Summarize this article with AI
Send it straight to your assistant and get the short version first. The link travels with the prompt, so the model reads the real page.
Opens in a new tab with the prompt already written. Nothing is sent until you press enter.
The short answer
Ask for the denominator, the repetitions, the engines, the baseline and the refusal. Those five answers tell you everything about whether there is a method behind the pitch.
1. What is your denominator?
Any claim about citations needs an out-of-how-many-asks attached. If they cannot give you one, they are describing an impression of a few conversations with an assistant, not a measurement.
2. How many repetitions per query?
Retrieval is not deterministic. The same prompt returns different sources on different runs. Below five repetitions you are reporting a coin flip with a percentage sign on it. We run seven.
3. Which engines, and which models?
Fewer than three engines is a sample rather than a measurement, and engines disagree with each other far more than people expect. We have seen one engine never cite a business at all across a run while three others cited it regularly.
4. What will you measure before you start?
This is the question most consultants have not thought about, and the one that matters most. No baseline means no proof later. It also means they cannot tell the difference between their work and normal fluctuation.
5. What will you refuse to promise?
A good answer includes rankings, citation rates and lead volume. Anyone promising those has told you they did not measure a baseline, because if they had, they would know how variable this is.
A note on the local market
Las Vegas is competitive for construction and almost entirely unworked for this. Most local competitors have no structured data, a stale Google profile and no off-site presence beyond the licensing board. The foundation gap is wide open, which means the bar to clear is lower here than the pricing in this category implies.
Three answers that should end the meeting
Beyond the five questions, some responses are disqualifying on their own. Not because the person is dishonest, but because the answer reveals there is no method underneath.
- We guarantee you will appear in ChatGPT. Nobody controls retrieval. This is either a misunderstanding or a sales line, and both cost you the same.
- We have a proprietary file that gets you indexed by AI. Google's documentation states no special markup is required for eligibility. There is no secret file.[1]
- Here is a screenshot of ChatGPT recommending our client. One answer is a coin flip. Ask how many times they ran it and what the rate was.
A good provider will volunteer the limits of what they can promise before you ask. That single behaviour is more diagnostic than any case study in the deck.
Questions builders actually ask
Should I hire local?
Not necessarily, though someone who knows Clark County permitting, the local trade directories and how homeowners here search has a genuine head start.
What should this cost?
Wide range, and price is a poor signal on its own. The five questions above are a far better filter than the number at the bottom.
Can my SEO agency do it?
Often yes for the overlapping two thirds. Ask them the five questions and see whether they answer or deflect.
Sources
Every claim above that could be checked links to the primary documentation rather than to another agency blog. That is the standard we hold client content to as well.
Share this article with AI
Finished it? Hand the whole thing to a model and ask what it means for your business specifically, including anything in it you think is wrong.
Opens in a new tab with the prompt already written. Nothing is sent until you press enter.
