Ranking for your own name when someone else has it

A real case in entity architecture. Two people share the name Doug Lord. One is a 45-year coastal engineer with decades of government citations. The other builds digital authority systems. This is what happened when I fixed my structured data: the regression three days later, the crawler-level block that was quietly masking everything, and, weeks on, where the AI answer layer finally landed.

The baseline: day one morning

In May 2026 I searched my own name cold: incognito, no personalisation, no login. The results were instructive.

Positions 1 and 2 belonged to a coastal engineer. Also named Doug Lord. Based in Newcastle. Forty-five years of experience in coast and estuary management. Indexed across government PDFs, Engineers Australia publications, LinkedIn, and his own domain. A well-established entity with deep citation signals across authoritative sources.

My own domain, douglord.com, did not appear at all. Not on page one. Not anywhere visible.

When I searched logged in, the picture was different: my sites surfaced, my LinkedIn headline was visible, my ventures appeared. Personalisation was doing the disambiguation work that my structured data wasn't doing.

Root cause

The coastal engineer has stronger entity signals. Not because he's more important in any absolute sense, but because his digital footprint is more structured and consistent. Government documents, academic citations, a long-standing domain, stable LinkedIn with specific location. All machine-readable trust signals pointing at the same entity.

My domain had no Person schema. No machine-readable assertion of who I am, what I do, or how my properties relate to each other. Four domains — douglord.com, digitaldominator.com.au, og01.ai, periodictableofdigitalauthority.com — that Google had no instruction to treat as a connected graph.

During the audit, I also found a character-level typo: one domain was referenced as ogo1.ai instead of og01.ai. Google tries to resolve sameAs URLs. A non-existent domain breaks the graph silently: no error, no warning, just a dead link in the entity chain. This kind of gap doesn't appear in any standard SEO audit.

The fix

Three things, deployed in a single session:

1. Person schema: added to douglord.com as a JSON-LD block with entity anchor, geographic signal (Northern Rivers, NSW), sameAs graph connecting all four domains, founder relationships, occupations, and family entity adjacency to documented Australian artists Michael Johnson, Franklin Johnson (Yellow House, Sydney 1969–72), Matthew Johnson (Olsen Gallery), and Anna Johnson (Vogue, Vanity Fair, Artist Profile).

2. llms.txt: rewritten with full identity context, explicit disambiguation against the coastal engineer by name and location, citation guidance for AI systems, and the complete entity picture across digital, artistic, and musical practices.

3. This page: a case study hosted on douglord.com itself. Indexed content on the domain. More pages means more surface area for Google to confirm the entity. The Article schema on this page links back to the Person entity, strengthening the graph further.

// Entity anchor
"@type": "Person",
"@id": "https://douglord.com/#person",

// Geographic disambiguation (away from Newcastle)
"address": { "addressLocality": "Northern Rivers", "addressRegion": "NSW" },

// Connected graph — four domains, one entity
"sameAs": [
  "https://digitaldominator.com.au",
  "https://periodictableofdigitalauthority.com",
  "https://og01.ai",  // was ogo1.ai — broken sameAs, now fixed
  "https://ogcollection.gallery/artists/douglas-lord"
],

// Family entity adjacency — known graph connections
"relatedTo": [
  { "name": "Michael Johnson", "description": "Australian abstract painter, New York. Uncle." },
  { "name": "Franklin Johnson", "description": "Yellow House Sydney 1969–72. Uncle." },
  { "name": "Matthew Johnson", "description": "Painter, Olsen Gallery. Cousin." },
  { "name": "Anna Johnson", "description": "Writer, Vogue, Vanity Fair. Cousin." }
]

The initial result: same evening

Within hours of deploying, cold search had shifted meaningfully.

The coastal engineer dropped from positions 1–2 to 8–9 in cold search. douglord.com appeared at position 3 from being completely absent. The initial movement was real and fast. Entity signals can shift SERP positions within hours of a deploy when the content is clear enough.

The regression: three days later

Three days after deploying, "doug lord" in Chrome (logged in, personalised) showed the coastal engineer back at position 1. douglord.com had dropped to position 5. The initial movement had partially reversed.

