26 years ago I loaded GIF ads onto a server we rebooted twice a day.
The only question: did the ad show up?
Today my book is out. Same question, different machine.
A buyer asks AI for 5 companies. Yours isn't one. Your analytics never see it.
https://t.co/L2uA8tV8sQ
2004: the Japan team flew to Redmond with binders documenting every mistake. My predecessor had been fired. I inherited the job.
AI keeps a binder on your company too.
You never get to read it. Your buyers get its answer.
https://t.co/D036dOFEJ6
0% visible. 88.7% visible.
Same company. Same two months. 9 AI visibility tools.
My first reaction: come on, it just can't be.
Before you trust a score, ask what it measured.
Chapter 10 of The Invisible Audience: https://t.co/sxQgV2tRSN
I say it on almost every call: ask AI the same question 100 times, get 99 different answers.
This time I counted.
13 firms. 22 slots. 3 showed up all three days.
"Variance is weather. A gap is climate." (Emma Jones, Amp'd)
https://t.co/3XttqPjEP7
I ran my own company through nine AI visibility tools.
Scores ranged from 0% to 88.7%. Same company.
Any single score from a tool that won't show its math is a guess. More in my @LeadersPMag interview:
https://t.co/YxKNupudAj
The neutral-looking "best tools for X" list AI cites? Often written by a contestant.
Of 40 source pages behind AI vendor picks, 30 came from companies selling in that category. 22 ranked their own products.
Thanks @TechBullion and Victor Miller.
https://t.co/joNhJDcUQX
By the time a buyer contacts you, an AI system may have already decided whether you're in the running.
Thanks @tech_financials for covering The Invisible Audience.
https://t.co/3mR3p2sd0h
AI knew who I was. It just didn't recommend me.
When someone asked for an expert without using my name, I showed up in about 1% of answers. That gap is what I diagnose now.
Thanks @LeadersPMag for the interview.
https://t.co/YxKNuptFKL
@dynbuzzdigital That's the right number to track! re: showing prompts, some did, others did not, but all 9 showed an average score vs. broken down by branded vs unbranded
One AI visibility tool scored my company 0%.
Another scored it 88.7%.
Same company. Same two months.
I tested nine tools. They weren't measuring the same thing badly.
They were measuring different things.
https://t.co/L2uA8tVGio
You think a rebrand resets your AI visibility.
It doesn't.
AI trusts your old identity more. It has more proof.
Across 10 industrial rebrands I studied, the old name came back in 78% of the questions.
https://t.co/L2uA8tV8sQ
Your competitor shows up in the AI answer. You don't.
Same symptom. Seven possible causes.
More content fixes one of them.
Don't fix the tactic until you know which layer failed.
https://t.co/L2uA8tV8sQ
When ChatGPT stopped citing Reddit, where did the answers come from?
I traced 1,855 citations. Not analysts. Not press.
Mostly companies ranking their own product.
Your competitor's “Top 10” page is telling AI who to recommend.
https://t.co/L2uA8tVGio
A European company swapped the Spanish “y” in its name for “and.”
Its homepage headline still used the “y.”
To anyone inside, obviously the same business.
To a machine, possibly two.
https://t.co/L2uA8tVGio
You optimized one page.
AI checked 50 others.
Google asks: is this page credible?
AI asks: do the sources it trusts describe this company the same way?
No pattern. No recommendation.
https://t.co/L2uA8tV8sQ
A dentist told me patients pick practices whose sites list TMJ or sedation. Then they get referred out.
To a patient, or an AI system building a list, the claim can look exactly as good as the work.
https://t.co/L2uA8tVGio
Your dashboard says you're winning.
Clicks. Sessions. Conversions.
That's the last 20%.
Upstream, AI already built the shortlist and compared the options.
None of it shows up in your metrics.
https://t.co/L2uA8tVGio
Revenue is a terrible predictor of AI visibility.
$400B lost to $2.5B.
1,000+ employees lost to 200.
In my Algorithmic Authority Index, the biggest didn't win.
The most clearly defined did.
https://t.co/L2uA8tV8sQ
Two companies. Same market.
AI called the 15-year veteran “established.”
It called the newcomer a “specialist.”
Established tells a buyer you've been around.
Specialist gives them a reason to call.
https://t.co/L2uA8tVGio