AI search visibility is now measurable, and most consultants are not measuring it. Google AI Overviews now appear on roughly 25% to 60% of searches depending on the tracker, and 84% of the sources they cite are earned media, not brand-owned pages, according to a 2026 statistics review from Instant Press. If your client's business does not show up when ChatGPT, Perplexity, or Google's AI answers a buyer's question, that client is invisible in a channel already replacing the search results page.

That gap is a diagnostic product. You can build the dashboard that proves it inside HubSpot's free CRM tier, this week, with zero new software spend.

The bottleneck nobody is watching

At Munich Re, we called it the 90-Day Bottleneck Audit. Every engine room casualty drill starts the same way: find the constraint, measure it, fix it, verify the fix held. You don't fix what you can't measure. That's the whole discipline.

AI search visibility is the new bottleneck for consultants, and almost nobody has instrumented it. Your clients are spending money on SEO retainers built for a search engine that is disappearing under their feet. Meanwhile the AI engines replacing it operate on rules SEO never had to answer to.

If your clients don't show up when ChatGPT answers, they don't exist. That's the sentence that opens the sales conversation for this diagnostic. Read that back to a prospect and watch them reach for their notepad.

Why AI citation behavior is a different game than SEO

A study of 1,000 Google AI Overview citations by The Stacc found that the top 1% of domains, about 12 websites, capture 47% of all citations. Wikipedia alone accounts for 24.3%. The next 9% of domains split 31% of what's left, and the remaining 90% of the web fights over 22% of citation share. That's a power-law distribution steeper than organic search rankings ever were.

It gets worse for anyone assuming their SEO rank protects them. The same research found only 17% to 38% of AI-cited pages also rank in the organic top 10. AI visibility and search rank are separate systems now. A business can dominate page one of Google and still never get named when a prospect asks ChatGPT "who's the best [service] in [city]."

A separate Semrush analysis of roughly 150,000 AI citations found Reddit is the single most-cited domain across AI engines overall, at 40.1% of citations, ahead of Wikipedia at 26.3% and YouTube at 23.5%, according to research compiled by Mentionova. Reddit signed a content-licensing deal with Google worth roughly $60 million a year in February 2024, and OpenAI followed with a similar deal near $70 million a year months later. That's not organic authority. That's a purchased pipeline into the training and grounding data these models read from.

For B2B and commercial queries specifically, a study of 1,259 citations across software categories by Derivatex found third-party "best of" lists earned 63% of citations, vendor-owned pages just 17%, and YouTube 9%. In the CRM category alone, YouTube pulled 26% of citations and vendor sites just 6%. Your client's own website is aimed at roughly 12% of where AI citations actually originate. That's the diagnostic finding that justifies the invoice.

What to track: the three metrics that matter

Strip this down to what a consultant can actually sell and defend in a client meeting. Three metrics.

AI citation frequency. Run a fixed list of 15 to 30 buyer-intent prompts across ChatGPT, Perplexity, Google AI Overviews, and Claude. Log whether the client's business is named, cited, or absent for each prompt. This is your baseline number: cited in 3 of 20 prompts, or 14 of 20.

Track it monthly. It moves.

Source attribution. For every prompt where the client is cited, or should be, record which domain the AI actually pulled from. Was it the client's own site, a review platform, a directory, Reddit, YouTube, or a competitor's content that mentioned them? This tells you where the real work is. If 60% of your client's potential citations run through Yelp and G2 profiles that are three years out of date, that's the fix, not a new blog post.

Competitor mention share. Every time you run the prompt set, log which competitors get named alongside or instead of your client. Over 90 days this builds a share-of-voice number you can chart. It's the same logic as market share, just measured inside an AI answer instead of a sales report.

Building the dashboard in HubSpot's free tier

HubSpot's free CRM gives you unlimited custom properties on the Contacts or Companies object, up to 1,000 marketing contacts, and basic reporting: 3 dashboards with up to 10 reports each, according to HubSpot's own pricing page and a 2026 breakdown from EngageBay. The full custom report builder is a Professional-tier feature. You don't need it. Standard single-object reports on custom properties do the job for this product.

Step 1: Create the client record structure. Go to Settings > Data Management > Properties. Select the Companies object, since each client is a company, not a contact. Create a property group called "AEO Audit" so these fields don't scatter across your CRM.

Step 2: Build the tracking properties. Add these as custom properties under the AEO Audit group:

  • `aeo_citation_count` (number): how many of your fixed prompt set returned a citation this cycle
  • `aeo_prompt_total` (number): total prompts tested, usually 15-30
  • `aeo_citation_rate` (calculation or manually updated number): citation_count divided by prompt_total
  • `aeo_primary_source` (dropdown): Client Site, Review Platform, Directory, Reddit/Forum, YouTube, Competitor Content, None
  • `aeo_top_competitor_mentioned` (single-line text): name the competitor showing up most
  • `aeo_last_audit_date` (date)
  • `aeo_audit_tier` (dropdown): Snapshot, Full Audit, Retainer

Use dropdowns wherever the answer set is fixed. Free text creates reporting chaos, and HubSpot's free tier has historically limited total custom properties per object, so don't waste slots on fields you can standardize.

