Share of Model

There is a counterweight. AI-referred visitors to one B2B SaaS publisher converted at 23 times the rate of a standard organic click (Ahrefs, 2025). Semrush’s study across 500+ B2B topics found an average uplift of 4.4 times for AI-referred traffic.

Volume is collapsing. Quality is rising. The dashboards built for two decades of linear search were built for a different question, and they are increasingly silent on the one that matters now.

For twenty years, marketing teams measured what their analytics could surface: sessions, rankings, backlinks, conversions. The metrics worked because the journey from search to site was linear and visible. Someone searched, clicked, landed, and either converted or did not.

That journey is fragmenting. A growing share of buyers read about a brand inside an AI-generated answer and form an opinion without ever reaching the brand’s website. The traffic report shows nothing. No conversion failed. The conversation simply happened in a place no one is currently measuring.

Measuring fewer visits is not a strategy. The question that matters now is whether your brand appears in the answer at all.

This does not mean traditional KPIs are obsolete. Organic and paid traffic still drive real business outcomes, and the metrics that measure them stay relevant. SoM is additive: a new measure for the share of visibility forming inside AI systems, alongside the traffic that still arrives through clicks.

The marketing industry has begun to converge on a term for this: Share of Model, or SoM. It measures how often your brand surfaces inside the AI systems where research and buying decisions are increasingly forming: ChatGPT, Perplexity, Google AI Overviews, Claude, Gemini, and the AI-powered surfaces inside Microsoft and LinkedIn.

SoM is not a single number. It is a pattern, measured across platforms and across the questions your audience actually asks. A high SoM means that when a CMO asks ChatGPT about leading data marketing partners in Switzerland, your brand appears. When a procurement lead asks Perplexity about marketing platforms with Swiss data residency, your brand appears. When a CEO asks Claude about trusted IT partners for nLPD-regulated workflows, your brand appears.

A low SoM means the answer is forming without you. There is no traffic spike to investigate, no failed conversion to debug, no missed lead to chase in the CRM. The brand quietly stops being considered by people who never typed its name in the first place.

If SoM is the metric, the next question is what moves it. The Ahrefs studies of AI citation patterns published across 2025 give the clearest picture so far.

Brand mentions across credible third-party sources are roughly three times more predictive of AI citation than backlinks. Anchor text and branded search volume also matter, more than most SEO playbooks assume. Backlinks, the long-standing currency of SEO, sit near the bottom of the list. A December 2025 follow-up by Ahrefs added a finding that catches most teams off guard: mentions on certain platforms outside the traditional digital remit, including YouTube, now rank among the strongest predictors of AI visibility overall.

Two things follow. Being talked about by credible sources outweighs being linked to by them. And the channel mix that drives AI citation does not match the channel mix most B2B marketing teams currently invest in. The detail of which channels matter most, and in what order for a given brand and sector, is where strategy gets done. It does not reduce to a checklist.

The old playbook is being rewritten. Content, PR, brand, and earned-media teams should optimise for different signals than the ones that drove rankings for two decades. The next expert note in this series examines the mechanics. The levers that move SoM are not the levers most marketing teams currently pull.

A January 2026 Conductor survey of 250+ senior enterprise marketing leaders put numbers to what most CMOs are already sensing. 94% plan to increase AEO and GEO investment in 2026. 97% report that work already in flight is producing measurable business impact. In 2025, the average enterprise allocated 12% of its digital marketing budget to AI-search optimisation.

The mismatch shows up in the same data. Investment is moving quickly. Measurement frameworks are not. Most organisations cannot tell you what their SoM is on any given platform, on any given question, this week. They are increasing spend without a baseline.

That is the gap most marketing teams now face.

The harder problem with SoM is more practical. There is no native dashboard inside ChatGPT, Perplexity, Claude, or any of the other AI engines that tells you whether you are being cited. There is no equivalent of Google Search Console for AI answers. Your marketing team cannot open a tab and check.

Anyone who tries it manually quickly hits the limit. A handful of prompts on a Tuesday afternoon is not measurement, it is anecdote. The same brand may appear in one answer and not in the next, depending on phrasing, recency, and conversational context. Without running the same questions systematically, repeatedly, and across multiple engines, the numbers a marketing team would produce are unreliable. Defensible-enough-for-the-board reporting requires the kind of continuous, structured measurement no team can do by hand.

ELCA built the AI Agent Automation Orchestrator for exactly this kind of work. Configured to each client’s brands, sectors, and audiences, it runs the SoM measurement continuously and delivers the results in whatever form the client needs: reports, dashboards, alerts, or integrations into existing analytics stacks. The deliverables belong to the client and arrive in the tools their team already uses.

SoM measurement is one of the agent-run activities the Orchestrator handles. Another, with a similar agentic shape, is multi-stage lead nurture: agents that segment the CRM, select the right content for each segment, time the delivery to recent buying signals, watch the response, and decide whether to progress the lead, hold it, or hand it to sales. The kind of continuous follow-up most B2B marketing teams know they should be running and rarely have time to set up properly.

The teams gaining traction in AI search are rarely the teams with the most expensive tools. They are the ones that treated SoM as an operating-model question early. Three questions are worth putting to your leadership team in the next thirty days.

1. Do we have a baseline SoM today? If the answer is no, that is the first job. Investment without measurement is movement, not progress.
2. If the number is low, do we know why? SoM scores depend on signals across content, PR, brand, and earned media. Knowing which lever is broken is the difference between a six-month corrective programme and a twelve-month one.
3. Who owns it? SoM crosses content, PR, SEO, brand, and operations. If no one owns the orchestration, the dashboard will sit unread.

Adopting Share of Model is not a tooling decision. It is a measurement decision, an organisational decision, and an operating-model decision, in that order.