How to Improve Customer Onboarding by Subtracting

I work at Moonb, a creative studio, and a big part of what we make is onboarding and explainer videos for software companies. So I have watched a lot of onboarding flows up close, from the inside, right at the moment a real person decides whether to keep going or close the tab.

Here is what I have come to believe, and it runs against almost every “16 tips” article you will find. Improving onboarding is mostly an act of subtraction. Fewer steps and fewer things to read before a person gets to the point. The window to prove your product is worth it is much shorter than teams assume, and every extra thing you add spends time you do not have.

Your onboarding window is measured in days, not weeks

Most teams plan onboarding like it is a semester. In reality it is closer to a first date.

Amplitude’s 2025 Product Benchmark Report, which looked at more than 2,600 companies, found that Day-1 activation tops out around 21% even at the 90th percentile, and it falls to roughly 9% by Day 14. Read that again. The best products in the study still lose the majority of new users in the first day, and the number keeps sliding through the second week. There is a related “7% rule” in Amplitude’s data: if at least 7% of a new cohort still comes back on day 7, you are already in the top quartile for early value.

So here is the framing I use. You have days, not weeks. Sometimes you have a single session. Whatever your onboarding is going to accomplish, it needs to accomplish most of it fast, because the audience thins out by the hour.

That single fact should change how you design the whole flow. If the clock is that tight, a five-screen product tour is not thoroughness. It is you spending your scarcest resource on things the user did not ask for yet.

Why customers actually quit during onboarding

It is tempting to blame the user. They were lazy. They were tire-kickers who never really needed us. I have sat in those retros. It is comforting and it is almost always wrong.

When people abandon onboarding, it usually traces back to one of four things, and none of them is laziness.

Misaligned expectations. They signed up believing the product did one thing, and the first screen tells a different story. The gap between the promise and the first click is where trust leaks out.

Cognitive overload. Too many choices and too many fields to absorb before anything useful happens. The brain treats a wall of setup the same way it treats a wall of legal text. It looks for the exit.

No early engagement. Nothing has happened yet that felt like a win. The user has given inputs but received nothing back that made them think, oh, that is handy.

No sense that this is for them. A generic path that ignores who they are and why they came. A solo founder and a 200-person marketing team get the exact same eleven steps, and both feel like the product was built for someone else.

Notice that three of those four are about pace and relevance, not features. People rarely quit because a product lacks capability. They quit because getting to the capability took too long or felt like work.

Stop adding steps: the friction math nobody wants to hear

Here is the part that stings, because it means undoing work teams are proud of.

Flow length predicts completion, and the drop-off is brutal. In onboarding benchmarks compiled by Userpilot, roughly 3-step flows complete near 72%, while 7-step flows fall to about 16%. Four extra steps, and you have quartered your completion. That is not a rounding error. That is most of your funnel.

The instinct when onboarding underperforms is to add. Another tooltip to explain the confusing part. Another checklist item to nudge. Another modal to celebrate. Every one of those feels helpful in isolation. Stacked together they are the reason the flow is failing. You are treating the symptom of confusion by adding more things to be confused by.

So before you add anything, do the boring audit. Write down every step between signup and the first moment the product does something valuable. Then, for each step, ask one question. If I deleted this, would the user still reach value? A shocking number of steps exist to serve the company (collect this field, show off that feature, ask for this preference) rather than the person. Cut those. Move them to later, after the user already cares.

Subtraction is unglamorous. Nobody gets promoted for the modal they removed. But it is the single highest-leverage thing you can do to an onboarding flow, and it costs you nothing but ego.

Anchor everything to one metric: time to first value

If you only track one number for onboarding, make it time to value: how long from signup until the user gets a real result they came for. Not “finished the tour.” Not “connected an integration.” A result. The email got sent, the report generated, the design exported, the thing they actually wanted.

The benchmark here is sobering and useful at the same time. Userpilot’s 2025 study of 547 SaaS companies found a median time to value of roughly 1 day and 12 hours, while top performers get a user to first value in under five minutes. That is an enormous spread, and it maps almost perfectly onto who retains and who does not. The products that pull ahead are not carrying the most onboarding content. They compress the distance to the first good outcome.

Once you make time to value the north star, a lot of decisions get easier. Any onboarding element either shortens that distance or it does not. A field that delays first value is now clearly a cost, not a feature. A short piece of guidance that gets someone unstuck faster is clearly worth it. The metric turns arguments about taste into questions of arithmetic.

Where a 40-second video beats every tooltip you could build

This is where I am biased, so I will be upfront about it. We make video, and I think video earns its place in onboarding for specific, unsexy reasons, not because it looks premium.

A tooltip can point at a button. It cannot show a hand actually doing the thing, in the right order, with the messy real interface, in the time it takes to make a coffee. Some concepts are motion. “Drag this here, then it snaps into that” is three confusing tooltips or one obvious eight-second clip. When the action is spatial or sequential, showing beats telling by a wide margin.

