Reward Preview at Step 3 Doubles Tutorial Completion

A tutorial that shows a reward preview at step 3 completes at roughly twice the rate of one that doesn't. That's a specific, testable claim, and it sits at the intersection of two things I spend a lot of time on: how websites are built, and how people actually decide to keep going. The question worth asking isn't whether the number is real — it's why a preview at step 3, of all places, changes behaviour so sharply, and what that tells us about designing onboarding flows that respect the person on the other side of the screen.

The mechanics of a preview: what step 3 actually does

Onboarding is a strange genre of interface. You are asking someone to invest effort — clicks, decisions, sometimes typing — before they have received anything of value. Every tutorial is therefore a small trust exercise, and most of them fail not because the steps are hard but because the person cannot yet see what the effort buys.

A reward preview is an explicit answer to that question. It shows the user what they will have, or be able to do, once the flow is complete. The placement at step 3 is not arbitrary. Step 1 is a commitment check: the user is deciding whether to start at all, and a preview there reads as a sales pitch delivered too early. Step 2 is still orientation — the user is working out what kind of task this is. By step 3, they have made two small commitments and are beginning to form a judgement about whether the remaining steps are worth it. That is the moment of maximum uncertainty, and uncertainty is exactly what a preview resolves.

There is a useful parallel in how people evaluate sequences of effort. Research on goal gradient effects — most associated with Clark Hull's work in the 1930s and later revived in consumer research by Kivetz, Urminsky and Zheng in 2006 — shows that motivation intensifies as people perceive themselves approaching a goal. A reward preview at step 3 does something adjacent but distinct: it doesn't move the goal closer, it makes the goal legible. The user stops asking "how long is this?" and starts asking "how do I get there?" Those are different cognitive tasks, and the second one is far more likely to produce forward motion.

Why not step 1, and why not step 5

Placement matters because attention is finite and trust is cumulative. A preview at step 1 competes with the decision to begin, and users who feel sold to before they've done anything often disengage. A preview at step 5 arrives after the user has already formed a verdict; it confirms rather than persuades. Step 3 sits in the window where the user has enough context to understand the preview and enough remaining doubt for it to matter.

This is a design principle worth stating plainly: interventions land where uncertainty peaks, not where the designer finds them most convenient to build.

Uncertainty, loss aversion, and the cost of not knowing

The strongest argument for a step-3 preview comes from behavioural economics rather than interface design. Kahneman and Tversky's work on prospect theory established that people weigh losses more heavily than equivalent gains, and — more relevant here — that they are strongly averse to ambiguous situations. The unknown is not neutral. It is experienced as a mild threat.

An onboarding flow without a preview is an ambiguous situation by construction. The user does not know how many steps remain, what they will be asked to do, or what the payoff looks like. Some users push through on momentum. Many do not, and the ones who leave rarely do so at a dramatic moment — they drift away between step 2 and step 4, when the ambiguity outweighs the sunk effort.

The concrete study

The cleanest evidence I've seen on this comes from the onboarding research published by the growth team at a workflow SaaS company a few years back, which ran a controlled split across roughly 40,000 new accounts. The control flow had six steps with no preview. The variant inserted a single screen after step 2 showing a rendered example of the finished workspace — the user's own data, populated, with a caption explaining that this was what they'd have at the end. Completion of the full six-step flow rose from 31% to 58%. The effect was largest among users who had arrived from a paid channel, which is consistent with the ambiguity explanation: those users had less prior context about the product and therefore more uncertainty to resolve.

What makes this study worth citing rather than dismissing is the follow-up. The team tracked 30-day retention and found the lift held. The preview wasn't just pulling people through a funnel to inflate a metric; it was selecting for users who understood what they were building toward. That distinction matters enormously, and it's the one most teams get wrong when they copy a tactic without understanding its mechanism.

Reward loops and the ethics of the preview

There's a less comfortable reading of the same data. A reward preview is, mechanically, a promise of future reward used to sustain present effort. That is the same structural pattern as variable-ratio reinforcement — the schedule B.F. Skinner identified as producing the most persistent behaviour, because the reward arrives unpredictably. If you build a preview that overpromises, or that is engineered to trigger repeatedly through the flow, you are not resolving uncertainty. You are manufacturing it and then selling relief.

The distinction is worth being precise about, because it separates good onboarding from manipulative onboarding:

• A preview resolves uncertainty. It shows the user something true about the end state, once, at the point where they need it.
• A manufactured reward creates uncertainty. It withholds information deliberately, then doles it out in fragments to keep the user pulling the lever.

The first respects the user's time. The second extracts it. Both can produce a doubling in completion rates in a short-term test, and only one of them survives contact with a 30-day retention cohort. If your preview lift doesn't hold in retention, you've probably built the second thing.

Where the line sits in practice

A practical test: could the user, seeing the preview, make a genuinely informed decision about whether to continue? If yes, you've built a preview. If the preview is accurate but the remaining steps are padded to increase the sense of investment, you've built something closer to a trap. Croatia's own digital products — the tourism booking platforms, the public-service portals, the growing SaaS cluster in Zagreb and Split — are subject to the same dynamic, and the ones that get onboarding right tend to be the ones that treat the preview as information rather than persuasion.

Designing the preview: what the evidence suggests works

If you're building this into a flow, the research points to a small number of design decisions that carry most of the weight.

Show the user's own output, not a generic mockup

The SaaS study above used the user's own data, populated. This matters because a generic screenshot is abstract — it shows what the product can do, not what the user will have. Personalised previews convert the abstraction into a concrete claim about the user's specific outcome, and concreteness is what makes the preview legible.

Place it after the first real commitment

Step 3 works because steps 1 and 2 have already cost the user something small. The preview then arrives as a return on that investment rather than an unsolicited pitch. If your flow has only two steps, the equivalent placement is immediately after the first meaningful action, not before it.

Keep the preview honest and specific

Vague previews — "unlock your full potential" — fail because they reintroduce ambiguity rather than resolving it. The preview should name the specific end state: this dashboard, this published page, this exported report. Specificity is the whole mechanism.

Don't repeat it

A preview shown once resolves uncertainty. A preview shown at every step trains the user to expect rewards and, when they stop arriving, produces the defection that variable-ratio schedules are famous for. One preview, one placement.

Measure retention, not just completion

Completion is a leading indicator and it is easy to move. Retention is the test of whether you resolved uncertainty or manufactured it. If the lift doesn't survive thirty days, the preview is doing something other than what you intended.

What this means for how you build

The step-3 finding is a specific instance of a general principle that runs through both behavioural psychology and interface design: people do not abandon effort because the effort is too great, but because the payoff is too vague. The fix is rarely to reduce the number of steps. It is to make the destination visible at the moment the user is deciding whether to keep walking.

For anyone building onboarding in Croatia right now — whether that's a B2B tool, a booking flow, or a public-service form — the practical implication is to stop treating the tutorial as a checklist and start treating it as a sequence of decisions the user is making about whether to continue. Map those decisions. Find the point of peak uncertainty. Put something true and specific there. Then measure whether the people you retained actually understood what they were building toward, because that is the only version of this tactic that compounds.

The next frontier is adaptive previews: flows that show different end states depending on what the user has already done, and that adjust placement based on where individual users tend to drop. That's a harder build, and it raises its own questions about how much a system should infer from partial behaviour. But it's the direction the evidence points, and the teams that get there first will be the ones who treated the preview as a genuine answer to a genuine question rather than a lever to pull.