Effect SEO Records

What your fixes have produced. Measured every month.

You fix your SEO. A month later, the only question that matters is: did it move? Effect SEO Records audits your entire site every month and gives you the gap against the previous survey — fixed, appeared, falling back, improving, page by page.

The eight records €149 for the year →

Six pieces, and they turn on their own.

A merchant who fixes their site does it once, then moves on to something else: it's the measurement that has to come back on its own. This one comes back every month, and it comes to you.

The survey

Your entire site, read the way a search engine reads it: titles, descriptions, canonical URLs, redirects, structured data, images, speed. Up to ten thousand pages, as of today's date. It's the same survey as the Audit SEO Records — the one you can buy as a one-off.

The gap

The month's report is compared with the previous month's, signal by signal and page by page: for each page, the title, description, response, indexing and canonical as they stood then, then as they stand today. An address that redirects to a page in today's report is called moved, its redirect followed. Two reports go in, one difference comes out: the calculation states what the two of them state, and it stops there.

The report that comes to you

You receive the survey by email, every month, with the gap up front and the full report one click away. Your job is to fix; ours is to tell you what the fix achieved.

The baseline

The first survey goes out within twenty-four hours: it's your starting point, the one the eleven that follow will be compared against. A measurement needs an origin, and the origin is set before the fixes.

Monitoring, between two reports

Every day, monitoring re-reads the pages cited by your latest report, the home page and robots.txt — some forty addresses, re-read one by one: your site is watched over, it stays quiet. One response, one destination address, one title, one noindex or one canonical that changes, and the email goes out that same day. If it was intended, monitoring picks up from there.

The certificate

Every report goes out sealed: the dated record of your site's condition, attached to your email. You keep it, and anyone can verify it on their own machine, offline, in their browser. It's the document you show to whoever asks you to account for yourself.

The gap

Five movements, and two surveys are enough to state them.

The gap is a pure calculation: two reports of the same site at two dates go in, five movements come out. It says what the two surveys say, and it stops there — that's what makes it defensible.

Fixed

A signal present in the previous survey, absent from this one. The fix worked, and the date it worked is written down.

Appeared

A signal the previous survey knew nothing about. A redesign, an import, a module added: the month it happened is the month you find out, rather than the following quarter.

Falling back

A signal still there, but on fewer pages than before. The exact count for both dates is given: thirty-eight pages yesterday, nine today reads at a glance.

Improving

A signal gaining ground: present on more pages than at the previous report. It's the line you want to see early, and it arrives that very month.

Pages back in the clear

The URLs that the signals cited in the previous report and have stopped citing. The list names them: these are your pages, put right, one by one.

The records — reading of 07/09/2026

Eight records, and all of them answer the same question.

For a product that measures, only one question counts: did every reported change actually happen? Each of these eight figures answers it, and each can be replayed with the command shown below. The day one of them fails, its line leaves this page.

100% of movements found

Every announced line traces back to the two readings

6,576 changes, checked one by one across 300 pairs of readings built by the very same computation that produces the reports served to customers. A "fixed" item must be present in the previous reading and absent from this one; an "appeared" item, the reverse; a signal that regresses must carry both exact counts; a page that has become clean again must have been flagged yesterday and no longer be. All 6,576 check out.

npm run records:effect

300 / 300pairs weighed in both directions

The differential weighs the same forwards and backwards

A computation that leaned toward good news would be a salesman's computation. On each of the 300 pairs, the delta is recomputed in the other direction: what is counted as fixed on one side must be counted as appeared on the other, and "regresses" must become "gains ground". The switch is exact on all 300.

npm run records:effect

900 / 900shuffles, a single verdict

The order of the signals leaves the verdict unchanged

Two readings of the same site list their signals in whatever order they come, and an order-sensitive computation would invent changes. The signals from both readings are shuffled at random 900 times, the differential is recomputed and checked against the original verdict — codes, counts, and named pages. The verdict holds all 900 times. This record failed at first: the reading of 07/09 showed 809 shifted verdicts out of 900, and sorting the pages before truncating them made it exact.

npm run records:effect

12readings delivered in 365 days

What is sold is what is delivered

A full year of sweeps, simulated pass by pass — four passes a day — with the product's real decision-making: a thirty-day cadence, holding the reading in flight, the six-hour belt, and one reading in twenty deliberately left hanging. 12 launches, 12 readings delivered, one at a time. A reading that drags is retried, and the following month arrives on time.

npm run records:effect

1 460 / 12round passes · reads of the site

The sweep wakes your site twelve times a year

The sweep runs four times a day, 1,460 times a year. It requests a reading from the measured site twelve times: the other 1,448 passes look at the state of the reading on our side, and your site ignores them. A product that measures a site should cost it as little as possible — and this count says so.

npm run records:effect

100% of the figures in the email

Every figure in the email comes from the differential

300 emails are generated from 300 real differentials, then re-read: each of the 1,372 figures highlighted in their text is checked against the count it comes from. An email that inflated a figure by one would be caught here. All 1,372 trace back to their source.