Step 3: Populate the data manually each cycle. Run your prompt set once a month per client. Log the results directly into the company record. This is manual work on the free tier since workflow automation requires a paid plan, but for a diagnostic product billed per engagement, manual entry is 20 minutes, not a cost center.

Step 4: Build the report. Navigate to Reporting > Reports > Create report, and build a standard report on the Companies object filtered to your AEO Audit property group. Chart `aeo_citation_rate` over time by company. Add a second report breaking down `aeo_primary_source` across your client base, so you can see at a glance how many clients are dependent on Reddit or review platforms they don't control. Pin both to a dashboard titled "AEO Client Visibility."

Step 5: Export the client-facing view. The free tier's reports export to PDF and CSV. Screenshot the trendline, drop it into a one-page summary alongside the raw citation logs (prompt, engine, cited yes/no, source), and you have the deliverable that anchors your invoice.

Packaging it as a paid diagnostic

Real consultants are already charging for this. Avante Visibility sells a GEO/AI Visibility Audit at a flat $2,500: 28 live citation tests across five engines, a 12-point score, a three-competitor benchmark, and a 90-day roadmap, delivered in 48 hours. Misha Manko runs a nearly identical structure at $2,500 with 30+ citation queries and an 18-page report. On the higher end, Maria Dykstra's Algorithmic Authority Audit runs $4,500 for a three-week, 30-to-45-page diagnostic with a 90-day re-measurement built in.

The pattern across all three: fixed price, fixed scope, a scored report, a competitor benchmark, and a roadmap. None of them are selling a retainer up front. They're selling the audit, then letting the findings sell the implementation work.

That's the 90-Day Bottleneck Audit playbook again: measure first, propose second. You don't pitch a fix before you've proven the bottleneck exists. This connects directly to the operator-independent systems doctrine I keep repeating in this series.

Structure your own offer the same way. Price the initial diagnostic at $2,500 for a single business, scaling to $5,000 for multi-location or multi-brand clients. Deliver the HubSpot-generated trendline, the source-attribution breakdown, and a ranked list of what to fix first: claim and update the review platform, get cited on the Reddit threads that already discuss the category, fix the schema markup that determines whether AI crawlers can even parse the site.

Re-run the audit at day 90 and show the delta. That delta is what turns a one-time diagnostic into a retainer. It's the same build-to-sell logic behind the FOCUS Strategy: narrow the offer, prove the result, expand from there.

Doctrine Connection: Process beats ego

Every consultant wants to walk into a pitch meeting with an opinion. "Your marketing needs work." That's ego talking, and clients have heard it from six agencies before you. Process beats ego.

Walk in with the citation log instead. Fifteen prompts, four engines, a number: cited in 2 of 15, competitor cited in 11 of 15. That's not an opinion. That's a casualty report.

Nobody argues with the instrument reading. This is the same discipline behind the Sovereignty Stack: build the system that produces the proof, then let the proof do the selling.

FAQ

Q: Do I need HubSpot's paid tier to build this dashboard? No. The free CRM tier supports unlimited custom properties on the Companies object and includes 3 dashboards with up to 10 reports each, which is enough to run this diagnostic for a full client roster before you'd need to upgrade.

Q: How many prompts should I test per client audit? Fifteen to thirty, covering the buyer-intent questions a real prospect would ask: "best [service] in [city]," "[service] vs [competitor]," "how much does [service] cost." Consistency across audits matters more than volume.

Q: Which AI engines should I test? At minimum ChatGPT, Perplexity, and Google AI Overviews. Add Claude and Gemini if the client's buyers skew technical or enterprise. Each engine pulls from different sources, so citation rates will vary by engine, not just by client.

Q: What if the client isn't cited anywhere? That's the sale, not a failure. Zero citations means zero AI-driven referral traffic in a channel already producing traffic growth of roughly 796% year over year from generative AI platforms, per the Instant Press statistics review. Frame the audit findings as the starting balance sheet, then price the fix work.

Q: How is this different from a standard SEO audit? SEO audits measure organic rank. This measures whether an AI model names the business at all, and where it's pulling that answer from. The overlap between the two is small. Only 17% to 38% of AI-cited pages also rank in the organic top 10, so a client can pass an SEO audit and fail this one.

Build the instrument first. Every AI Search Sensor product I've reviewed in this series follows the same procedure: measure the bottleneck, price the proof, sell the fix. Consultants who skip the measurement step are still pitching opinions in a market that now runs on citations.

*Jeff Barnes, MBA has no personal position in any company, fund, or platform named in this article. demg.ai has no current commercial relationship with any party mentioned. demg.ai provides marketing systems and education services, not investment advice. Past performance does not guarantee future results. All business decisions involve risk, including loss of capital.*