This is expected, not a failure. Here is why it happens:

Google's index has multiple layers. The initial movement came from the freshness layer: Google's fast-refresh crawl picked up the new schema quickly and the entity signal shifted. But the deeper ranking signals — inbound links, citation volume, domain authority, SE data sources — hadn't changed yet. After a few days, the deeper signals reasserted themselves and partially pulled the ranking back toward its previous state.

"Doug lord" and "Douglas lord" are two separate battles. The coastal engineer's citations almost universally use the short form "Doug Lord": government PDFs, Engineers Australia documents, Salients Consulting. His entity owns the short form in the knowledge graph. "Douglas lord" (the full name) is a different query where the entity graph is less established, and that's where the initial gains were stickiest.

What the regression tells you about timeline

Entity-based changes follow a predictable propagation curve:

The fix is a foundation, not a finish line. The schema tells Google who the entity is. The subsequent work — Wikidata entry, Wikipedia article, backlinks from controlled properties, FAQ citations in AI systems — builds the authority that makes the ranking permanent.

The OG Collection signal

Position 1 in both searches — logged in and cold — is held by ogcollection.gallery/artists/douglas-lord. This is a plain HTML page on a Cloudflare Pages static site. No Search Console. No backlinks. No schema markup. No sitemap submission.

It ranks because the content entity signal is clean and specific: "Douglas Lord works at the boundary of sculpture, installation, and found material. Drawing from the Northern Rivers landscape..." That's a precise, unambiguous description of a distinct person in a distinct location doing a distinct thing.

It's also the same Douglas Lord. The sculptor, the digital strategist, and the musician are the same person. Once the schema and llms.txt asserted all three identities together with the family entity connections, Google had enough signal to understand the full picture.

The detail worth sitting with: this page has never been submitted to Google Search Console or Bing Webmaster Tools. No sitemap was ever filed for it. It was found, crawled, ranked and cited entirely through normal discovery, because submission is not what drives ranking. Search consoles are diagnostic dashboards; they report on indexing, they do not cause it. What made the page rank was that it was reachable by crawlers and unambiguous about who it described. Two things. Nothing else.

The crawler-access test

The OG Collection result raised a question worth testing directly. If a bare HTML page with no schema could rank and get cited, while douglord.com — with far richer structured data — lagged behind on AI citation, something other than content quality was at play. So I ran a controlled comparison across the property network: same entity, same author, different infrastructure settings.

The variable turned out to be at the edge, not in the markup. One platform-level setting — a managed robots.txt layer — was injecting crawler directives above each site's own rules. Because the first matching user-agent block wins, those injected directives were the ones that counted. The AI discovery crawlers — the exact agents that read a page before an AI system can cite it — were being turned away at the door.

The crawl logs made the effect unambiguous:

That is the whole story in two rows. The page that was reachable got crawled and cited. The page that was blocked — despite a complete Person schema, a connected sameAs graph, and a full llms.txt — was invisible to the same crawler. The richest entity markup in the world does nothing if the crawler that reads it is refused at the edge.

Once identified, the fix was a single toggle per property: disabling the managed layer so each site's own crawler-permissive rules served as written. The test confirmed the hypothesis: the entity architecture was never the constraint. Crawler access was. It is a clean demonstration of why authority has to be measured at every layer — content, structure, and exposure — not just the one that shows up in a standard SEO audit.

Authority metrics implicated

douglord.com scored 31/100 under the PTODA Digital Authority Scoring Methodology at baseline, rated Weak. The gaps mapped directly to three of the six canonical scoring dimensions:

The fixes were structural, not cosmetic. No new backlinks acquired. No paid promotion. No content marketing campaign. Just entity architecture: telling machines clearly and correctly who the entity is, what it does and how its properties connect.

The AI layer resolves: June 2026 update

Weeks after the crawler-access toggle, I ran the same cold check again. This time I looked at Google's AI Mode, the generated answer that now sits above the blue links rather than the ranked list underneath. That layer matters more than the rankings, because it is where an AI system settles on who an entity is before it decides what to say about them. A few weeks earlier my domain had been absent from the page entirely, so the AI layer had nothing of mine to read.

