem. And now you know exactly how to fix it.