Three months in. September is the month the Google side of the work started compounding: fourteen tracked terms improved against six that slipped, the average position of the ranking terms gained eight places, and open technical issues fell from 121 to 72. UK visitors rose by 36% at the same time. The AI side held its ground rather than gaining it, and section 05 explains precisely why.
Month two produced one page-one result. Month three is the month that result stopped looking like luck. Fourteen tracked terms improved, six slipped, and the average position across every term that ranks moved from 52.5 to 44.3, a gain of eight places in a single month. The technical foundation improved sharply as well: open issues almost halved and the link profile grew for the first time since the baseline.
This report tracks two kinds of visibility. Traditional Google search, measured by where the site ranks for the terms buyers type. And AI search, measured by how often assistants name the firm when a buyer asks them who to use. The second is measured across a set of fifteen buyer questions put to ChatGPT and Gemini every cycle, sixty answers in all this month.
Two things to hold in mind this month. First, the AI measurement field is not fixed. The number of brands being compared grew by roughly eighty over the month as the assistants named more firms, bodies and directories across the same fifteen questions. Any percentage calculated against that field will fall when the field grows, even with the firm's own mentions unchanged, so this review leads on how many of the fifteen questions return the firm, which is a stable count, and shows the percentage alongside it. Second, the two guides published in mid-September are too new to judge. Pages typically take four to eight weeks to settle into a position, so they are noted but not counted.
Of the 105 tracked keywords, 14 rose, 6 slipped and 85 held. That ratio is the headline of the month: last month it was ten up against eleven down, which is noise. Better than two improvements for every decline is a trend. The average position across the terms that rank improved from 52.5 to 44.3, and because that average is taken across a near-identical set of terms, the gain is movement rather than arithmetic.
Nineteen terms hold a position, against twenty last month. One term, "employer negligence duty of care mental health", fell out of the measured range from 83. The count is effectively flat while the positions behind it improved, which is the preferable way round.
The best average position of the four months measured, and the first month in which it has improved.
480 searches a month. Holding page one through a second month matters more than reaching it: it is the point at which a position usually stops being provisional.
These are the firms you named, five on the employment side and five on intellectual property and commercial, fixed since August so that each month compares against the same names. Shared keywords counts the search terms where both sites appear in Google, and it is the measure to watch: it rising is ground being taken rather than ground being estimated.
Two points of housekeeping before the table. The comparison returned eight of the ten firms this month; Harper James and Spencer Shaw did not report, and both are being reinstated so the full set is back for Month 4. And the shared-keyword counts moved down on three of the eight while the site's own ranking terms moved up, from 107 to 115. That combination normally means the overlap has shifted rather than shrunk: as the site's terms become more specialist, fewer of them collide with a generalist firm's pages.
| Firm | Shared keywords · Sep | Aug | Move |
|---|---|---|---|
| wilkes.co.uk Employment | 3 | 7 | ▼ 4 |
| freeths.co.uk IP and commercial | 3 | 3 | ◆ held |
| mfgsolicitors.com Employment | 1 | 3 | ▼ 2 |
| schofieldandassociates.co.uk Employment | 1 | 1 | ◆ held |
| hgf.com IP and commercial | none | none | ◆ |
| forresters-ip.com IP and commercial | none | none | ◆ |
| neateandpugh.com Employment | none | none | ◆ |
| codified.legal IP and commercial | none | none | ◆ |
| harperjames.co.uk IP and commercial · not reported this month | pending | 51 | ◆ reinstating |
| spencershaw.co.uk Employment · not reported this month | pending | 2 | ◆ reinstating |
| burleylaw.co.uk (you) | 115 ranking terms | 107 | ▲ 8 |
Fifteen buyer questions went to ChatGPT and Gemini through the month, producing sixty answers. The firm is named on two of the fifteen questions, the same two as last month, first on one and fourth on the other. Where it is named, it is named earlier in the answer than last month: average position improved from 1.60 to 1.43. Every mention, in every answer, was positive in tone.
