Faragiri

Learn Content quality On page

A Peak District locksmith audited its own guide against the six things a reader arrives needing, and found four of them missing and two nobody had ever checked.

4Sections
4Min read
4Depth

The working test is whether a reader goes back to the results afterwards. Longnor Locks, a locksmith in the Peak District, had a 1,900 word guide on door lock security sitting at position 14 for two years, and the audit that fixed it was six rows long and took an afternoon.

The six things a reader needed, and four of them were missing. Somebody who has just been burgled, or has just failed a home insurance question, arrives wanting: what the lock on their door actually is, whether it meets the standard the insurer names, what a replacement costs, how long it takes, whether they can do it, and who to ring at nine at night. The guide answered the first and the fifth.

What was added, and what it did

The repair was structural before it was factual: answer the question in the first two sentences, then earn the rest, which meant the four hundred words of introduction went and the price table came up the page. Whether an assembled answer at the top of the results has already ended the search before any of this is read is being cited when the answer is generated.

The repair

What moved Longnor's lock guide from position 14 to 6

Four hundred words of introduction left the page, and two concrete facts replaced them near the top.

Position 14Two years stuck behind a four-hundred-word preamble.
  1. 01Answer in the first two sentencesRemoved the long intro before stating the core point.
  2. 02Move the price table upPlaced repair figures where readers find them immediately.
  3. 03Name British Standard 3621Matches what a home insurer's question is actually checking.
  4. 04Quote 85 pounds and 40 minutesLocal Buxton cylinder pricing and labour a summary cannot invent.
Position 6Reached over four months after structural and factual repair.

An assembled answer can define a standard, but only a working locksmith can publish local 2026 rates and common town cylinders.

Two facts did most of the work. British Standard 3621, and what a euro cylinder costs to change in 2026, which at Longnor is 85 pounds for the cylinder and about 40 minutes of labour. Neither had been on the page, both are what an insurer's question is actually about, and the page went from 14 to 6 over four months.

A number, a date and a name are what a page has that a generated answer does not. An assembled answer can define a British Standard perfectly. It cannot say what a locksmith in Buxton charged in July 2026, and it cannot say that the two most common cylinders on the doors in the town are a particular pair, which is the kind of thing only somebody who has changed four hundred of them knows.

Length, and the thing it is mistaken for

Length is a consequence of answering, never a target. The guide was 1,900 words before the audit and 2,150 after, so almost nothing was added by volume; four hundred words of introduction came out and six hundred and fifty words of answers went in.

Word count

Length came from answering, not padding

The guide moved from 1,900 words to 2,150 by trading repeated filler for specific local proof.

  • Removed as duplicate fillerCost nothing to delete because top results already held it.
    • Four hundred words of opening preamble
    • Lock type descriptions repeated on top three pages
    • Four purchased stock photographs
  • Added as earned answersSix hundred and fifty words built around real jobs.
    • Direct answers to missing reader questions
    • Three phone photos taken in a Longnor hallway
    • The exact cylinder shown in its own door

Longnor Locks door guide

Total length barely changed, because replacing generic paragraphs with job-specific evidence keeps word count disciplined.

A page that repeats what the top three say has no reason to be preferred. The old guide's whole middle section was a description of lock types that appeared, in different words, on every page ranking above it. Removing it cost nothing, because a reader who wanted that had already read it somewhere else, and an engine holding three copies of it has no reason to want a fourth.

One photograph of the actual job beats four stock images. Longnor replaced four purchased photographs with three taken on a phone in a hallway in Longnor, one of which shows the exact cylinder the page is about, sitting in the door it came out of.

Where this stalls: auditing what a machine can count

Every other item in this cluster can be counted. Title length, heading order, alt text, valid markup, all of them appear on a report as a number that can be driven to zero. This one cannot, and so it is the item that gets postponed while the countable ones are perfected.

The audit

Six things a reader arrives needing, marked against the page as it was

Four were absent, one was answered, and two rows are marked unknown because nobody had ever asked a reader which they wanted.

The door lock guide, February 2026

  1. YesWhat the lock on their door actually isAnswered, with photographs, and it was the only strong section.Yes
  2. NoWhether it meets the standard an insurer namesBritish Standard 3621 was not on the page at all.No
  3. NoWhat a replacement costs85 pounds for a cylinder and about 40 minutes, added in March.No
  4. YesHow long it takesOne line, and it was accurate.Yes
  5. Not checkedWhether a reader could do it themselvesNobody has asked whether readers want to be told they can.Not checked
  6. Not checkedWhether price matters more than timing at nine at nightThe phone would know and nobody has kept a note.Not checked

Two unknown rows on a six row audit is a good result. An audit with no unknown rows has usually been filled in by somebody guessing on behalf of the reader.

The audit above is the substitute and it is not automatable: six rows written by somebody who has answered the phone at nine at night, marked against the page as it stands. The unknown rows in an audit are the honest ones, and Longnor had two of them: nobody knew whether readers wanted the standard explained or just named, and nobody had ever asked whether the price mattered more than the timing.

What can be read afterwards

Search Console does not report quality and it reports two things that stand in for it. The number of distinct queries a page collects, which rose from 60 to 174 at Longnor, and the position for queries the page was never written for. Both are free, and it replaced the keyword data Google removed from analytics in 2011.

Afterwards

What Search Console and Clarity showed after the rewrite

Free tools cannot score quality directly, so they track query breadth and whether readers reach the answer.

Longnor Locks measurements before and after moving answers up the page

Before repairAfter repair

  1. Distinct queries collected60174Search Console query count rose across terms the old page missed.
  2. Readers reaching price table (per 100)2268Microsoft Clarity scroll depth in 2026 after moving pricing higher.

Both metrics come from free reporting that replaced older paid session tools and removed keyword analytics.

Microsoft Clarity shows how far down people read, free with no cap on sessions, and it replaced paying by the session for recordings. As of 2026 the reading to trust from it is not a heat map but the scroll depth on a page where the answer is halfway down: at Longnor, 68 in a hundred reached the price table after it moved up, against 22 before.

The claims a reader skims before deciding to read are an outline a reader and a crawler both use, and the promise that got them here at all is the most valuable line on the page. Both are inside what a single page controls about its own ranking, and this is the one item that decides whether the other six were worth doing.

Alongside this

Above this