Most people who leave your website should leave. They were looking for a different kind of business, they were checking your opening hours, they were a competitor, or they were a student researching something for an assignment. No message, however clever, turns those visits into customers, and trying to stop them just irritates people.
But inside that stream of departures is a smaller group worth caring about. They were interested. They got as far as the pricing page or the service they needed. And then something made them hesitate, and hesitation on the web looks exactly like leaving.
That hesitant group is the only realistic target when you try to reduce website bounce. The job is not to trap everyone. It is to offer one well-timed piece of help to the people who might take it.
Why visitors leave, and which reasons you can do anything about
Before designing any recovery message, it helps to be honest about the reasons.
| Why they are leaving | Can a message help? | What would actually help |
|---|---|---|
| Wrong business or wrong service | No | Clearer headings so they realise sooner |
| Got what they needed (hours, address) | No, and that is fine | Nothing, this is a success |
| Price looked high or unclear | Sometimes | An answer on what is included, or a range |
| A specific doubt the page did not answer | Often | A quick answer to that doubt |
| Not ready yet, still researching | Occasionally | A way to leave a question for later |
| Page was slow or broken | No | Fix the page |
| Form looked like too much effort | Often | A shorter route, such as asking in chat |
| Distracted, phone rang | No | Nothing you can control |
Look at the right-hand column. Several of the fixes are not messages at all. If lots of people leave your service page, the better headline, the visible price range and the shorter form come first. A recovery message is for the doubts that remain after the page is as good as you can make it.
The single offer that works better than a pop-up
Exit-intent recovery grew up in online retail, where it mostly meant a big box offering money off in exchange for an email address. For a service business, that approach tends to misfire. It cheapens the offer and it asks for something before giving anything.
A better pattern is one small, dismissible offer of help, tied to the page the visitor is on. The wording carries most of the weight:
- On a pricing page: "Not sure which option fits? Ask me and I'll explain the difference."
- On a service page: "Any questions about [service] before you go? I can answer straight away."
- On a booking or enquiry form: "Stuck on anything in the form? I can take your request here instead."
- On a location or coverage page: "Want to check we cover your postcode? Just type it in."
Each one names something specific the visitor might be wondering. None asks for an email. None offers a discount. And none appears twice.
Timing and the rules to set
Getting the message right is half the job. The other half is showing it rarely enough that it stays useful.
Once per visit, at most. If they dismiss it, it is gone. Showing it on every page turns help into nagging.
Only on pages where hesitation is likely. Pricing, key services, booking forms. Not the blog, not the contact page, not the privacy policy. The guide to which pages need an assistant is a good way to pick them.
Not in the first few seconds. Someone who arrives and leaves within moments was probably in the wrong place. Save the message for people who spent real time on the page.
Desktop and mobile handled differently. Exit intent works by spotting the mouse heading towards the browser controls, which does not happen on a phone. On mobile, a gentle greeting after a period of reading is a more honest substitute. When to nudge with proactive greetings covers timing across devices.
SpideyChat includes both a proactive greeting and an exit-intent greeting, so you can set one wording for each and keep them page-specific. The rest of the options are on the features page.
A worked example of what recovery is worth
Numbers here are made up to show the method. Replace them with yours.
Say your main service page gets 1,500 visits a month. Your analytics show that around 600 of those visitors spent more than 30 seconds on the page and then left without enquiring. That 600 is your hesitant pool; the quick exits are not.
Suppose the exit-intent greeting is shown to those 600, and 5 in every 100 open the chat. That is 30 conversations. If one in three of those conversations ends with the visitor leaving their details, you have 10 extra enquiries a month.
If you normally win one job in four enquiries and a job is worth $400, that is roughly 2 or 3 extra jobs, around $1,000 of work a month. Change any number and the result moves a lot, which is exactly why you should measure your own instead of trusting anyone else's.
Signs you have overdone it
Recovery messages have a failure mode, and it is easy to miss because the damage is quiet.
Watch the transcripts for visitors typing things like "no thanks" or "stop" into the greeting. Check whether time on page or pages per visit dropped after you switched the message on. Look out for complaints on reviews or social media about pop-ups. And if the conversations the message starts rarely lead anywhere, the wording is probably too vague.
When any of those show up, pull back: fewer pages, later timing or softer wording.
Set up one message and watch it
Choose your single most important page, usually pricing or your best-selling service. Write one recovery message that names the doubt visitors most often have on that page. Show it once per visit, only after they have spent real time there. Leave everything else untouched for a month, then count the enquiries that began with that message and read every transcript it started. You can try an assistant on your own site first to see how it sits on the page.