Skip to content
allanninal.dev/build
Collaborate

Part 5 of 7 · Search rank reporter series ~5 min read

How the ranking report reads

A weekly report that shows different things each week cannot be skimmed, and a report that cannot be skimmed does not get read for long. This one shows the same six lines in the same order, forever, and that consistency is most of why it survives.

Key takeaways

  • The same six themes, in the same order, every week.
  • Each line: position, clicks, impressions, and the direction against last year.
  • The page carrying each theme is named, with what share of it lands there.
  • Correlations are stated as correlations. The report never claims a cause.
  • One monthly addition: the top twenty unmatched queries.

The weekly page

Week ending 26 July

  • Emergency callouts — pos 7.8 (was 4.1), 340 clicks (−60%), 4,100 impressions. 92% lands on /emergency-plumber.
  • Boiler installation — pos 6.2 (steady), 210 clicks (−4%), 2,900 impressions. Year on year: +8%.
  • Bathroom fitting — pos 11.4 (was 12.1), 88 clicks (+12%), 1,600 impressions.
  • Commercial — pos 9.0 (steady), 41 clicks, 640 impressions. Below the volume floor for movement.
  • Brand — pos 1.1, 520 clicks, 700 impressions. Steady.
  • Unmatched — 18% of impressions. Top queries listed monthly.

Six lines, the same six, in the same order. The first one is the finding this week and it is obvious at a glance precisely because the other five look like they always do. A report that only showed movement would have shown one line, and a reader would have no idea whether the other five were fine or missing.

Naming the page

The page carrying a theme is what somebody can actually change, and the share matters as much as the name. “92% lands on /emergency-plumber” says the theme has one page and the finding is about that page. A theme spread across nine pages at eleven per cent each is a different situation entirely and the report says that too.

How a search theme's movement is attributed to pagesA vertical chain of five steps entered by a box labelled A theme with movement, confirmed. Step one asks which pages carry it, using the by-page fetch and clicks per page. Step two asks whether it is concentrated, meaning one page carries over seventy per cent; if so it exits to Name that page, since the finding is about it. Step three asks whether it is spread, with no page over thirty per cent; if so it exits to Say it is spread, indicating a site-wide movement. Step four asks whether a page changed, using the deploy or CMS feed. Step five states the correlation and never the cause. A note says a concentrated theme and a spread one need completely different responses.AWS ACCOUNTA theme with movementconfirmedWhich pages carry it?from the by-page fetchPage dataclicks per pageConcentrated?one page over 70%Name that pagethe finding is about ityesSpread?no page over 30%Say it is spreada site-wide movementyesDid a page change?from the deploy or CMS feedState the correlationnever the causeA concentrated theme and a spread one need completely different responses.
Fig 1. How a theme’s movement is attributed to pages. Concentration is the first question, because it determines whether this is a page problem or a site problem.
  • Database
  • App integration
  • Analytics
  • Front-end & mobile

What it refuses to claim

Correlation, stated as correlation

  • It says: “/emergency-plumber was edited on the 21st. The position change first appears in the week beginning the 20th.”
  • It does not say: “the edit on the 21st caused the drop.”
  • Why: search rankings move for reasons entirely invisible from your side — a competitor’s change, an algorithm update, a new result format. Claiming causation from a timestamp is how businesses spend a fortnight reverting an edit that was fine.
  • What it will say: “nothing changed on this page in the last 60 days”, which is genuinely informative and rules something out.

That restraint is not pedantry. A confident wrong cause is worse than no cause at all, because somebody acts on it. Presenting a timing correlation as a timing correlation lets a person weigh it against everything else they know, which is what they are for.

The monthly addition

The three search reporting cadences and what each addsA horizontal row of five boxes. Weekly: six themes. Monthly: plus the top twenty unmatched queries. Quarterly: plus year-on-year comparison. Never: a query-level export. Available: as a link, always. A note says the raw data is one link away and is never the message.THREE CADENCES, AND ONE THING THAT NEVER APPEARSWeeklysix themesMonthlyplus unmatched top 20Quarterlyplus year-on-yearNevera query-level exportAvailableas a link, alwaysThe raw data is one link away and is never the message.
Fig 2. The three reporting cadences. The raw query-level data remains available and deliberately never arrives in an inbox.
  • Networking
  • Security & identity
  • Analytics

The unmatched top twenty is a monthly rather than weekly item because it changes slowly and reading it takes five minutes of genuine attention rather than nine seconds of skimming. Putting it in the weekly report would mean it was skimmed, which would waste the most valuable thing in the system.

Next: what all of this costs to run.

All posts