The Website Produced 75 Real Leads Last Month. The CRM Knew Where Three of Them Came From.
Three instruments on one appliance repair site gave three different answers in the same month. Search Console said the channel had doubled. Analytics said the forms produced two to four leads a day. The CRM said the website booked three jobs all month. Here is where the chain breaks between a form submission and a job on the board — and the four checks that find it on any service site in an hour.
Is a $70,000 AI Visibility Contract Different From a $10,000 One, or Are You Buying Seats?
Three separate buyers asked me a version of the same question this month: a COO holding a mid-five-figure quote, a CFO deciding whether to bid two vendors against each other, and a public-affairs finance lead who suspected the tiers were seat count with a story attached. Here is the arithmetic that answers it, using the only enterprise-grade numbers in this category that are actually published.
AI Crawled 1,703 URLs and Pulled 207 Into Answers. The Gap Is the Whole Job.
Being crawled is not being cited. On one client site AI agents touched 1,703 distinct URLs in eight days and pulled just 207 of them live to answer a real person — 12%. On a second site it was 4%. Here is what the 207 have in common, what the other 1,496 were missing, and the uncomfortable thing I found sitting inside the live-fetch log on the second site.
What "AI Visibility" Actually Means: Four Different Numbers, and Only One of Them Is Worth Paying For
When somebody sells you AI visibility, they could be selling any of four completely different measurements — and three of them can go up while your business gets nothing. I measure all four on live sites. Here is what each one is, which instrument sees it, and the specific case where the biggest number was the least useful: 1,132 requests from one crawler that was never answering anybody.
What AI Visibility Actually Costs in 2026 — and Why the Same Work Is Quoted at $99 and at $15,000
Published prices in this category span two orders of magnitude for work described in nearly identical words. That is not a market finding its level; it is four different things being sold under one name. Here is what each price band actually buys, with the vendor numbers as published, and the three questions that tell you which band you need.
llms.txt Takes Ten Minutes to Publish and Is Almost Impossible to Measure. Here's the Trap in Every Report You've Read.
Two live sites, 6,676 requests from AI agents in the last week of August, fourteen distinct crawlers and assistants. Requests for llms.txt: zero on both. I am not going to tell you that means AI ignores llms.txt, because the same logs prove that zero is a measurement artifact rather than a finding. Here is what the logs do prove — including the one file AI agents request more than any other page on either site — and how you would measure llms.txt properly.
If You're Hiring for AI Visibility, Here Are the Questions I'd Want You to Ask Me
Most job specs in this field ask for tools and years. Neither predicts whether the hire can prove the work paid. Here are the six questions that separate someone who reports on AI visibility from someone who measures it — with my own answers underneath, including where they are weak.
GEO, AEO, SEO: Three Labels, One Real Question — Does the Machine Repeat Your Answer?
Every agency now sells GEO, AEO, or both, and most of the definitions floating around were written to justify a new line item. Here is the distinction that survives contact with data: I have six weeks of measurements from one client site where AI responses grew about 41x while organic traffic moved 22%, and server logs showing 813 live ChatGPT fetches against 26 recorded visits. Plus the search-demand numbers that show which of these three labels buyers actually type.
Every AI Visibility Tool Samples Prompts From the Outside. Not One of Them Can See Your Server.
I pay for one of these tools and I built the instrument they're all missing, so this is a buyer's guide written from both sides. What the category actually measures, the check-budget arithmetic that decides how noisy your dashboard is, the engine that was fetching a client's pages 129 times in eight days while the panel reported it as absent, and the honest weakness of my own logger.
Home Services Is the Hardest Place to Prove AI Visibility Works. That's Exactly Why I Test It Here.
HVAC, plumbing, appliance repair and cleaning companies have four structural features that break every standard AI visibility playbook: the conversion is a phone call, the job is created by hand with no source field, the service area is a radius rather than an address, and nobody links to a repair company. Here is what I measured on two live sites in this industry, and the one asset type that worked when service pages did nothing.
AI Share of Voice Is a Ratio Built From a Sample. Here's the Arithmetic Nobody Shows You.
Share of voice in AI answers is the metric every dashboard leads with, and it is a fraction whose numerator and denominator are both estimates. I watched a frozen metric in one tool fall from 194 to 25 while nothing about the site changed. Here is how the number is actually built, the check-budget arithmetic that decides how noisy it is, and the three things I track instead.
GA4 Says AI Sent You 23 Visitors. My Server Log Says 425 in a Fifth of the Time.
I pulled both instruments on the same site for the same period. Google Analytics recorded 23 sessions from all AI platforms combined across a month — 0.47% of traffic. The server log recorded 425 live answer fetches in eight days. That is a factor of roughly seventy, and it is not a tracking bug. Here is why GA4 cannot see this channel, what it does see, and how to set it up so the fraction you can measure is at least measured correctly.
