Answer one question about your own site before you read anybody's case study

Before you read a vendor's case study, answer one question about your own site: does it already have pages in Google's index and links pointing at them? A yes and a no put you in two different populations, and a result measured on the population you are not in cannot forecast anything for you however accurate it is. Our proof page is built around that split. It carries two cases, one domain with history measured over 92 days and one domain with none measured over 9 days, published as two separate case studies and deliberately never combined into one. The reader's job is to work out which of the two they are, and this page is how.

The question, asked properly

It takes two checks and about three minutes, and it needs doing on your own site rather than estimated.

• Indexed pages. Open the Page indexing report in Google Search Console and read the indexed count. Not the number of pages you have published, the number Google kept. The gap between those two is often the whole story.
• Inbound links. Read the Links report in the same property. What matters is whether anything outside your own domain points at you at all, and whether those links sit on pages that themselves have standing.
• Age. How long the domain has been resolving, which is a separate fact from how long the current site has existed. A rebuilt site on a ten year old domain is not a new domain.

If the answer is yes, the case with history is the one that speaks to you

The clawlaw.in case covers 2026-07-13 to 2026-10-12, 92 days with data, exported on 2026-10-14. It records 18,537 clicks and 929,405 impressions, with the daily click figure moving from 59 at the start of the window to 349 at the end. Monthly daily averages run 59, then 154, then 282, then 349.

That domain had history and existing pages before the window opened. So what the case demonstrates is compounding on a working site: whether each month beat the one before it. If your own site has an index footprint and some links, that is the question your next ninety days will be judged on, and the thing to agree in advance is the monthly reading rather than a target.

What the case cannot tell you is how long it takes to get to a starting point of 59 clicks a day, because everything that produced that figure happened before the first row of the file.

If the answer is no, the case with no history is the one that speaks to you

The aiknowsus.com case covers 2026-10-04 to 2026-10-12, 9 days with data, exported on the same date. It records 5,965 impressions and 3 clicks, with a best day of 11 October 2026 at 2,685 impressions and an average position of 41.6. That domain was registered days before the window opened: no history, no links, no age.

The 5,965 is a discovery result and it is not traffic. Impressions mean Google is showing the pages. Clicks follow position, and position takes longer than nine days, which is exactly why the click column reads 3 and is published anyway.

So what this case demonstrates is speed of discovery from nothing, and that is the only question a brand new site can have answered in its first fortnight. Judging it on clicks is judging the wrong column at the wrong time.

Why two unlike cases are more use than one blended one

The instinct, when a company holds two results, is to average them into a single claim. That would destroy both. The value of each case here is precisely the thing the other one is not: one has a starting base and the other has none, and a figure drawn across both would describe a site that does not exist.

Kept apart, they do something a single number cannot: they let a reader self select. Two cases on two unlike domains give a buyer a decision to make rather than a claim to accept, and the decision is the more useful object.

Putting the same split to work on a supplier's own result

The same split applies to work you are being sold. Ask what the site was on day one of the window: its age, its indexed count, whether it had links. Then ask whether anything else ran in the same period, and what share of the clicks were brand searches. The general version of this reading, including what to ask when the chart has no axis values, is set out in how to read a case study and how to write one that earns trust.

And if the answer to the opening question was no, there is a second decision waiting behind it, which is whether to build the measurement in house or buy it. A new site has fewer existing systems to protect and more of the work still ahead, and the five wrong answers an in house build arrives at quietly is the part of that decision most teams discover late.

Our commercial interest in all of this is stated plainly: we sell the measurement and the work behind both of those cases, which is why each one is published with its own window, its own export date and its own limits rather than as a single reassuring figure.

Deciding which of two published cases fits a site like yours

What if my site is somewhere in between?

Most are. Read the case nearer your own starting conditions and treat it as the shape rather than the forecast, and take your own baseline before any work starts so that your next reading is compared against you rather than against somebody else.

Why not publish one combined result instead?

Because there is no site the combination describes. The two cases differ in domain age, link profile, window length and the claim each one supports. A single figure across them would have a denominator nobody could state.

Is 92 days long enough to judge an established site?

It is long enough to see whether the work compounded, which is what four consecutive monthly averages answer. It is not long enough to cover a full seasonal cycle, so in a seasonal category a year on year comparison is the stronger reading.

Is nine days long enough to judge a new site?

It is long enough to establish that discovery happened and to date it. It is not long enough to establish a trend, and the published claim says so in those words rather than leaving the reader to infer it.

What should I take a baseline of before starting?

Indexed pages, inbound links, non brand clicks and impressions over a stated window, all dated. 101 of 102 build spec conversations run on ChatGPT and Gemini on 15 October 2026 asked for a before and after with the window stated, and the before only exists if somebody records it first.