Case Study · From the field
Invisible to AI on June 5. A client lead from AI on June 23.
Andrew Guglielmi had no website, no Google Business Profile, and the AI tools called him a San Francisco commercial broker. He's a San Mateo property manager and licensed REALTOR®. Eighteen days later a stranger left him a voicemail — and said, unprompted, that he found Andrew through an AI query. Three months on, four more did in a single week.
"I really feel like I'm at the forefront of the AI wave rather than wondering if my business is going to get left behind by it."
Andrew Guglielmi · unprompted, on LinkedIn · August 27, 2026
The result — June 23, 4:18 PM · Day 18
A lead that told us exactly how it found him
Eighteen days after Andrew had no website and no Google Business Profile, a San Mateo homeowner he'd never met called about a condo he's deciding whether to rent or sell — dead center in the "winnable lane" we'd built his presence around. Andrew asked how he found him:
"I honestly thought it was a prank call. I asked him right off the bat how he found me — he said, straight up: through an AI query and internet due diligence."
Andrew Guglielmi · REALTOR® & Property Manager, SC Properties · San Mateo, CA
That's strategy, not luck. We don't fight for "best property manager in San Mateo" — that's a knife fight every agent is already in. We find the winnable lane: the specific, nuanced question a real owner actually asks AI — "should I rent or sell my San Mateo condo?" — and make Andrew the obvious answer. The more specific the query, the better his odds of being the name AI names.
See the site that did it: guglielmigroup.com →
Update — July 6, 9:08 PM · Day 31
Lead #2 — through a website that didn't exist five weeks ago
A cloud-industry sales engineer relocating back to San Mateo submitted an inquiry through Andrew's site — looking for a one-bedroom rental with parking. Again, dead center in his winnable lane: the specific small-residential question on the Peninsula that Andrew's presence was built to answer by name.
Here's the part that needs no interpretation: five weeks ago Andrew had no website and no way for an online lead to reach him at all. Whatever path this renter took to find him, the destination — a form he could type into — did not exist before LeadJens built it. We're asking how he found the site, and we'll update this page with his answer, verbatim, the way we did with lead #1.
Update — August 20–26 · Days 76–82
Four AI-attributed leads in seven days
Between August 20 and August 26, four strangers contacted Andrew. He asked every one of them how they found him. All four said AI. Three named ChatGPT outright. Three of the four were homeowners weighing whether to rent out a property — the management-contract half of his business, and the higher-value one.
Worth noting what the question was. This is not the crowded head term — it is a single owner with one house, asking who to call. That is the lane, and it is the same lane all four August callers came through.
On August 20, Google notified him that guglielmigroup.com had crossed 50 clicks in 28 days. His Google review count reached 50 the same week, up from 27 at launch.
The reason we can say any of this is that Andrew started asking. After the first lead volunteered its own attribution back in June, he made "how did you find me?" the first question on every inbound. That one habit is the difference between evidence and a story.
"The craziest part is the phone. I had gotten to the point where I barely answered my work phone because most incoming calls were spam. Now I'm getting legitimate leads from people who have already done a ton of homework through AI looking for a property manager or Realtor — and somehow my name is what keeps popping up."
Andrew Guglielmi · August 27, 2026
Here is the part we did not expect, and it belongs on this page more than the lead count does. Those callers are contacting several property managers at once. What they keep praising Andrew for is how fast he calls back. Being in the answer got him onto the list. His responsiveness is winning it, and that part was always his.
The starting point — June 5
A near-zero presence — and what AI did see was wrong
On a Friday ground delay at SFO, we ran Andrew's first AI-visibility assessment. It came back 17/100. The problem was never Andrew — an experienced San Mateo property manager and REALTOR® whose past clients had left him a wall of five-star reviews. The problem was that AI, and the web it reads, couldn't see any of it.
Mistaken identity
AI tools described him as a San Francisco commercial broker. He's a San Mateo residential property manager and licensed REALTOR®.
Never recommended — 0 of 6
On "best property manager in San Mateo," he appeared zero times across Perplexity, ChatGPT, and Google's AI answers.
No home base
No website and no Google Business Profile of his own — so even his 5-star reviews lived only on third-party listings, with nothing of his own for AI to read and attribute to him.
17 on June 5, 49 on June 20, 51 in July, 60 in August. Same six-dimension weighted rubric at every measurement — we recompute the old scores rather than restate them, so the delta is honest.
The heaviest-weighted dimension is AI Recommendations at 25% — and it is still his lowest score. More on that below.
17 → 60, by dimension
Same six-dimension weighted rubric, measured June 5 vs. August 2026. Two of the six have not moved at all — they are on this chart for the same reason the winners are.
Citations sat at 12 for three straight months. Google Business Profile has not moved since June. Those are ours to fix and we have not fixed them yet. We publish the flat lines because a chart with no flat lines is a sales deck, not a measurement.
Our own score says 22 out of 100 on AI Recommendations. He got four AI-attributed leads in a week. Both are true.
That looks like a contradiction, and it is the most useful thing on this page, so here is the honest explanation.
