The progress bar hits 100% at step 12 — users quit at step 4

You've built a twelve-step onboarding flow. Analytics says the drop-off cliff is at step four — right after you ask for the ABN. The completion bar is beautiful, the copy is clean, and none of it matters, because most people never see step five. So what exactly is happening in a user's head between step three and step four, and can you design your way out of it?

The bar is a promise, and step four breaks it

Progress bars are one of the most studied widgets in interface design, and the research is more ambivalent than most developers assume. In 2010, researchers at the University of California ran a series of experiments on what they called the "goal-gradient effect" — the finding, first observed in rats running mazes, that motivation intensifies as a goal gets closer. Participants who were given a ten-stamp coffee card with two stamps already filled in completed it faster than participants given an eight-stamp card with none filled in. Same effort required, different perceived distance. The bar itself does work.

But there's a catch, and it's the catch that's quietly killing your funnel. The goal-gradient effect depends on the user believing the remaining steps are roughly equivalent to the ones they've already cleared. Step four — the ABN, the business details, the "connect your accounting software" screen — is not equivalent. It's a wall. The user has been coasting through name and email, and suddenly the flow is asking for information that requires them to leave the tab, find a document, or make a decision they haven't made yet.

This is where behavioural economics earns its keep. Kahneman and Tversky's work on loss aversion showed that losses loom roughly twice as large as equivalent gains. In a signup flow, the "loss" isn't money — it's the sunk effort of steps one through three, now at risk of being wasted. The user isn't thinking "I'll gain a dashboard." They're thinking "I've already spent four minutes and now it wants my ABN, and if I get this wrong I've wasted the four minutes." The bar that was supposed to motivate has become a receipt for time already spent.

Variable rewards explain why the exit is so abrupt

B.F. Skinner's variable-ratio reinforcement schedule — the finding that unpredictable rewards produce the most persistent behaviour — gets cited constantly in product circles, usually to justify notification badges. But it's more useful as a diagnostic tool. Ask yourself what your onboarding flow rewards, and how predictably.

Steps one through three usually deliver something. You type your name, the field validates, a tick appears. Small, immediate, certain. Step four delivers nothing until it's complete. No tick, no preview, no partial credit. You've switched from a steady drip of confirmation to a single large ask with a delayed payoff, and the behavioural literature is unambiguous about which one people abandon.

The practical fix isn't to gamify the ABN field. It's to break the large ask into smaller confirmations. ABN lookup tools return a business name and status in under a second — show that result. Let the user see the system has recognised them before asking for anything further. You've converted one uncertain, high-effort step into three small certain ones, and you've kept the reinforcement schedule ticking over.

The specific study worth knowing about

Jared Spool's team at User Interface Engineering ran a well-known study on form abandonment that's still the clearest illustration of this. They tested a multi-page checkout against a single-page version and found the multi-page flow performed worse — not because of the number of fields, but because users couldn't see how much was left. When the team added a clear "you are here" indicator showing total steps and current position, completion improved. The lesson wasn't "fewer steps." It was "make the remaining effort legible."

That distinction matters enormously for Australian business software, where ABN verification, GST registration status, and ATO-linked details are often mandatory. You can't remove step four. But you can make it legible, and you can make it feel like a small step rather than a cliff.

Concretely, that means three things. First, show the total step count honestly, including the annoying ones. Users who discover a hidden step feel deceived, and deception is disproportionately punished — back to loss aversion. Second, move any step requiring external documents or lookups to a position where the user has already invested enough that abandoning feels genuinely costly, but not so early that they haven't seen any value. Third, and most importantly, give every step a visible output. Even "we've saved your details" is a reward.

Where this is heading

The interesting shift over the next few years isn't better progress bars — it's the collapse of the multi-step form altogether. Australian businesses can now verify an ABN, confirm GST status, and pull company details through APIs that return in milliseconds. Every step that exists purely to collect a verifiable fact is a step that shouldn't exist. The steps that remain will be the ones requiring genuine human judgement: choosing a plan, describing a business, deciding what to connect.

That reframes the design problem. If you can't eliminate a step, you should be asking whether it's a decision or a data entry. Decisions deserve context, examples, and a way to change your mind later. Data entry deserves to be pre-filled or deleted.

The teams that get this right will stop optimising the bar and start questioning the steps behind it. Pull your funnel data and look at where the cliff is. Then ask, for that specific step, what the user has to gain, what they stand to lose, and whether the interface is telling them the truth about both.