TL;DR
- Board metrics create bad incentives - Boards want growth metrics. SEO dashboards show traffic and rankings. This creates a bias toward high-volume keywords that everyone fights over.
- Zero-volume keywords convert - The keywords that actually convert show "0 volume" in tools. They're too specific. Too niche. Too valuable.
- LLMs fan out queries - LLMs don't search your keyword. They "fan out" into 4-6 sub-queries. Your board tracks the main keyword. The LLM retrieves based on the sub-queries.
- Third-party sources matter - ChatGPT often doesn't care what you say about yourself. It cares what G2, Capterra, and Reddit say about you.
- Introduce Zero-Volume Pipeline Attribution - You can't walk into a board meeting and say "traffic doesn't matter." Instead, introduce Zero-Volume Pipeline Attribution.
The Quarterly Review Problem
Every quarter, the same ritual plays out in funded startups across the world.
Marketing presents to the board. The slide deck has a traffic chart (up and to the right, hopefully). A keyword ranking table. A content output metric showing how many blog posts shipped. The board nods. Growth looks good. Keep doing what you're doing.
Here's what's not on that slide: the keywords that actually brought in customers.
The specific, weird, long-tail queries that real buyers type when they have real problems and real budget. Those keywords don't show up in SEO tools. They have "0 volume." They don't make the deck. So nobody optimizes for them.
When your success metric is "traffic growth," you optimize for traffic. When it's "keyword rankings," you optimize for rankable keywords. Neither correlates with revenue. But both look great in a board deck.
There Are No Hidden Gems
Let's talk about "hidden gem keywords."
There aren't any.
Every SEO tool pulls from the same data sources. Ahrefs, Semrush, Moz. They all see the same queries, the same volumes, the same difficulty scores. When you find a "hidden gem" in your keyword research, your competitor found it too. Last week. They're already writing the article.
The entire industry is playing a game where everyone has the same map. The only variable is execution speed and domain authority. If you have less authority, you lose. If you're slower, you lose. There's no information edge.
But here's what SEO tools can't show you: the queries with "0 volume." These are the specific questions real buyers ask. Too niche to register in keyword databases. Too specific to aggregate. Too valuable to ignore.
The "Source of Truth" Layer
Here's a tactical reality check: ChatGPT often doesn't care what you say about yourself. It cares what others say about you.
When we analyze AI responses for B2B queries, we see a pattern. The AI builds answers from what we call "Trust Hub" domains: G2, Capterra, Reddit threads, industry publications, news sites. These are the sources it returns to repeatedly.
Your company blog saying "We're the #1 solution" carries almost no weight. But positive sentiment on Reddit, reviews on G2, mentions in trusted publications, your research quoted on industry sites — that's what builds consensus in the AI's mind.
If you have 50 blog posts but no presence across Trust Hub domains, you're invisible to the inference engine. The strategy isn't just "write content on your domain." It's "build presence across the domains AI trusts for your category."
The Trust Hub Audit
Map Your Trust Hub: Run 30-50 queries in ChatGPT/Perplexity for your category. Track every domain cited. That's your target list.
Prioritize Diversity: It's not just Reddit. You need mentions across multiple Trust Hub domains: G2, Capterra, industry publications, news sites, niche blogs.
Work Systematically: If you aren't on those domains, your own blog content won't move the needle yet. Define the sites, then nail them one by one.
The Math Nobody Does
Keyword A: 10,000 monthly searches. Informational intent. "What is [category]?" Your SEO tool found it. Your competitor's found it too. You both write articles. You split the traffic. But with AI Overviews and LLMs giving answers directly, your click-through rate tanks. Maybe you get 200 visitors. Conversion rate: 0.05%. You get 0.1 leads.
Keyword B: 0 monthly searches (according to tools). High intent. "[Product] for [specific role] at [specific company type]." No one optimizes for it because it doesn't show up in research. But here's the thing: when you create content for a specific query, you also rank for surrounding keywords in that cluster. That "0 volume" page might bring 50 visitors from related searches. Conversion rate: 10%. You get 5 leads.
50x the outcome. Only one makes the deck.
Why Specificity Wins (The Technical Reality)
The shift to AI search isn't just a marketing trend. There's an engineering reason why specific content wins.
Traditional SEO works on lexical matching. Count keywords. Match "best CRM software" to people searching "best CRM software."
LLMs work on semantic matching. They match meaning, not keywords. When someone asks ChatGPT a question, the system converts that question into a vector (a numerical representation of meaning), then searches for content with similar vectors.
A specific query like "CRM for 10-person agency using Linear that needs Slack integration" creates a tight semantic cluster. Generic content about "best CRM software" is too broad to match well. The more constraints in the query, the more specific content wins. This isn't philosophy. It's how vector search works.
The "Fan-Out" Reality: How AI Actually Searches
There's an architectural reason why zero-volume keywords work, and it has to do with Query Decomposition.
When a user asks Perplexity or SearchGPT "what is the best inventory software?", the model doesn't just look up that string. It "fans out." It generates 4-6 sub-queries to build a comprehensive answer.
Entity Positioning: Why This Matters
To an LLM, "HIPAA compliant CRM" isn't a keyword string. It's a relationship between two entities: [Concept: CRM] and [Constraint: HIPAA].