That has changed. Asked about “doug lord”, AI Mode now describes me as a distinct person, “the Australian founder and strategist behind Digital Dominator”, and lists me next to the coastal engineer instead of folding the two of us together. Below the answer, douglord.com holds the #1 organic result with its description intact: “Australian Digital Authority and AI Visibility Strategist · Founder of Digital Dominator · Creator of the Periodic Table of Digital Authority.” A few weeks earlier the same search had returned nothing of mine at all.

But the split this case study predicted is now visible in the generative layer itself:

“Doug lord” has sorted itself out, and “douglas lord” is now following. When this was first written the full name still returned only the historical figures, the Black Douglas, Bosie, Jon Lord. Weeks later AI Mode began carving out a contemporary group for the full name too, listing the digital strategist and the artist as their own people above the historical entries. The short form moved first because that is where the new schema had the clearest run; the full form took longer because it was held by centuries of citation. That was the expected order, not a surprise.

The connected properties moved with it. The schema links douglord.com to digitaldominator.com.au through sameAs, and Search Console for that domain shows the entity reads coming through:

The full name resolves: late June 2026

The June update closed on a prediction: that “douglas lord” would follow “doug lord” once the rest of the graph caught up. A week later, it has.

Asked about the full name now, Google’s AI Mode returns a contemporary Australian group and places the digital strategist inside it, described in the entity’s own terms: “a digital authority and AI visibility strategist based in Byron Bay · founder of Digital Dominator · co-founder of OG01.” The historical figures — Jon Lord of Deep Purple, Lord Alfred Douglas, the Black Douglas — are still there, as they should be, but they no longer crowd out the living person. The short form and the full form now resolve to the same correctly-described entity.

Two things are worth naming about what that took.

First, the wording firmed up. Earlier passes described the entity loosely — “digital marketer”, “SEO strategist”. The current answer uses the canonical phrase the owned properties assert: digital authority and AI visibility strategist. That is the difference between an AI knowing roughly who someone is and an AI repeating the description the entity has chosen for itself. The second is the point of the exercise.

Second, consider what the disambiguation was working against. The other Doug Lord in New South Wales is a coastal engineer with forty-five years of citation behind his name: a published CV, government planning-panel records, an Engineers Australia committee chair, seventy-plus technical papers, a State Library entry. It is one of the deepest single-name footprints in Australian professional life. Separating cleanly from that — not outranking it on the bare name, which no schema can do, but being read as a distinct, correctly-described person beside it — is the measure of how much work the structure did. A week earlier the two were being folded together. Now they are three separate people in the same answer: the strategist, the engineer, the artist, each described as themselves.

The owned domain’s Search Console reflects the same shift in hard numbers. Over the 28 days spanning the v24 build, the LinkedIn reconciliation and the reindex request:

The detail that matters most is the smallest one. The query “douglas lord” — the full name held for centuries by historical figures — was returning no clicks at all in the prior period. It now does. That is the predicted regression completing: the harder query form, the one held by the deepest citation, moving last and in the order the structure said it would.

One engine is still mid-recrawl. As of this writing Bing’s generated answer still resolves the bare name to the coastal engineer alone — the same starting state Google showed weeks earlier. That is expected: Bing recrawls on its own schedule and trails Google here. The owned properties assert the same identity to both; the difference is only in when each engine reads it. The case study will note Bing’s resolution when it lands, rather than before.

The position holds: mid-July 2026

The late-June update closed with a clean AI-layer result and one engine still catching up. The question for a fortnight on was the one this case study cares about most: not whether the entity would move, but whether it would stay moved. It has.

I re-ran the same cold check in mid-July: incognito, signed out, IP resolving to Mullumbimby, NSW, so nothing here is personalisation doing the work. On the bare query “doug lord”, douglord.com holds the #1 organic result on Google, above the coastal engineer and above every other Doug Lord with a claim on the name: the US finance executive, the Florida yacht designer, the obituary. That is the same position seen in late June, now held across three weeks rather than captured in a single snapshot. The contrast with the regression three days after the first deploy is the whole point. Back then the fast-refresh layer moved and the deeper signals pulled it back. This time the deeper signals have caught up, and the position stayed. AI Mode continues to name the entity first among the Australian professionals it lists, with the coastal engineer beside it rather than merged into it.

