Most advice about conversion is about persuasion — better headlines, stronger calls to action, urgency. That advice assumes the visitor wanted to buy and needed convincing.
In practice, a large share of the people who leave had already decided to act and then hit something that stopped them. Fixing those obstacles is cheaper than persuasion, and it works on people who were never going to be persuaded anyway.
The three questions every page must answer in four seconds
- What is this? Not your tagline — the plain category. “Project management for construction firms” beats “Reimagining how teams build.”
- Is it for me? Say who it is for, and ideally who it is not for. Excluding people builds more trust than including everyone.
- What do I do next? One obvious action. Not three of equal visual weight.
Test it honestly: show your homepage to someone for five seconds, take it away, and ask what the company does. If they cannot say, no amount of copywriting below the fold will help.
Where visitors actually drop out
| Stage | Common cause of loss | Fix |
|---|---|---|
| First 3 seconds | Cannot tell what this is | Plain-language headline |
| First scroll | Slow load; layout shifting under them | Reserve image space; defer scripts |
| Deciding | Price hidden | Publish pricing, or say why you cannot |
| Form | Too many fields | Ask only what you need to act |
| Error | Form clears, or vague message | Preserve input, name the field |
| Final step | Surprise fee or requirement | Show it at the start |
Speed is a conversion feature
People do not consciously abandon a slow page. They lose momentum, glance at something else, and never come back. The abandonment does not feel like a decision, which is exactly why it does not show up in any survey.
Three fixes cover most of it, and none require a rebuild:
- Compress and size images properly. Serve the size actually displayed, in a modern format. This is usually the single largest win and it takes an afternoon.
- Reserve space for anything that loads late. Content jumping as the page settles causes mis-clicks, and a mis-click at a decision point is a lost visitor.
- Audit third-party scripts. Analytics, chat widget, two pixels, a font service, a cookie banner. Each one was added by someone who is no longer thinking about it.
Test on a mid-range phone on a normal mobile connection — not your laptop on office wifi. That is where the audience actually is, and the difference between the two is enormous.
Forms: ask less, ask later
Every field you add costs you completions. Not proportionally — the drop is worst at the point where the person realises the form is longer than they expected.
Two rules that consistently hold up:
- Ask only what you need to take the next action. Company size, job title and phone number can be collected later, by a human, in context.
- Show the length. If it must be long, show progress. Uncertainty is worse than effort.
And label fields above the input, not inside it. Placeholder-only labels disappear the moment someone types, which means anyone who gets interrupted has to delete their answer to remember what was asked.
What builds trust, in order
- Real specifics. Actual prices, actual limits, actual customer names. Specificity is the cheapest trust signal there is.
- A visible way to reach a human. Most people will not use it. Its absence is noticed.
- Honest constraints. “Not suitable for teams under five” reads as confidence, not as a lost sale.
- Screenshots of the actual product. Not illustrations of abstract dashboards. People want to see what they are buying.
What does not build trust: stock photography of models in an office, testimonials with no name or company, and badges nobody recognises. These are so common that they now read as noise at best.
Common questions
Should we A/B test everything?
Only if you have the traffic to reach significance — realistically thousands of conversions per variant. Below that, tests produce noise that feels like insight. Fix the obvious problems first; they need no test.
Do pop-ups work?
They lift email signups and they cost you goodwill and mobile ranking. If you use one, trigger it on exit or after real engagement, never on arrival, and make the close button obvious.
How long should a landing page be?
Long enough to answer the objections a real prospect has. Write the objections down first, answer each one, then stop. Length is an output of that process, not a decision.
Where do I find the objections?
Support tickets, sales call notes, and the questions people ask before buying. Same source as most UX insight — we went through that in the five frictions.
Leave a Reply