A website can look finished and still be broken in the only way that matters. The design is signed off, the pages are populated, everybody is pleased, and for three weeks the enquiry form sends to an inbox nobody owns. Growth Bully, a Malta performance marketing agency, inherits sites in that state regularly, and the damage is never in the design. It is in the journeys nobody tested.
Launch testing is not a design review. It is a check that the paths carrying money survive contact with production.
What should you test before a website goes live?
The journeys that make money, in production conditions, before anything else. Forms that reach a real inbox, tracking that fires into the live account rather than a test one, payment or booking flows completed end to end, and the search settings that decide whether the site can be indexed at all.
Everything else is secondary. A slightly wrong heading size costs nothing. A quote form that fails silently on mobile costs you every lead you paid to generate until somebody notices, and the people it failed for do not come back to tell you.
How do you test a lead form properly?
Submit it as a stranger would and follow the lead all the way out. A real submission, from a phone, on mobile data, with a genuine email address. Then check the notification arrived, the autoresponder sent, the record landed in the CRM, and nothing went to a spam folder.
- Submit from a phone on mobile data, not from the office network on a desktop. Mobile is where most enquiries come from and where most forms break.
- Use a real address you can open. Test addresses often sit on an allowlist that hides a deliverability problem.
- Trigger the validation deliberately. Leave a required field empty, enter a malformed email, and confirm the error is visible rather than silent.
- Follow the record into the CRM. A notification email proves the form sent. It does not prove the lead was captured anywhere you can work it.
- Check the reply path. Make sure a reply to the notification reaches the enquirer rather than a no reply address.
Do this for every form on the site, including the ones nobody expects to be used. The newsletter box and the careers form fail just as often, and they fail unnoticed for longer.
What breaks most often on launch day?
The invisible things, because nobody looks at them. Tracking left in test mode, form notifications pointing at an inbox that belonged to someone who left, the staging block on indexing shipped to production, old URLs with no redirect, and phone numbers that are text instead of tappable links.
Two of those are quiet catastrophes. A staging noindex that ships to production removes the site from search entirely and can go unnoticed for weeks. Missing redirects throw away every link and every ranking the old URLs had earned. Both are a ten minute check and both are skipped constantly because the site looks perfect while they are happening. Ask for them in writing whenever a new build is handed over.
Which defects should actually delay a launch?
Only the ones that break revenue or indexing. A broken enquiry form, a failed payment step, tracking that does not fire, or a site that search engines cannot crawl all stop the launch. Spacing, a slightly wrong shade and a missing team photo do not. Sort every defect into those two piles.
That sorting is what keeps launches moving. Teams that treat every defect as equal either ship late for cosmetic reasons or ship on time with a dead form, because the real problems were buried in a list of forty items. Two piles, one rule: does this stop a customer or a crawler.
Who signs off that a website is ready?
One named person on the business side, against written evidence, not a walkthrough call. Sign off means someone looked at a screenshot of the test enquiry in the CRM and the tracking event in the live account. Approval given on the strength of how the site looks is not approval of anything measurable.
Ask for the evidence pack before you approve: the test enquiry in the CRM, the conversion event recorded in the live property, the redirect map from old URL to new, and confirmation that indexing is open. If your builder cannot produce those four things, the site has not been tested, whatever the call said. The same discipline is what makes tracking survive the next change rather than breaking again at the first redesign.
What should you check in the first 24 hours after launch?
Repeat the money tests on the live site, because production is not staging. Submit another real enquiry, confirm the first genuine visitors are recorded, and watch for traffic arriving at URLs that no longer exist. On a pest control launch we ran, sales were tracked inside 24 hours of going live.
That only happens when the measurement was proved before the switch, not after it. Put a reminder in the calendar for the same checks at day seven as well, because consent banners, caching layers and plugin updates all change behaviour after launch day. If the first week of data looks wrong, assume the tracking before you assume the market, and read it alongside how your landing pages convert rather than in isolation.
If you are approving a build soon and want to know which checks actually protect your pipeline, send us the launch date and we will tell you what to demand before you sign.