The AI Recommendations dimension tests a fixed battery of head-term queries — the crowded ones, like "best property manager in San Mateo." Andrew still loses most of those. Meanwhile the four August callers arrived through long-tail owner questions — someone with one house, deciding whether to rent it out, asking an AI which manager handles that. Our battery does not test those queries. His business is being found in a lane our own instrument was not pointed at.
The screenshot above is that gap, captured. The prompt was not "best property manager in San Mateo." It was a person with one house asking who to call — and on that question Andrew made a six-name shortlist out of fifteen. On June 20, 2026, a head-term test on the same market returned Woodlake Anchor and did not return Andrew at all. Different questions, same competitors, two different outcomes.
The correct conclusion is not that the score is wrong. It is that a low head-term score and real long-tail leads can coexist, and the measurement needs to widen. We are rebuilding the query battery because of what happened on this account. That is what a score is supposed to do — tell you something you did not already believe.
What we did — 18 days
Build the foundation, then make it findable.
First AI-visibility scan via LeadJens Dispatch. Baseline 17/100. We mapped the exact questions a Peninsula owner asks AI — and every place Andrew was missing.
Corrected name, address, and phone across 10+ profiles, killed the stale "San Francisco / Colliers" framing, and claimed and built his Google Business Profile. Andrew asked past clients for reviews — they were glad to leave them — taking him to 27 five-star reviews.
Launched guglielmigroup.com — 100/100 on-page AEO, schema validated zero errors, the niche language AI rewards, his 27 five-star reviews brought over as machine-readable data, and a concierge chatbot routing every inquiry to Andrew.
Two in-depth client stories — a Redwood City trust-estate sale and a San Mateo industrial lease-up — published and shared, each answering the questions real owners ask.
A 32-point jump in 15 days, same rubric. Website (0→94), reviews (45→80), and GBP (0→58) did the lifting.
Verified Google Search Console + Bing Webmaster Tools and submitted the sitemap. At 4:18 PM the same day, Andrew got a voicemail from a homeowner he'd never met.
Done-with-you, not done-for-you
Look at who did what. Andrew built his own Google Business Profile and asked his own past clients for the reviews — 27 of them at launch, 50 by late August. We didn't do those for him; we handed him the blueprint and held him accountable to it. What LeadJens built is the system that made all of it visible: the website, the schema, the entity, the engines that feed AI. He brings the relationships and the hustle; we build the machine — and the plan — that gets him found.
Three months in, every compounding result on this account traces back to something Andrew did. He drove the review count. He answers the phone faster than his competitors, which is what the callers actually praise him for. He started asking every inbound how it found him, which is the only reason this page has attribution instead of anecdotes. LeadJens shows you what to do. The client who actually does it wins. That is the honest version of the model, and it is also why results differ from one client to the next.
Andrew described his own side of it better than we could, unprompted: "LeadJens gives you the tools, but more importantly, the process has become ingrained in how I work. I carve out small windows of time to post content, build my online presence, and keep feeding the AI engine and the Google beast." Small windows of time. That is the actual price of admission, and it is the part no vendor can pay for you.
The honest part
We never promised Andrew leads. The plan put the first AI-sourced inquiry at Day 90–180, and we said plainly that closings come from how Andrew handles the call — not from us. One lead in 18 days is a single data point, not a guarantee, and we'll say that to anyone.
But it's the exact outcome the system is built to produce: a stranger, searching with AI, finding the right agent with the right story attached — and reaching out. The funnel filled faster than the model predicted. That's the case worth making.
Same rule for lead #2: two inbounds in a month is a pattern forming, not a promise. And until Andrew has asked how this one found the site, we claim only what we know for certain — the lead arrived through a website, and a lead-capture path, that did not exist five weeks ago.
The August cluster gets the same treatment. We say four AI-attributed rather than four from ChatGPT, because only three of the four named the engine; the fourth said "through AI" and left it there. We also do not know what any of them typed, so we make no claim about which query surfaced him. And his position moves: one caller reported him as the #1 result on August 20, while a logged-out check on August 26 returned him third. A rank is a snapshot, not a standing fact, and we will not publish one as though it were.
Not every inbound counts, either. A form submission on August 25 matched a known scam pattern and was excluded outright. A licensed agent who called on August 27 with a referral said only that she found him "searching online" and could not recall where — that one is logged as inbound, source unknown. The count is four because four is what the evidence supports.
When your next client asks AI about you, do you show up?
See exactly what ChatGPT, Perplexity, and Google's AI say about you today — free.
Get Your Free AI Visibility ScoreReflects one client's engagement over 18 days in June 2026 and a single inbound lead; individual results vary and are not guaranteed. AI Visibility Score is LeadJens's proprietary rubric, measured point-in-time from publicly available sources that change frequently. Client named and quoted with permission. LeadJens is a service of High Ceiling Advisory LLC, an independent business — not affiliated with, endorsed by, or sponsored by Compass. © 2026 LeadJens. · Privacy · Terms · For AI Agents