What to do after a website launch: a calendar for the first thirty days

Launch is not the finish, it is when the data starts

Before launch every decision rested on guesses: we thought people needed this, so we arranged it that way. After launch guesses give way to observation, and the first month is the most informative of all.

It is also the riskiest. Mistakes missed at handover run for the whole of it: a form landing in spam, a page the search engine cannot see, an analytics script that never installed.

Below is the order of checks by day and week. It is built so the most expensive thing is checked first: what is losing money right now, then what will matter in months.

And separately: what counts as normal fluctuation and what counts as an actual fault. Most of the panic in the first weeks is caused by normality mistaken for an emergency.

Day one: six checks in an hour

Done on launch day, an hour in total.

Send an enquiry through the form. Not check that the button presses, but follow it to the end: did the email arrive, is it in spam, did a record appear where it should. Repeat from a phone.

Open the site on a phone over mobile data rather than home wifi. Half of all speed problems are visible only that way.

Check the site opens both with and without www, and that one leads to the other rather than both working separately.

Request the sitemap file in the browser and confirm it is served and contains your pages. We once broke ours with an edit that looked harmless — the check takes thirty seconds.

And look at the page source of any page for an indexing block. Staging versions are closed to search, and that block regularly rides along into production with everything else.

Week one: make sure the search engine found you

The most common loss of a first month is a working site nobody sees, because nobody told anyone about it.

Add the site to the search consoles and verify ownership. It is done once and unlocks everything else: indexing reports, crawl errors, queries.

Submit the sitemap in the console. Do not rely on the crawler finding it through robots.txt: it will, but later.

Check several key pages individually through the URL inspection tool. It shows whether the search engine can see the page and what exactly it sees in it.

And confirm analytics tags are on every page type, not just the home page. A missing tag on product pages usually surfaces a month later, when the data can no longer be recovered.

What appears in the console, and when

So you do not go looking for data before it physically exists.

First twenty-four hours: usually empty. That is normal and means nothing.

First week: crawl data appears along with the first indexed pages. Usually the home page and whatever has the most internal links pointing at it.

First month: query impressions appear. There may be no clicks at all, and that is also normal: an impression in twentieth place does not produce one.

And keep in mind that console data lags two to three days. We exported the report three days running and got two identical exports in a row. Looking more often than weekly is pointless.

First enquiries: check the route, not the send

A form showing thank you and an enquiry reaching a human being are two different events.

Between them sits a chain: submission, email, spam filter, integration, notification, person. Any link can fail silently.

So in the first month, send a test enquiry yourself once a week and walk the whole route: did the email arrive, did it land in the right folder, did a record appear, how many minutes before somebody saw it.

Check separately what happens to an enquiry outside working hours. An auto-reply stating when somebody will be in touch is cheap and holds the people who would otherwise write to a competitor.

And time the real gap to first reply. It is the most underrated measure of a first month: enquiries are lost less often on the site than in the pause between arrival and call.

What to read in analytics, and what to ignore: The first month after launch is the most informative and…

In a first month there is little data, and most conclusions drawn from it will be wrong. So a short list of what is genuinely informative.

Worth reading: which pages people leave without acting; which pages were never opened at all; how many visits come from phones against computers; which search queries bring people in.

Not worth it: comparing conversion by day of week, drawing conclusions from twenty visits, chasing bounce rate on pages where bouncing is the normal behaviour.

Pages nobody opened all month are the most useful finding. Either nothing links to them or they are not needed. Both conclusions save money.

And watch a few session recordings in full if you have that tool connected. Ten recordings explain more than a week of staring at graphs.

Normal fluctuation and genuine warning signs

A distinction that removes half the unnecessary panic.

Normal: traffic swinging several times over from day to day at low numbers; positions on new queries appearing and vanishing; pages indexing one at a time over weeks rather than all together.

Normal after a migration: a twenty to thirty per cent dip in the first weeks while the search engine carries accumulated signals from old addresses to new ones.

Worrying: a fall concentrated in one section rather than spread; the count of indexed pages flat for a second week running; enquiries stopping abruptly while traffic holds.

That last one almost always means a technical fault in the form or the enquiry route rather than a change in demand. It takes five minutes to check with a test submission.

Old addresses, if the site moved

This section is for anyone who had a site before this one. Work that cannot be deferred.

Collect the old site's address list from two sources: the previous system and the search console. The lists will not match — the search engine remembers what you have forgotten.

After our own move there were 673 such addresses. Each needed its own decision: is there a page on the new site that answers the same question?

If there is, set a permanent redirect to it. If there is not and the page is gone for good, 410 is more honest than 404: it states directly that the address is not coming back, and the search engine clears it from the queue faster.

What not to do: redirect everything to the home page. Search engines treat that as a soft error, decline to index anyway, and you lose the data about what people were looking for.

Speed: measure on visitors, not on your own machine

A site that opens instantly for you may take four seconds for half your visitors.

The difference is that you have fast internet, a recent device and a warm cache. The speed measures search engines use are collected from real visits, not from a laboratory.

So in the first month read the field data report once enough visits have accumulated. Until then use a lab measurement, remembering that it is the optimistic one.

And check the three things that most often spoil speed on a new site: uncompressed photographs, fonts that load ahead of text, and third-party scripts added at the last minute.

What to add in the first month, and what not to

The urge to keep finishing things is strong, but a first month is better spent otherwise.

Worth adding: pages for the specific queries that have already produced impressions; answers to questions people ask in enquiries; a price or a range if there is none yet.

Not worth it: changing the address structure, rebuilding navigation, adding language versions. All of that changes what the search engine has only just begun to learn and discards part of the work it has done.

The rule is simple: in the first month add, do not rebuild. Schedule rebuilding for the third month, when there is data to base it on.

A calendar for the first thirty days

A summary you can print and tick off.

Day one: an enquiry from the form and from a phone, a mobile-data check, address consolidation, the sitemap, the indexing block.

Week one: consoles and verification, sitemap submission, key page checks, tags on every page type, redirects from old addresses.

Weeks two and three: a weekly test enquiry, first indexing reports, the list of pages nobody opened, a speed measurement.

Week four: first queries in the report, pages added for them, a decision on what moves to month two. And no rebuilding before month three.