How to Read AI Share of Voice (and What to Do About a Low Score)

If a buyer at one of your target accounts opens ChatGPT tomorrow and asks "What are the best [category] platforms for B2B SaaS?" — does your company show up in the answer?

For most B2B SaaS companies, the honest answer is: we don't know. And it matters now in a way it didn't eighteen months ago. Recent industry data suggests that between 30% and 50% of buying-research queries that used to start in Google now start in ChatGPT, Perplexity, Claude, or Gemini. That share is growing. The companies that get cited in AI answers get added to shortlists. The companies that don't, don't.

This is the gap your AI Share of Voice score is measuring.

What AI Share of Voice actually means

AI Share of Voice (AI SOV) is citation share in AI-generated answers. It is not the same thing as either of the older metrics that sound like it.

It is not paid Share of Voice (your impression share against competitors in ad auctions). It is not SEO Share of Voice (your aggregated organic ranking weighted by query volume). Both of those metrics tell you something about the legacy search world. Neither tells you anything about whether an AI engine cites your company when answering a buyer's question.

Here is what AI SOV measures, mechanically: a defined set of buyer-intent questions in your category (for example, "What are the best B2B SEO audit tools?" or "How do I improve technical SEO for a Webflow site?") is submitted to one or more AI engines that perform real-time web retrieval. Citations are parsed out of the answers. Your domain's share of total citations across that question set is computed as a percentage.

The score updates over time. The score is comparable across competitors in the same category. And — this is the part most companies miss — the score is largely independent of your traditional SEO rankings. A site that ranks #1 in Google for a query can have a 0% AI SOV for the same question, because the AI engine simply didn't pick it up.

How to read your score

Kennext reports AI SOV as a single percentage, alongside historical trend, on your Health dashboard. The number is small, in absolute terms, for almost every business. That's expected. AI engines cite many sources per answer, and the long tail is wide.

Here's a rough reading guide for B2B SaaS:

0% — invisible. No AI engine in the test set cited your domain for any of the buyer-intent questions you're tracking. Common for sub-Series-A companies and for older sites that haven't updated their content in two years. Recoverable.

1% to 5% — present, but quiet. You're showing up occasionally, probably for branded or near-branded queries. You are not showing up when a buyer asks an unbranded category question. This is where most B2B SaaS companies sit when they first measure.

5% to 15% — earning citations. You're getting picked up across multiple non-branded questions. You probably have at least one piece of content that's become a reference for AI engines. The work now is to make sure those citations point at the right pages.

15% to 40% — strong category presence. You are one of the recognized voices in your category. Buyers using AI to research are seeing your name. This is where the AEO investment starts to pay back materially.

40%+ — category leader. You dominate the citation share for the questions you care about. Rare. Usually requires either a large content investment, a strong domain authority, or both. Defensible if maintained.

The single most important thing to look at is trend, not absolute level. A 3% AI SOV that's been climbing 0.5 points a week for three months is a far better signal than a 12% that's been flat for six months. Citation share compounds. Stagnant scores warn you that whatever you've done so far isn't working.

Why your score might be lower than your SEO rankings would suggest

If your traditional SEO is decent but your AI SOV is near zero, one of five things is usually true. They are listed in rough order of how often we see them with B2B SaaS customers.

1. Your content is positioned for ranking, not for answering. AI engines retrieve content that answers a question directly — typically the first 100 to 200 words of a page. If your blog post opens with a 400-word context-setting introduction before getting to the actual answer, the AI extractor will skip you in favor of a page that gets to the point.

2. Your schema markup is missing or weak. AI engines lean on structured data to understand what a page is about. If your pages don't have FAQ schema, HowTo schema, Article schema, or BreadcrumbList — and the relevant ones for the page type — you're leaving free signal on the table.

3. You have no obvious authority on the topic. AI engines are skeptical of single-source claims. They preferentially cite domains that have multiple pages on a topic, that are linked to from other authoritative sources, and that have a discernible point of view. A single thin blog post that mentions a topic once won't earn a citation.

4. Your brand name isn't consistent in your own content. This sounds trivial. It isn't. If your homepage, your About page, and your blog use slightly different brand phrasing — "Acme", "Acme Inc", "Acme Software", "the Acme platform" — AI engines have a harder time aggregating citations under a single canonical brand. You end up split across multiple lower-confidence references.

5. AI engines literally cannot read your content. This is the killer. JavaScript-only rendering, infinite scroll without server-side fallback, content gated behind interactive components — these break the retrieval step entirely. The AI can't cite what it can't read. If your site is heavy on client-side rendering and you've never tested what raw HTML it serves, this is a likely root cause.

What to do about a low score

In rough order of effort to impact:

This week — fix what's already broken. Run a scan that checks for missing schema, broken pages, and content that AI engines can't read. Fix the technical floor. Until your site is reliably crawlable and structured, none of the content work will compound.

This month — rewrite your top five pages to lead with the answer. For each of your top five pages by traffic, rewrite the first 150 words so they directly answer the question the page is targeting. Move the context-setting introduction below. This single change has produced the largest week-over-week AI SOV gains we've seen across our customer base.

This quarter — build out topic depth. Pick two or three topics where you want category authority. Write three to five interlinked pieces of content per topic. Make sure each piece is genuinely good — AI engines are getting better at filtering thin content. Quality and depth beat volume.

This year — earn external authority. Citations from third-party publications, industry reports, partner blog posts, and earned media disproportionately move AI SOV. They tell the AI engines that you're a recognized source, not a self-promoting one. Most B2B SaaS marketing teams underinvest in this work because it's slow. It's also the most defensible move you can make.

Why now matters more than later

There is a temptation to wait. AEO is new. The metrics are still being defined. Best practices are still emerging. It feels reasonable to put off the work for six or twelve months until the playbook is clearer.

We think this is wrong, for a specific reason: AI engines learn cumulative patterns. The pages an AI engine cited last quarter are more likely to be cited this quarter. The brands that built early authority are now compounding it. The companies that wait are starting from zero against competitors who are not.

This is not a fashion cycle. It's a content-and-authority flywheel that started turning in 2024 and is accelerating. Every quarter you wait, the gap to the category leaders gets harder to close.

The Takeaway

Kennext measures AI SOV continuously for the buyer-intent questions in your category and reports the trend on your Health dashboard, alongside the technical and content-quality factors that drive it. We tell you what your score is, what specifically is hurting it, and — because we generate the actual code fixes for the issues we find — what to ship to move it.

You don't have to fix it all at once. You do need to know your starting point. Run a free scan at kennext.ai and we'll show you your current AI SOV alongside the issues that are most likely keeping it low. Have questions? See the FAQ or reach out through the contact page.