Cost Per Lead Is a Dying Metric. Start Measuring Share of Answer.

Cost-per-lead math assumes a buyer clicks, lands, and fills out a form. That funnel is disappearing. When an AI engine reads your catalog and answers the buyer's spec question directly, the lead either arrives ready to transact or never arrives at all — and the metric that tells you whether you're in that conversation isn't CPL, it's share of answer.

The funnel stopped funneling

Start with what's actually happening to the top of the funnel. In 2026, roughly 68% of U.S. Google searches resolve without a click to any website, and that rate climbs past 80% when an AI Overview appears in the results — figures tracked by SparkToro's ongoing zero-click research. B2B sites have been hit disproportionately hard: industry trackers now put average year-over-year organic traffic declines in the 30-40% range for B2B, with some sectors seeing 70-80% erosion, because the how-it-works, spec-comparison, and application-guide content distributors publish is precisely the content AI Overviews and chat answers are built to summarize.

Forrester's 2026 buyer research found 94% of B2B buyers now use generative AI somewhere in their purchase process, and more buyers name conversational AI as their single most useful research source than name vendor websites, product pages, or sales reps combined. Buyers aren't abandoning research. They're doing it inside an interface that never sends you a session to count.

Distribution Strategy Group has been circling the edges of this shift without naming it directly. Their 2025 piece on marketing automation is right that automation is doing more of the work — the "two-funnel" example they cite, where a distributor separates a transactional storefront from a high-touch nurture track, is a real pattern. But the piece frames automation as better cadence tooling: drip sequences, triggered emails, a "Country Club" tier for qualified accounts. That's optimizing the mechanics of a funnel that's shrinking at the mouth. If 68 to 83% of the queries that used to become site visits now get answered before anyone reaches your welcome series, no amount of email sequencing recovers that volume. The nurture track only works on people who arrive.

Cost per lead was measuring the wrong thing even before this

Go back further. Distribution Strategy Group's 2023 explainer on cost per lead tells distributors to track CPL, then immediately hedges: prioritize lead quality, not volume, because "not all leads hold equal value." That hedge is doing a lot of work. If the headline metric requires a permanent asterisk to mean anything, it's not the right headline metric — it's a proxy for something the industry didn't have a cleaner way to measure yet. CPL survived as long as it did because the alternative, actually tracking whether your product content answered the question a buyer asked, wasn't measurable at scale. It is now.

What share of answer actually is

Share of answer is simple to define and increasingly straightforward to measure: of the spec, compatibility, and application questions a buyer could ask about a category you sell, what fraction get answered — and your product cited — by the AI systems buyers are actually using. It's the AI-search analog of share of voice, except the currency isn't impressions, it's citations inside an answer a human reads and acts on.

The mechanics are less mysterious than "AI visibility" vendors make them sound. AI engines parse structured data, comparison tables, FAQ content, and clean HTML far more reliably than they parse a PDF spec sheet or a JavaScript-rendered product grid. Research on AI citation behavior finds each missing structured-data element — schema markup, FAQ markup, comparison tables, crawlable (non-JS) rendering — costs a brand roughly 6-8 points of prompt coverage, the gap between a catalog that shows up in 25% of relevant AI answers and one that shows up in 70%. Separately, brands that do get cited in AI Overviews see about 35% more organic clicks and convert at 4-9x the rate of ordinary organic traffic, because a citation is a pre-qualified referral — the buyer already got the answer and is now confirming the source.

This is exactly what we built the Digital Readiness Index to measure. AEO readiness is one of its four pillars, scored the same way an answer engine would score you: crawlable product data, structured markup, spec completeness, machine-parseable comparison content — measured from each distributor's own live site, not self-reported. Across the 200+ distributors in the Top Distributors 2026 index, readiness on this pillar tracks almost nothing about company size and almost everything about whether product data lives in structured, syndicated form or is trapped in a legacy PIM export nobody's touched since a spec sheet PDF.

Where the budget actually needs to move

None of this argues for abandoning automation — nurture sequences still matter for the buyers who do land on your site. It argues for reordering the budget. Marketing dollars currently rent clicks: paid search, retargeting, sponsored listings, all metered against a funnel that's leaking at the top. Share of answer redirects a slice of that spend toward owning the underlying asset — catalog data structured well enough that an AI engine can parse it, cite it, and hand your product to a buyer who's already decided. That's not a marketing-automation problem. It's a catalog-quality mandate that marketing, sales ops, and whoever owns the PIM all have a stake in.

That reframing is Anglera's whole thesis: your PIM stores the data, but getting it into the shape an AI engine will actually cite — complete specs, clean structure, consistent formatting across thousands of SKUs — is enrichment work most distributors don't have the headcount for. We built the tooling to do that at scale, live from a flat file in weeks, without ripping out anything you already run. Measuring share of answer only matters if you can move it — and moving it starts with product data an answer engine can actually read.