You pay for the first 90 days either way. If the pages Vera improved don’t end up ahead of the pages she left alone, the next 90 days are free.
Applied automatically: nothing to claim, nobody to email. Once per customer. The first three months aren’t free and aren’t refunded; the free ones are the next three, and only if the first didn’t work.
| Months 1 to 3 | Months 4 to 6 | |
|---|---|---|
| The improved pages came out ahead | Paid | Paid |
| They didn’t | Paid | Free |
This page is dated and versioned on purpose. If the rules ever change, the change gets a new version and a date, and plans already running keep the version they started under.
A guarantee that can be reinterpreted after the result is known isn’t a guarantee.
So everything that decides the outcome is written here, before any result exists, in words you can check against your own results page.
The pages Vera improved ended up with more visitors than the pages she left alone, each compared with how it was doing before your plan started. That’s the whole test. It depends on which way it went, not on how sure we are.
You’re billed as normal and the next plan starts. Your results page still shows the likely range, and says plainly whether the result can be told apart from chance.
Your next 90 days are free. We undo the improvements that didn’t help, keeping only the fixes to things that were plainly wrong, like a dead link or a 2023 price.
No guarantee either way, and your results page says why. These rules are shown to you before you agree to your plan, not discovered afterwards.
Why it’s about which way it went, not certainty.
With around 40 pages on each side, the numbers are noisy enough that a real, worthwhile improvement often can’t be told apart from chance in 90 days. If the promise depended on certainty, we’d be giving away most second quarters while the work was actually working. So it depends on which way it went: simple, unarguable and something you can check yourself. How sure we are is published alongside it, every time.
Visits from Google search, as Google reports them, once its figures have settled. Not how often a page was shown, which can rise while visits fall, and not average position, which hides the thing being measured.
Each page against how it was doing in the 90 days before your plan started, averaged across pages. So one page having a big week doesn’t decide the result, and a quiet season affects both groups equally.
Days 1 to 90 of your plan, leaving out the last two days, because Google revises its most recent figures.
Pages you edited yourself, pages you advertised during the 90 days, pages that stopped loading, and pages with too few visits to show anything in 90 days. They’re reported as too quiet to call, rather than counted.
If one page brings most of your visitors, its weight is capped and your results page says so.
If you approved fewer than the minimum number of improvements, there was nothing to measure. Someone who publishes nothing can’t collect.
Vera checks those pages throughout. Telling us is fine: the page is simply dropped from both groups. Changing one without telling us means the result can’t count.
A change of domain, or a move to a new website builder, during the 90 days.
Before the 90 days were up, so the final figures couldn’t be read.
Small websites with few visitors are exactly the ones where nothing can be measured in 90 days. If the free check finds your website is under the minimum number of pages or visitors, we tell you before you pay, and either say no or offer the work without the guarantee.
We don’t guarantee a particular increase, a ranking or a number of visitors. We don’t control Google. What’s guaranteed is that the test was run properly, that the pages left alone were never touched, that the result is reported whichever way it came out, and that a failure costs you the next three months’ fee rather than your goodwill.