npm run records:effect

100% of attested fields

The sealed attestation says what the delta says

The seal is the document you produce against a third party: what it certifies must be exactly what the difference contains. 300 certificates are sealed from 300 real differences, then re-read field by field — pages read, signals, fixed, appeared, declining, improving, pages back in good standing, and the difference page by page — changed, new, gone, moved. All 3,300 fields have their source. A certificate that embellished by a single unit would be worse than no certificate at all: it would be signed.

npm run records:effect

100% of the annual report fields

The annual report says what the twelve statements say

At the twelfth statement, the annual report goes out sealed: the twelve statements, and what changed from the first to the last — signals gone, arrived, declining, improving. 300 years of twelve statements are seed-generated, and each of the 7,200 fields in their reports is checked against an independent recalculation from the twelve summaries. All 7,200 have their source.

npm run records:effect

The third record failed at first. When measured on 7 September 2026, the reading showed 809 verdicts shifted out of 900 : the counts held up, but the list of pages that had become clean again was drawn in the order the signals arrived. Two readings of the same site could therefore name different pages. The pages are now sorted before being retained, and the reading returns 900 out of 900. A reading that finds something is a reading that has been of use.

World first

Every report is handed to you sealed.

Your monthly report comes with its attestation : the dated statement of your site's condition, Ed25519-sealed by the vendor. It's yours to keep. And anyone can verify it on their machine, offline, in their browser: verification belongs entirely to them.

What it says

Your domain, the report date, the pages read, the signals recorded, and the delta against the previous report — fixed, appeared, declined, improved, pages returned to clean. The same figures as your email, sealed.

What it proves

That this statement was delivered exactly as it stands on that date, and that it has remained intact since. Verification is byte-exact: it recognises the original record, and it alone. Ranking, for its part, remains whatever Google makes of these entries — the seal covers the measurement.

Who it's for

To show a client what their search visibility has become this month. To a partner, to a buyer during an acquisition audit, to anyone holding you to account. A dated document is worth more than a dashboard you'd have to be taken at your word for.

The verifier, online →The public key is printed on the page; the sealing key stays with the publisher.

What you get.

Twelve reports on your entire site over the year, launched automatically, up to ten thousand pages each — the same report as the Audit SEO Records, on the same date each month.
Eleven dated change reports, from the second report to the twelfth: fixed, appeared, declining, improving, and your pages back in the clear, named.
The difference page by page: for each page that has changed, the field, the previous value and today's — title, description, response, destination address, indexing, canonical — and the moved addresses, followed through their redirect.
Daily monitoring between two reports: the pages cited by your latest report, the home page and robots.txt are re-read every day, and any change of response, indexing, title or canonical reaches you that same day.
The monthly email, with the changes up top and the full report one click away — readable, archived, and something you can show anyone who asks what SEO delivered.
The reading is done from our end, from the outside: your site keeps its files and its state, and it is measured exactly as search engines see it.
Twelve sealed certificates, one per report: the dated record of your site's condition, which you keep and which a third party verifies on their own machine, offline — the document to show a client, a partner or a buyer.
Search Console, connected in under a minute: you grant our account read access to your property, and every report compares your fixed pages against your control pages, before and after — clicks, impressions, click-through rate, each with its 95% confidence interval. When volume is insufficient, the report says so, and states the volume that would be required.
Every report also reaches you as a PDF — the report, the variance, page by page, the year's curve and the seal reference. On the Agency tier, the PDF carries your agency's name at the top: the document to hand over to your clients, and the measurement stays sealed.
The year's curve in every report, statement by statement: pages read, signals, and each signal tracked month after month. At the twelfth statement, the annual report goes out sealed: what changed from the first to the last, verifiable offline just like the attestation.

What we measure, and what belongs to Google.

What we measure

Everything Google reads on your pages: every signal recorded across your entire site, as of a given date, and what changed between two reports. The calculation can be replayed, both reports remain readable, and every line cites the pages it is based on.

What belongs to Google

Rankings. That's what Google does with these inputs, and it also depends on your content, your links, and your competitors. Anyone selling you a position is telling you too much; we sell the measurement, and it can be verified.

Twelve reports for the price of three.

You pay once, for one year, and the price shown is the price you pay. The year runs to the day, then stops on its own. The first report goes out within twenty-four hours, the other eleven follow by themselves. A ten-thousand-page audit costs €49.90 on its own: the year includes twelve, for €149. Each report reaches you sealed — the certificate is included, and its verifier is public.

One site

149 €for one year

Paid once: the year runs to the day, i.e. €12 per month. One domain, twelve reports of ten thousand pages, eleven dated change logs and twelve sealed certificates — the price of three standalone audits, for twelve.

Take a year

Three sites

Recommended

349 €for one year

Paid once, for one year. Three domains, i.e. ~€116 per site. Each one recorded, compared and reported separately, with its own monthly email.

Take a year

Agency

799 €for one year, up to 10 sites

Paid once, for a full year: ten domains fully monitored, your clients' or your own — that's about €80 per site. Each one gets its twelve reports and variances, as a PDF bearing your agency's name, and we stand behind them.

Take a year

You provide your domain at checkout. The three- and ten-site tiers cover the others: reply to the welcome email with the additional domains, they enter the reporting cycle that same day and each receives its own.