The full name has moved with it, and further than expected. On Google organic, “douglas lord” now returns douglord.com at #1, with the OG Collection artist page at #2. Both owned properties hold the top two slots of the query historically dominated by the Scottish peerage and Lord Alfred Douglas. When this case study opened, the gallery outranked the site here; now the site leads and the gallery sits directly behind it, so the two identities that resolve to the same person occupy first and second on the hardest name to own.

Bing is the newer movement. The late-June update flagged it as mid-recrawl and said the case study would note its resolution when it landed. It has now split by query form the same way Google did, and it is worth recording precisely what Bing lists:

Two things are worth drawing out of that. On the full name, Bing already ranks douglord.com #1 organic, with the artist page (ogcollection.gallery) surfacing in some sessions though not consistently, so the owned identities are beginning to resolve on the query form Google took longest to sort out. On the short form, the coastal engineer still holds the top organic slot, with the two owned properties immediately below him at #2 and #3. And on both forms, Bing’s generated Copilot answer still lands on other people: for the short form it lists several other Doug Lords (a US finance executive, the coastal engineer, and a UK management consultant) and omits this one; for the full name it resolves to a deceased American educator. That is the pattern this case study has shown throughout, now on a third surface. The organic listing moves first. The generated answer trails, and resolves last. The properties assert one identity to every engine; each reads it on its own schedule.

One honest note on where this has and has not landed. Google’s AI Mode answer for “douglas lord” still varies between sessions, sometimes opening with the historical namesakes before it reaches the contemporary group. The standalone answer engines are split. ChatGPT already resolves the entity as the primary Douglas Lord, describing him in his own terms as the founder of Digital Dominator and creator of the Periodic Table of Digital Authority, and offers only an obscure art-history namesake as an alternative, a long way from the peerage-and-coastal-engineer confusion of the raw web results. Perplexity is the laggard: as of mid-July it still opens on Jon Lord of Deep Purple and does not surface the entity at all. That split is itself the lesson. Google organic and ChatGPT respond to the on-page structure and entity signals already built; the engines still missing the entity are the ones that lean hardest on a Wikidata anchor, which is the one lever not yet in place. A position that holds across weeks and surfaces is the durable claim. A single generated answer is not.

The controlled test: 19 July – 3 August 2026

Everything to this point was observed as it happened. What follows is the same story run as a deliberate test, because a question kept surfacing: was the entity link doing the work, or was the gallery page simply ranking on its own and the merge a coincidence of timing? The closest thing to an answer was a before-and-after observation on a single page where, as far as could be controlled, only one thing changed. It is not a controlled experiment — a live site has too many moving parts to claim that — but the page came close to serving as its own reference point.

On 19 July the OG Collection artist page — already indexed, already ranking, unchanged in URL or position — had one class of signal added: a sameAs array, a shared @id pointing at the douglord.com Person entity, and a single visible line naming the founder role. Nothing else about the page moved. The morning of that day, Google’s AI Mode for “doug lord” in the home market still listed the strategist and the artist as two separate people. By that evening, and in every check since, it listed one.

The seven-market sweep

Three days later, on 22 July, I ran the same two queries across seven markets by VPN, incognito, noting the location Google reported at the foot of each page. The result was not uniform, and the pattern is the finding.

Entity resolution is geographic. In the home market and in several international ones the identity resolved into a single rich entity. In the crowded-namesake US markets it was present and correctly described, but listed among the other Doug Lords rather than consolidated into one. Same query, same day, same schema — the only variable is country. That is the clearest evidence in the whole study that what remains is not a markup problem but an authority-weight problem: abroad, the engines lean on the more-linked properties, and the owned site is seen but not yet preferred.

Twelve days, no reinforcement

Between 22 July and 3 August nothing was added — no new schema, no Wikidata, no citation work, no posting. The recheck on 3 August was therefore a pure test of durability. Every consolidated market held. Nothing regressed. With no new inputs, the state produced by the 19 July fix was stable at twelve days.

One query, three engines, three answers

The most useful single comparison on 3 August was the same identifying question put to different engines on the same day. They disagreed, and the disagreement is the point.