When you write specific, zero-volume content, you aren't just ranking for a long-tail keyword. You are creating a strong vector relationship between your brand entity and a specific constraint. When the AI fans out its query to check for that constraint, your content is the mathematical match.
The Protocol: How to Mine Zero-Volume Gold
You can't find these opportunities in Ahrefs. You have to extract them. Here's the protocol:
The Extraction Protocol
Mine for Constraints (The "But" Search)
Search your sales call transcripts (Gong, Chorus, whatever you use) for the word "but."
"We love HubSpot but we need HIPAA compliance."
"We need a project tool but our devs refuse to leave the terminal."
Every "but" is a vector constraint. It defines the specific territory where you can win.
Reverse-Engineer the Prompt
Take that constraint and write the prompt that solves it.
Don't write: "HubSpot Alternatives" (Volume: 50k, Competition: Brutal)
Write: "HIPAA compliant CRM for healthcare startups using HubSpot for marketing" (Volume: 0, Value: High)
Optimize for the Answer (The 40-Word Rule)
LLMs are prediction engines, not readers. They want the answer up front.
• Start with a direct 40-50 word answer to the specific question
• Use lists and tables. LLMs parse structured data better than prose.
• If you help the LLM generate its answer, it will cite you.
Structure is the Signal (The Table Heuristic)
LLMs are not reading your prose. They are parsing your data.
Prose is unstructured data. It's hard to extract specific specs, pricing, or constraints from a wall of text. Tables are structured data. They're essentially CSV files rendered in HTML.
Don't write: A paragraph comparing Tool A and Tool B.
Do write: A Markdown comparison table with rows for "SOC2 Compliance," "Seat Minimums," and "Integration Cost."
We've found that LLMs cite structured tables at a significantly higher rate than prose. Why? Because it lowers the computational friction of extraction. If you make it easy for the AI to steal your homework, it will cite you as the source.
Where Converting Keywords Actually Come From
Here's where it gets interesting. SEO tools show you what's measurable. But some of the best keyword opportunities sit in places tools can't reach: sales calls, support tickets, client conversations.
Most SEO work happens on the screen. But the real goldmine is a 30-minute call with someone who talks to customers every day. Your sales team hears exact phrases on calls. Support sees the questions people actually ask. Founders get market intelligence from investors and industry conversations.
You don't realize how much SEO value sits in these conversations until you start mining them.
The Extraction Protocol
Talk to Sales
Ask what phrases prospects use in calls. Not what analytics say. What customers actually told them.
Listen to Call Recordings
If you use Gong or Chorus, transcribe them. Look for specific language. "We need a way to..." is gold.
Mine Support Tickets
If someone asks "can you integrate with X?" in support, people are searching that too.
Ask About Market Intel
Gartner reports, VC insights, industry trends. Your team often knows about emerging categories before SEO tools do.
I spoke with the CPO of a security company. He mentioned a keyword he started hearing more often among CTOs on sales calls. I checked it in SEO tools: zero volume. Zero competition. We created a dedicated page and a blog article. Within two weeks, the client dominated search and LLM results for that term. This zero-volume article brought them more qualified leads in one quarter than their highest-traffic informational article.
Validating Before You Build
Sales calls can surface internal jargon that customers say on the phone but never type into Google. Before building a page, do a quick validation:
Quick Validation Check
Type the phrase into Google. If autocomplete suggests it, real people search for it.
Check "People Also Ask." Related questions mean you've hit a real query cluster.
Search in quotes. Any results at all means the exact phrase exists in the wild.
This takes 60 seconds. It bridges human insight and data without relying on volume estimates that don't work at this scale.
The Funded Company Disadvantage
Well-funded companies have resources. More writers. Bigger budgets. Better tools. This used to be an advantage.
In the AI visibility era, it's becoming a liability.
Resources create pressure to show ROI. ROI pressure creates metric optimization. Metric optimization creates a bias toward measurable, visible, volume-based content. This content is exactly what AI has commoditized.
Exceptions exist. Smart growth teams at places like Zapier or Canva have already figured this out. But for most Series B startups, the pressure to show immediate traffic growth makes this pivot politically impossible.
Meanwhile, a tiny startup with no board to report to can spend six months creating hyper-specific content for a narrow persona. No traffic. No rankings. Just customers.
The Bottom Line: How to Sell This to the Board
You cannot walk into a board meeting and say "Traffic doesn't matter." You will get fired.
Instead, introduce a new metric: Zero-Volume Pipeline Attribution.
Run a manual analysis of your last 20 closed-won deals. Cross-reference the first touch with search queries where possible. In our experience, many companies discover that a meaningful portion of their best revenue came from search queries that tools show as "0 volume" or "N/A." Your numbers will vary, but the pattern is worth investigating.
Present that data. Show that the high-volume keywords drive traffic, but the specific keywords often drive revenue.
For funded companies: Use this data to buy permission for a "Portfolio Strategy." 80% of effort goes to the board-pleasing volume keywords. 20% goes to testing zero-volume keywords with high buyer intent.
For bootstrapped companies: Ignore the volume entirely. Let the VC-backed giants fight over the generic keywords. You can build a profitable business on the queries they're too big to see.
Yuval Halevi
Yuval, an expert in SEO with over a decade of experience, helps startups simplify their digital marketing strategies. With a focus on practical solutions and a track record of success as a digital nomad and successful company builder, he drives growth through effective SEO, growth hacking, and creative marketing.