What having history changes about every number on a growth chart

A growth figure is a statement about a window, not about a method, and the first thing a window inherits is whatever the site already was on the day it opened. The clawlaw.in export on our proof page is the clear case. It covers 2026-07-13 to 2026-10-12, 92 days with data, and it was taken on 2026-10-14. Across that window the site went from 59 clicks a day to 349, with 18,537 clicks and 929,405 impressions in total. And the site had history and existing pages before the window opened, which means every one of those figures describes acceleration on something already moving. That is a real result and it answers a narrower question than most people read it as answering.

What the starting conditions actually were

clawlaw.in was an established legal research product when the window opened. It had pages in the index, it had inbound links pointing at them, and it had age. None of that is a footnote to the numbers. It is the reason the numbers look the way they do.

When a site already has standing, new work lands on top of a base. A page published on that domain gets crawled sooner, gets internal links from pages that already carry weight, and gets judged by a system that has seen the domain before. The same page on a domain with none of those things has to earn all of it first.

The monthly shape, and the one caveat it carries

Here are the daily averages from the same export, which is the form that shows the shape rather than the total.

Month	Clicks a day	Impressions a day	Days in the window
July 2026	59	2,253	19
August 2026	154	7,276	31
September 2026	282	14,703	30
October 2026	349	18,331	12

Note the day counts. July contributes 19 days and October contributes 12, because the window starts mid July and ends on 12 October. A daily average handles that correctly and a monthly total would not, which is the reason the daily figure is the one published.

The four questions that recover the starting conditions

When somebody shows you a growth chart, these four questions get you to the thing the chart is silent about. They take a minute and they are not hostile, because any team that measured properly has the answers already.

• How old was the site on the first day of the window, and how many pages were already indexed?
• What was the first day's figure in absolute terms? A multiple hides its base, and the base is what decides whether the multiple was arithmetically easy or hard.
• Was anything else running in the same period? A product launch, press coverage, paid media or a seasonal peak all sit happily inside an organic chart.
• What share of the clicks were brand searches? Brand demand follows advertising and flatters the result, so the non brand figure is the one the search work earned.

The reason to ask all four is that a chart can be completely accurate and still predict nothing about your own site. A site recovering from a technical problem produces a spectacular curve that says very little about steady growth work. A site with ten years of links produces a curve a two month old domain cannot reproduce at any budget.

What a window on an established site can and cannot settle

It can settle whether the work compounded. Four consecutive monthly averages that each beat the one before, from 59 to 154 to 282 to 349, is a shape that one lucky week cannot produce, and that shape is the actual claim in the clawlaw.in case.

It cannot settle how long a different site would take to reach the same place, because the window never contained that part. The export starts at 59 clicks a day, so everything that got the site to 59 happened before the first row and is not in the file. A reader looking for the time from zero is reading the wrong document.

This is also why the pace question is worth separating from the growth question. How quickly the first measurable movement arrives depends on domain standing, competition and how much is broken at the start, and the stage by stage version of how long search work takes to show results is the better reference for that than any single curve.

How to state your own starting conditions so the number survives

If you publish growth figures, three lines above the chart make the whole thing defensible. The window with both dates and the day count. The export date. And one sentence saying what the site was on day one: its age, roughly how many pages were indexed, and whether it had links.

That third line is the one almost nobody writes, and it is the one that makes the first two useful. Without it a reader has to guess, and a careful reader guesses the least flattering option.

The same logic decides what a client should expect from a new engagement. A ninety day window on an established site is measured on whether each month beat the last, and what the first ninety days should actually produce is worth agreeing in writing before the window opens rather than after the chart arrives.

Disclosure, because it belongs on the page and not in a footer: we sell the measurement this argument comes from, and we publish our own starting conditions for the same reason we ask other people for theirs.

Questions about growth figures on a site that already had history

Is a result on an established domain less impressive than one on a new domain?

It is a different result rather than a smaller one. Compounding on a working site is the harder thing to sustain over three months and the easier thing to start. The two cases answer different questions and each is worth publishing on its own terms.

Why publish daily averages instead of monthly totals?

Because the first and last months of a window are usually partial. July holds 19 days of this window and October holds 12, so monthly totals would show a dip in October that is purely a counting artefact. A daily average removes it without hiding anything.

How much of the clawlaw.in result came from more impressions rather than better position?

Impressions a day went from 2,253 to 18,331 across the same window, so the site was being shown far more often by the end of it. That is the figure worth reading beside the click line, and it is published on the proof page for exactly that reason.

What should I ask for if the chart has no axis values?

The underlying export. A performance export is a file, not a judgement, and a team that produced the chart from one can send it. 78 of 102 build spec conversations on 15 October 2026 asked for the question set behind a measurement to be published so the work could be re-run, and the same instinct applies to a search export.

Does the same window logic apply to whether an assistant names a business?

It applies harder, because there is no console behind an assistant's answer. A reading taken on one date is a reading about that date, so the date and the apps by name have to travel with the figure. That is the measurement we built and the gap we sell against, and the dates are on every number we publish.