In these checks, Grok and Google resolved the intended identity, while Perplexity remained ambiguous. The published schema was unchanged during the comparison, but that does not establish whether each system retrieved or interpreted the same material. Differences in retrieval, freshness, context or source selection could explain the results; this study does not isolate their contributions.

A methodological note that matters

One finding on 3 August is worth stating on its own, because it is easy to get wrong and it changes how these tests should be run. When an AI assistant is asked to “search from New York”, its search tools do not use the operator’s IP or VPN. They run from the model’s own infrastructure and return a general, non-geo-personalised view, and the model then infers what a New York user would see. Asked directly, one assistant predicted the Australian site would lead in New York too. A real New York exit IP in a normal incognito browser showed the more conservative truth: present, correct, but competing with US namesakes.

Which leaves the claim where it should sit, stated in tiers rather than as a single headline. In the home market both name forms resolve to one correctly-described entity, and that has held for weeks. Internationally the entity is present and correctly disambiguated, but not yet consolidated in the crowded-namesake markets. The remaining gap is closed by independent corroboration — outside parties discussing the work, which is what makes a research identifier or an encyclopedic entry meaningful in the first place — not by any further change to the markup, which has already done everything markup can do.

What the measurement actually shows

It would be easy to end here on the AI-answer results alone, but the honest picture needs the traffic numbers beside them, because they say something the answer boxes don’t. Over the same window, Google Search Console records roughly 270–360 impressions and 9 clicks in twenty-eight days, impressions up about 69%, clicks still in single figures. Almost every click arrives on the name itself — “doug lord”, “douglas lord” — and the geographic split is Australia 67%, United Kingdom 22%, United States 11%.

These are small numbers, and stating them plainly is the point. AI-layer resolution and click traffic are different curves, and the entity work moves the first one first. The identity now resolves correctly across engines and markets; the clicks are an earlier, separate, slower line that independent citation and promotion will move next. The geographic concentration in the traffic — two-thirds Australian — is the same fingerprint the AI answers show: strong at home, present but light abroad. Every independent source, the answer engines and the server logs alike, tells one story from a different angle.

Anyone can check the small numbers, so they are here in full rather than rounded away. A claim about AI visibility that hides its traffic data is not a case study; it is a brochure. The resolution is real and it is measurable, and so is the distance still to travel.

Search performance: 25 May – 22 August 2026

Google Search Console shows the entity resolving consistently rather than producing a one-off ranking spike. Over the period:

GA4 recorded 70 active users during the same period. Most traffic was direct (55 users), with Google organic contributing 12, Facebook 2 and Bing organic 1. The case study itself recorded 18 views from 10 users with a reported 0.27 bounce rate — the people who reach it are reading it rather than leaving.

The framework resolves, correctly: mid-August 2026

The reads to 3 August tested whether the entity resolved. A second question sat underneath it: when an engine describes the framework the entity is known for, does it describe it accurately? For weeks it did not. Different owned properties stated the model’s structure with different numbers, and the engines recited the disagreement back — one answer would say the framework grouped its signals one way, another answer a different way, because the source pages themselves disagreed. A framework whose own author’s sites cannot agree on its shape is not a rigorous instrument, whatever else it is.

In the second week of August the numbers were reconciled to a single description across every property: 63 observable signals, organised into 6 element groups, measured through 4 cards. Nothing was added to the entity — no anchor, no new citation — only the existing copy was made to agree with itself. Within days the generative reads changed to match. Asked to explain the framework in a multi-turn exchange, Google’s AI Mode now returns the reconciled structure verbatim: sixty-three signals, six groups, the four cards named — Understand, Trust, Cite, Surface — cited from the owned properties.

The consolidation goes global: mid-August 2026

The 22 July sweep found entity resolution to be geographic: consolidated at home and in several international markets, present-but-subordinate in the crowded-namesake US ones. A wider sweep in mid-August — nine markets, both name forms, real exit IPs — showed the consolidated set had widened, not narrowed.

The pattern from July held and extended: where an engine reads on-page structure, the entity resolves and the owned site leads; where the bare name is genuinely contested and the engine leans on independent citation weight, the identity is present and correctly described but not yet preferred. The variable is unchanged. What separates the consolidated markets from the open ones is not markup — that is constant everywhere — but independent corroboration, and that is the one lever the study has named from the start and not yet pulled.