I Think I Found a Repeatable Way Into AI Citations. It's Holding Across Very Different Businesses.
I'm not going to oversell this: I'm still mid-experiment. But the same approach to getting cited by AI is now showing up across three clients in three unrelated industries — and that's the point where a lucky result starts to look like a method. Here's what I'm seeing, and the tooling I'm building so a business can actually see the advantage its AI presence gives it.
600 Out of 7,801: The Silent Bug That Booked Jobs on Techs Who Were Off
An AI voice bot was reading only 600 of 7,801 calendar events before deciding a technician was free — no error, no crash, every run logged 'success.' Here's the anatomy of a silent bug, why it's the most expensive kind, and why you monitor what your automations output, not just whether they crash.
Getting Cited by AI Is the Easy Part. Proving It Brings Orders Is the Real Work.
On a multi-location appliance-repair client's site, AI citations went from 63 responses to 2,605 across six weeks — ChatGPT alone from 1 to 813. Encouraging. But AI presence is not revenue, and the number almost everyone ignores is the only one that pays the bills: orders that actually came from that presence. Here's the full measurement series, what I can and cannot attribute to my own work, and why the count has now plateaued.
I Stopped Chasing Startup Ideas and Started Collecting Business Pains
AI crushed the cost of building software, so implementation stopped being the moat. The real skill now is pattern recognition — hearing 'why are we still doing this manually?' and seeing a product. Here's how I build AI systems that don't lie, and why a supervised AI beats an unsupervised one every time.
35,500 to 148,000 Impressions in 6 Months: Why Templated SEO Gets You Nowhere and Custom Content Gets You There All at Once
Most SEO agencies sell you city page templates. Copy, paste, repeat 50 times. Here's what 6 months of doing the opposite looks like — 4x impressions, 4-5 organic leads per day, no ad spend.
The SEO Strategy That Actually Worked: 14,050% Impressions Without Backlinks or 'Hacks'
I ignored every LinkedIn SEO guru and ran my own test. 64 articles across 18 locations, zero backlink building, zero technical wizardry. Here's exactly what happened and why PAA questions are the most underused tool in local SEO.
From 64 to 1,174 Pages: How I Built a Local SEO System That Drives 4-5 Organic Leads Per Day Across 6 Cities
September 2025. A multi-location appliance repair company. 64 pages, no traffic, no content strategy, no physical storefront. Here's the exact system I built — and why AI-assisted content didn't kill the rankings, it accelerated them.
How to Dispute LSA Credits at Scale: $13,000+ Recovered in One Cycle
Google charges you for LSA leads that were never real customers. Most businesses pay anyway because disputing at scale is too slow to do manually. I built a system that does it automatically — $13,000+ recovered in a single dispute cycle across multi-location accounts.
You Don't Have a Lead Problem. You Have a Conversion Problem You Can't See.
Most home service businesses on LSA think they need more leads. They don't. They need to stop losing the ones they already paid for. Here's where the money actually leaks — and how to find it.
Keyword Cannibalization Is Killing Your Local SEO Rankings — And Most Tools Are Making It Worse
I built a tool that separates real cannibalization from architecture conflicts on local service sites. Blind redirects based on standard cannibalization audits can destroy rankings you spent months building. Here's what to look for instead.
From 50 to 488 Clicks: A Cross-Niche SEO Case Study (Music Education, Zero Backlinks)
I stress-tested my home service SEO framework on a music education site with zero authority. 28 days: 50 clicks. 16 months: 488 clicks, 29K impressions, average position 10.8. No backlinks. No paid tools. Here's the full data.
Before You Copy Someone's SEO Results, Understand Their Business Model
My partner sent me a case study: 10,000 pages, traffic exploded. Let's do the same. I started asking questions. The business model behind those numbers would have gotten us deindexed in three months.
I Analyzed 100 Service Business Calls With AI. What I Found Was Uncomfortable.
Built a script that pulls call recordings from CRM, transcribes with Whisper, and analyzes with Claude. Ran it on 100 calls. The patterns that kill bookings don't show up in any dashboard — until you actually listen.
Google's March 2026 Update and AI Content: What the Doomers Got Wrong
Fire emojis, shocked faces, 'STOP USING AI.' I ran the video through NotebookLM, compared it to 8 months of real data across 280+ AI-assisted pages. Here's what actually happened after the update.
3 AM, Database Down, AI Bot Dead: What I Learned From My Worst Night in Automation
My entire automation system crashed at 3 AM. PostgreSQL full, AI phone bot dead, 5 US locations offline. Here's what actually happened and how I fixed it — and what you should do before it happens to you.
Google AI Overviews: What Home Service Businesses Need to Know in 2026
AI Overviews are showing up for local service queries. Here's what I actually tested across 6 US markets, what triggers them, and how to get cited.
I have a full breakdown for each one.
Each page covers what's causing it, how I fix it, and real numbers from real clients.