The reported brand-visibility percentage fell from 10% to 6.7%. That figure is a share of a field, and the field grew. Over the month the assistants named 181 distinct brands across the same fifteen questions, up from roughly a hundred. The firm's own mentions held; the denominator beneath them nearly doubled. This is why the stable measure in this report is the count of questions, not the percentage.
Held, not gained. No tracked question was lost and none was won.
Eighty more brands competing for the same sixty answers. A constant number of mentions divided by a growing field produces a falling percentage without anything having got worse.
Read this chart alongside the two bar charts above it. The line returning to its June level describes the field, not the firm.
Visibility counts how often a firm is named. Position counts where in the answer it appears, and it is the number that decides whether a reader acts on it.
Being named at 1.43 means that when an assistant does recommend the firm, it is almost always the first or second name given. That is the most valuable property the firm currently holds in AI search, and it is the one that improved this month.
The fifteen questions sit in four topics. This is the clearest view of where the firm stands, because it shows coverage and contest side by side rather than averaging them into one figure.
| Topic | Questions | Returning you | Coverage | Your rank |
|---|---|---|---|---|
| Protecting IP for a small business | 4 | 1 | 100% | 1.00 · named first |
| Workplace mental health & duty of care · Birmingham | 4 | 1 | 17% | 4.00 |
| Commercial & tech contracts for software businesses | 3 | 0 | none yet | · |
| Collective redundancy consultation · Birmingham | 4 | 0 | none yet | · |
Every cycle, these fifteen buyer questions go to ChatGPT and Gemini, four answers each. This is the full picture, including the questions where the firm does not yet appear, because that is where the next month's work is decided.
| Tracked question | Topic | You |
|---|---|---|
| Burley Law compared with other UK intellectual property law firms for small businesses | IP · SME | 1st · every answer |
| Which law firm advises employers on mental health duty of care in Birmingham? | Mental health | 4th |
| Best intellectual property protection services for small businesses in the UK | IP · SME | not yet |
| Which intellectual property lawyer should I hire for a small business? | IP · SME | not yet |
| Affordable IP protection options for small businesses near me | IP · SME | not yet |
| Which law firm handles tech contracts for small software businesses in Birmingham? | Tech contracts | not yet |
| Recommended solicitors for drafting SaaS terms and conditions in the UK | Tech contracts | not yet |
| Which UK law firm should a SaaS startup use for its commercial contracts? | Tech contracts | not yet |
| How long does collective redundancy consultation take, and do I need a solicitor? | Redundancy | not yet |
| Best employment solicitors for employers making redundancies in Birmingham | Redundancy | not yet |
| Legal advice for employers on collective redundancy rules in the UK | Redundancy | not yet |
| Which employment law firm should an employer use for collective redundancy consultation in Birmingham? | Redundancy | not yet |
| Can an employer dismiss someone on long-term sick leave with mental health issues in the UK? | Mental health | not yet |
| Is an employer liable for an employee's mental health problems in the UK? | Mental health | not yet |
| Legal advice for an employer managing an employee off sick with stress in the UK | Mental health | not yet |
This is the section that changed most. Open issues fell from 121 to 72, a 40% reduction, and the biggest single contributor was title tags: pages with titles too short dropped from 88 to 34. Of 190 pages crawled, none are broken and none return an error, including the two booking links that were found pointing at retired addresses mid-month and repointed the same week. The on-page score held at 80 ("high").
| Item | Pages | Impact |
|---|---|---|
| Thin / low word count | 29 | High |
| Titles too short | 34 | Medium |
| Titles too long | 5 | Medium |
| Poorly formatted address | 4 | Low |
Thin pages are the only item carrying a high impact rating, and the only one that content work rather than tidying will fix.
| Item | Now | Was |
|---|---|---|
| Pages crawled | 190 | 189 |
| Returned cleanly | 153 | · |
| Redirected | 14 | · |
| Broken or erroring | 0 | 0 |
| Blocked from search | 23 | 24 |
Page headings were also corrected site-wide this month: every page template now carries a single main heading, which is how both Google and the assistants establish what a page is about.