The independent anchor begins to form: August 2026

Every read in this study has pointed at the same missing piece. The entity resolves on-page, the framework is described correctly, the owned properties rank — but the one lever left unpulled was independent corroboration: a source that is not mine, saying who I am, in a place a machine already trusts. On-site assertion can carry an entity a long way, as this study shows, but the citation-weighted engines wait for outside confirmation before they commit.

In August that piece started to arrive. One Vision Productions — an independent not-for-profit that has run for seventeen years — now lists me on its board as Director and Digital Strategist, with a live link to this site. It is not my property, not my words, and not something I can rank into place. It is a third party of genuine standing naming me and pointing at my canonical domain, which is exactly the class of signal the whole study has been waiting on.

The honest status, then: the anchor is forming, not formed. The reciprocal link from my side is already live, so the relationship resolves both ways. What remains is volume and durability — more independent sources, and the encyclopedic entry that only becomes defensible once enough of them exist. The direction is finally right.

Consolidation reaches the hard markets: late September 2026

The mid-August sweep left the pattern clear but incomplete: the entity resolved at home and across several international markets, while the crowded-namesake US markets kept it present but subordinate, and the full name “douglas lord” still met the Wikipedia wall of the medieval knight and the poet. On 29 September I ran the widest sweep yet — nine markets by real exit IP, incognito, both name forms, Google AI Mode and classic search side by side: Sydney, Edinburgh, London, Amsterdam, Japan, Mumbai, New York, Los Angeles and Miami.

The result is the one the study has been working toward, and it is worth stating at exactly the resolution the data supports. On the short form “doug lord”, Google’s AI Mode now opens on the correct entity in every market checked — Sydney, Edinburgh, London, Amsterdam, Japan, Mumbai, Singapore, and the crowded US markets of New York, Los Angeles and Miami. The answer leads with “Digital Strategist & Artist (Australia)” and describes the entity in its own canonical terms — an Australian Digital Authority and AI Visibility Strategist, creator of the Periodic Table of Digital Authority, founder of Digital Dominator, COO and CPO of OG01, sculptor, and musician and co-founder of Moonjuice — citing douglord.com, the OG Collection gallery and Digital Dominator as its sources.

The full form “douglas lord” is further along than in August but not uniform, and the variation is the honest finding. In the US markets (Los Angeles, Miami) AI Mode now leads with the living entity on the full name too. In others — Singapore is the clean example — the full name still opens on the Historical Figures (Lord Alfred Douglas, the Black Douglas, William Lord of Douglas), then carries a “Modern Professionals” group in which the entity is listed first, correctly described, above the librarian and the other namesakes. So the full name has moved from “held entirely by the peerage” to “first among the moderns, beneath the historical block” in the markets where centuries of citation are heaviest, and to outright lead in the US. That is real movement on the hardest query form, reported at the grain it actually occurs.

Two honest boundaries keep this a result rather than a headline. First, the entity leads the answer; it is not the only one in it. AI Mode still lists the other Doug Lords — the boxing manager, the finance executive, the coastal engineer, the peer — below the correct entry. “First and correctly described among a named group” is the accurate claim, and no schema can make a genuinely shared name belong to one person alone. Second, the generative layer has moved ahead of classic search in the hard markets: New York’s AI Mode leads with the entity while New York’s ordinary results still surface US namesakes and obituaries first. That order — the generated answer resolving before the ranked list — is the reverse of what this study saw at home a few months ago, and it is worth recording as its own finding: in a contested market, the AI layer consolidated first.

The other engines still split as they have throughout. Bing’s organic listings resolve the owned properties into the top few for both forms, and its AI Mode for the full name now returns the entity as the primary Australian “Douglas Lord”, correctly separated from Douglas Cooper the Cubist collector and Douglas Lord the library columnist — but Bing’s generated “deep dive” on the short name still resolves wholly to the coastal engineer. The pattern the study named holds on every surface: organic moves first, the generated answer trails, and each engine reads the same asserted identity on its own schedule.

