The diagnostic

A written read of the Search Console export you send, and only of what it contains: how often its pages and searches are clicked, which pages answer in a single line, and which patterns the export supports looking at first.

Price

On request, paid up front, tax inclusive

Turnaround

5 working days

You send

One export from Search Console

You get

One written report

Calls

None. Everything in writing.

Refund

Full, on request, and you keep the report

What arrives

One document, five sections, every figure carrying the cut it came from. There is no dashboard to log into and nothing is held back for a follow-up conversation.

Rather than describe it: read a complete sample report, run on a property we operate. Every figure in it is computed from that export rather than written into the page, and checked against findings already published from the same window. The export itself is not published, so that is a check we run rather than one you can.

1. 01

   Pages that answer in one line

   Some pages answer with a single value: a code, a number, a date. This section sorts every page in your export by what each one answers, before any click figure is looked at, then sets how often one group is clicked against how often the other is. Both sides of the comparison come from the same rows of your file. It reports the difference between the two groups and does not say what produced it.

   Related finding: Three clusters that rarely got a click

2. 02

   Seen often, clicked rarely

   The pages and the searches that come up most often and get clicked least, at similar average positions. The two lists are read separately, because your file cannot say which page came up for which search. It shows where being shown is not turning into clicks. It does not say why.

   Related finding: Different click rates within one band

3. 03

   Common searches against rare ones

   Your busiest searches set against your rarest ones, within each group of related searches, and only where both come up at similar average positions, with each group's click rate beside the other's. Where your file does not return the rare ones, the comparison is left out rather than estimated. Which way the difference runs is a result about your own file and this report does not generalise it.

   Related finding: Long tail advantage, and its limits

4. 04

   Search appearance breakdown

   Which kinds of search result your impressions and clicks came from, read on their own because one impression can count under more than one kind. On a property we operate, 28 days of this returned 184,133 impressions from translated results, which we had not examined separately before.

   Related finding: Impressions from translated results

5. 05

   What the exports do not single out

   Said plainly, in its own section. Things in your file that look like problems but that the data does not actually single out, and questions your file cannot answer at all, such as what a change would cost or what a click is worth. Each one says what it would take to go further.

The export

One file, out of Search Console, which you already have. It takes a few minutes and needs nothing installed. We send these instructions again after you get in touch, so there is nothing to memorise here.

• One export

  Performance > Search results > Export > Download CSV

  Last 3 months. The export button offers more than one format and only the CSV download can be read here: it arrives as a zip of the sheets, and the thing to send is that zip as it downloaded. Do not remove any sheet from it, and do not open and resave one, since a file that has been through a spreadsheet application usually loses the number formatting the parser needs.

What is in it, and what each sheet is for

• Queries

  The searches people used, grouped, and how often each was clicked. It holds at most 1,000 rows and leaves out searches Google withholds for privacy, so on a larger site this is part of your searches, not all of them.

• Pages

  The page classification and page-level shares, read separately from the queries: the two sheets cannot say which page appeared for which query. It also stops at 1,000 rows, so on a larger site the shares describe the pages in the file, not every page.

• Search appearance

  Which search appearance types your impressions and clicks came from, read on their own and never added to the page figures.

• Dates

  The daily chart. The window the report covers is taken from these labels rather than from what anyone says it was, and it is stated as a number of days only where the labels are daily, in order and unbroken.

• Filters

  Which filters were on when the export was taken. The appendix quotes them as the interface wrote them, because a figure read under a filter nobody wrote down does not say what it covers.

• Countries

  Two things. The split across markets, which shows whether the totals are concentrated in one of them and so qualifies how broadly the report can be read: a rate is still a rate over every impression in the file, wherever those impressions came from. And a total, which agrees with the daily chart when those two sheets describe the same window: that is evidence about those two sheets and about nothing else in the file.

• Devices

  That same total, against the same chart, and nothing else. No finding in the report is read from the device split.

Who this is not for

Saying this plainly is cheaper for both of us than a refund.

• Sites with very little search data yet. This report compares groups of pages and queries; sparse data may leave those comparisons too uncertain to guide a decision.
• Anyone who wants a call first. That is a reasonable thing to want and it is not what this is. The writing is the product.
• Anyone looking for a guarantee that rankings will rise. The report says how often your pages and queries are shown and clicked, and where they are rarely clicked. It does not measure costs or revenue.

Questions