This is the first month in which the visitor figures move in the same direction as the rankings. UK visitors rose to 170 from 125, a gain of 36% in the month legal demand starts to return, and they were the most engaged audience the site has recorded: a 55% engagement rate and 37 seconds a visitor. Estimated organic traffic rose to 161 a month and the site's organic keyword count to 115.
The first clear rise in the UK audience since measurement began, and it came with engagement rising alongside it rather than falling, which is what separates real interest from passing clicks.
Organic search sessions rose from 187 to 193. The AI Assistant line, which records visitors arriving from an assistant rather than a search engine, rose from one session to five. Small, and the first time it has registered as anything other than a rounding error.
The automated traffic fell as the UK audience rose: Singapore from 425 to 251, China from 157 to 68. The UK is now the second largest source of visitors and by a wide margin the most engaged.
| Home | 438 |
| Contact | 41 |
| Understanding Business IP | 40 |
| We are | 40 |
| Employment Law and HR | 19 |
| We think | 18 |
| Commercial Law and IP | 15 |
| IP and Trade Marks | 14 |
| Liz Burley | 14 |
One further page recorded 185 views from nine people during the month. That is site work rather than readers, and it is left out of this table so it does not distort the picture.
Legal search softens over the summer and builds from September through to the winter peak. The rankings in section 03 improved in the first month of that build, on pages published during the quiet months, and the UK audience rose with them. That is the sequence the plan was arranged around.
| Metric | Baseline · 30 Jun | Month 1 · 30 Jul | Month 2 · 28 Aug | Month 3 · 30 Sep | Move |
|---|---|---|---|---|---|
| Best non-brand position | 14 | 18 | 5 | 4 | ▲ |
| Average position (ranking terms) | 45.9 | 51.8 | 52.5 | 44.3 | ▲ |
| Keywords on page one (non-brand) | 0 | 0 | 1 | 1 | ◆ held |
| Tracked terms ranking | 17 | 18 | 20 | 19 | ◆ |
| Tracked terms improved / slipped | · | · | 10 / 11 | 14 / 6 | ▲ |
| AI questions naming the firm | 1 of 15 | 1 of 15 | 2 of 15 | 2 of 15 | ◆ held |
| Average position when named in AI | 1.0 | 1.0 | 1.6 | 1.43 | ▲ |
| Sentiment of AI mentions | · | · | · | 100% positive | ▲ |
| AI brand visibility share of a growing field | 6.7% | 9% | 10% | 6.7% | ◆ see 05 |
| AI share of voice share of a growing field | 1.4% | 1.6% | 2.3% | 2.0% | ◆ see 05 |
| Brands in the AI field | · | · | ~100 | 181 | · |
| On-page SEO score | ~77 | 80 | 80 | 80 | ◆ high |
| SEO issues open | 133 | 121 | 121 | 72 | ▲ |
| Pages with titles too short | · | · | 88 | 34 | ▲ |
| Thin pages | · | · | 28 | 29 | · |
| Broken pages | · | · | 0 | 0 | ◆ none |
| Est. organic traffic | ~163 | 146 | 157 | 161 | ▲ |
| Organic keywords | · | · | 107 | 115 | ▲ |
| Backlinks | 89 | 89 | 89 | 99 | ▲ |
| UK visitors (Google Analytics) | 145 | 126 | 125 | 170 | ▲ |
| Organic search sessions | · | 192 | 187 | 193 | ▲ |
| Organic engagement time | · | 31s | 27s | 27s | ◆ held |
| UK engagement rate | · | · | · | 55% | ▲ |
| Sessions from AI assistants | · | · | 1 | 5 | ▲ |
| Desktop load time | 1.49s | 0.64s | 1.09s | 1.09s | within target |
Earlier reviews · Month One Baseline · 30 June 2026 · Month 2 · 28 August 2026 · Month 3 · 1 October 2026