Length matters more than people think, and shorter wins. In Wistia’s State of Video report, videos under one minute hold the strongest engagement, averaging around a 52% engagement rate, and it drops off sharply as videos get longer. So the move is not a single fifteen-minute walkthrough that nobody finishes. It is a set of tight, contextual clips, each one dropped at the exact moment a user needs it. Right video, right place, forty seconds, done.

There is a second advantage that live help never delivers: a well-made clip explains the tricky step the same way at 3am to a user in another timezone as it does to the one who signed up while you were watching. Captions make it work with the sound off and across languages, which for most software audiences is not a nice-to-have. If you are weighing where video fits alongside written docs, we get into that trade-off more in our piece on customer education strategy.

The five onboarding moments actually worth filming

You do not need a video library. You need maybe five, placed with intent. These are the moments where a short clip does the most work.

1. The welcome and expectation-set. Fifteen to thirty seconds, right at the start, that says plainly what this product does and what the next few minutes will feel like. This directly attacks the misaligned-expectations churn. You are closing the gap between why they signed up and what they are about to see.

2. The first setup or first key action. The first thing the user has to do that is not trivial. Connect a source, import data, configure the one setting that matters. This is where cognitive overload strikes, so a short clip that shows the action end to end removes the “wait, what do I do here” pause.

3. The aha moment. The instant the product delivers its first real result. A clip here amplifies rather than explains. It frames the result so the user recognizes that yes, that was the point, that was value. Do not narrate the whole app. Narrate the win.

4. The predictable stuck point. Every product has one. The step where support tickets cluster, the setting everyone misconfigures, the screen where session recordings show people hovering and hesitating. Film that. This single clip deflects more tickets than the other four combined, because it meets confusion exactly where it reliably happens.

5. The “what’s next” after activation. Once someone has their first win, a short nudge toward the second valuable action turns a one-time success into a habit. This is the bridge from activated to actually adopted.

The discipline across all five is the same. Short and contextual beats long and comprehensive. A user will watch forty seconds that solves the problem in front of them. They will not watch twelve minutes about problems they do not have yet. If you want the mechanics of scripting and shooting these, we wrote a fuller guide on how to make instructional videos.

Make it self-serve so it deflects tickets and scales

There is a cost argument here that finance people care about even if product people do not lead with it.

Gartner data (reported via Zendesk) puts live support at around $8.01 per contact versus roughly $0.10 for self-service, and finds that good self-service can cut support costs by up to 25%. Onboarding is where a huge share of early tickets are born. The stuck points. The “how do I” questions. The misconfigurations. Every one of those you answer once, well, in a clip that lives next to the moment of confusion, is a ticket that never gets filed.

The pattern that works is video plus a searchable knowledge base, not video instead of it. The short clip carries the parts that are motion and sequence. The written doc carries the parts people want to skim or search for at 2am. Together they form a self-serve layer that handles the predictable questions so your team can spend human time on the hard, non-repeating ones. This is also where FAQ videos pull their weight, because the same questions recur across nearly every new cohort.

How to tell if any of it worked

You do not need a dashboard with forty tiles. Watch these four numbers and you will know whether your changes helped.

Metric	What it measures	Benchmark to aim at
Time to first value	Signup to first real result	Median is ~1.5 days; push toward minutes to hours
Day-1 / Day-7 activation	Who reaches value and comes back	7% still returning on day 7 is top quartile
Onboarding completion	Share who finish the core flow	~72% for a 3-step flow; it collapses as steps grow
Feature adoption / early churn	Second action taken; who leaves in week one	Trend matters more than the absolute number

Track them per cohort, before and after a change, so you can actually attribute improvement. If you cut two setup steps and time to value drops and day-7 return climbs, you have your answer. If you added a clip at the stuck point and tickets on that screen fall, that clip earned its production cost. Speaking of which, if you are trying to size the investment before you commit, our breakdown of educational video production cost lays out the ranges honestly.

If you only have one week, do this first

You will not rebuild onboarding in a week, and you should not try. Here is the order that gets you the most improvement for the least work.

Day one and two: map every step from signup to first value, then delete the ones that serve you instead of the user. This alone often moves the completion number more than anything else, and it is free.

Day three: find your single worst stuck point using support tickets and session data. Not five. The one where confusion reliably clusters.

Day four and five: get one short, contextual clip made for that stuck point, and place it exactly where people get stuck rather than in a help center nobody visits.

That is it. One subtraction pass, one well-aimed video. Measure the two cohorts. If time to value drops and day-7 return ticks up, you have proof, and you have earned the case to keep going.

At Moonb we make the onboarding and explainer videos that sit in flows like these, so if you reach the point where a specific moment clearly needs film, here is how we approach product demo videos. But please do the subtraction first. A great video cannot rescue a flow that is fourteen steps too long, and once the path is short, you will need far less video than you think.