One caveat must sit beside all of this, because the study does not hide the load-bearing weakness. This consolidation has reached the hard markets before the durable independent anchor is in place. The properly-sourced encyclopedic entry does not yet exist — the earlier Wikidata item was removed on notability and has not been recreated, deliberately, until the outside coverage that would let it survive is written. So some of what the engines are reciting rests on accumulated on-page structure and on cached signals rather than on a live external authority; a Wikidata assertion this study created months ago and then saw deleted is, tellingly, still being cited by one engine long after its source was removed — evidence that ingestion is sticky, and also that a mark left by a deleted source is an echo, not an anchor. The consolidation is real and it is measured across nine markets on one day. Making it durable is the work still ahead, and it is deliberately sequenced after the framework’s trademark completes, so the anchor is built once, properly, on a settled name.

The numbers beside the answers: 29 September 2026

The discipline this study set for itself is that every claim about the AI layer is stated next to the traffic reality, because a case study that hides its numbers is a brochure. Here are the numbers, pulled the same day as the nine-market sweep.

Google Search Console, the three months to 29 September, shows the branded queries holding the top of the page and the position tightening as the window narrows:

Both name forms improved their average position in the most recent 28 days against the three-month figure — “doug lord” from 2.92 to 2.23, “douglas lord” from 5.38 to 4.06 — and the full-name CTR nearly doubled. That is the deeper-signal layer catching up to the entity resolution, not a one-day spike. The geographic split is the same fingerprint the AI answers show: Australia leads on clicks (37 clicks / 386 impressions, position 3.89), the United States carries the most impressions but converts less (13 / 525, position 6.95), then the United Kingdom, India, Japan, Sweden and Brazil — strong at home, present but lighter abroad, exactly as the market sweep predicts.

Google Analytics over the same 90-day window records 140 active users (138 new), 611 events, and an average engagement of 23 seconds. Traffic is mostly direct (102 users), with Google organic second (23) and small referral tails from facebook.com, moonjuice.band and bing. The detail worth drawing out is the case study page itself: over 90 days it drew 27 views from 17 users at a 28% bounce rate — against the homepage’s 71.7%. The people who reach this page are reading it, not leaving. In the last 28 days the site held 58 active users and 229 events, with the same direct-led source mix.

These are deliberately small numbers, stated in full rather than rounded away — and their scale is itself the honest point. At roughly 140 users a quarter, the correlation between entity resolution and click lift that a high-traffic site can show cleanly sits, here, inside the noise: you cannot separate a handful of extra clicks from ordinary variance at this volume, and this study will not pretend otherwise. The traffic-lift proof is what a larger property demonstrates statistically; at this scale it is directional, not statistical. What is clean at any volume is the branded-query position — entity-driven and not swamped by traffic — and that moved: “doug lord” from 2.92 to 2.23, “douglas lord” from 5.38 to 4.06 in the most recent window. That, and the answer-layer resolution itself, is the proof the study rests on. The click curve is a later, slower line, reported honestly but not asked to carry the argument.

The same discipline applies to the framework’s own scoring instrument. Assessed under the PTODA methodology, the domain has risen from 31 (Weak) at the May baseline to 41, and now 48 (Developing Presence) — an internal instrument moving in the same direction as the external answer-layer reads. That agreement is worth noting as a parallel signal, not as a proven correlation: three small-sample lines — score, position, traffic — are not cross-plotted here to imply causation the data at this scale cannot support. Each is reported as its own evidence, pointing the same way.

One activity signal worth recording, with its limits stated plainly. Beyond the One Vision Productions board listing noted in August, the founder authored and lodged a Submission to the Senate Inquiry into Artificial Intelligence and Data Centres (Environment and Communications References Committee, September 2026) — a piece of public-policy work on AI, under the name this study is working to resolve. It is worth being precise about what that is and is not. It evidences authorship and civic participation on an AI topic; it does not constitute independent endorsement of the entity’s expertise, because a submission one writes oneself is one’s own assertion, not a third party’s assessment. Nor is it yet confirmed as published: parliamentary submissions are not disclosed until the committee authorises publication, so its status on the public record is pending that authorisation. The genuinely independent corroboration the study still needs — outside parties discussing the work, not the work itself — remains ahead.

What comes next

Results will be updated here as they come in.