30-second rundown
Key learnings:
- AI-assisted sales are often credited to direct or branded search because AI tools may leave no referral trail.
- Combine customer self-reporting, competitive AI visibility on buying questions, and branded-search lift instead of trusting one metric.
- Triangulation creates a more defensible estimate of AI Search revenue while controlling for campaigns that also increase brand demand.
This week: Set up a post-conversion survey asking the user how they found you, monitor branded searches, and select and monitor your brand mentions for five buying-intent prompts. Check the numbers after 90 days.
Coming 3 months: Automate weekly collection of self-reported sources, branded-search trends, and AI share of voice, with one analytics owner investigating synchronized changes each month. The resulting dashboard should show attributable customers, visibility movement, competing brand campaigns, and a recalibrated revenue estimate rather than raw referral clicks.
Eli Smith has found a corpse in the attribution dashboard, and the label says direct traffic. His point is that AI discovery often leaves no referral trail because a buyer sees a brand in an answer, opens another tab, searches the name, and converts through a channel that gets the credit.
1. Ask Buyers Before Trusting the Pipes
Self-reported attribution means asking customers how they first heard about the company on forms, surveys, sales calls, or win-loss interviews. Smith argues that this can reveal AI influence that analytics misses, although the question should be optional or moved after conversion if it harms form completion.
2. Measure Share of Voice Where Buying Happens
AI share of voice measures how often a brand is recommended compared with competitors. Track a focused set of five to ten buying-intent prompts each quarter and weight prominent recommendations more heavily than footnote mentions, because not every citation carries the same commercial voltage.
3. Look for Branded Search Lift
A buyer may hear a name from an AI tool and then search for that brand on Google. Compare branded searches before and after meaningful AI visibility changes, but annotate public relations, paid campaigns, podcasts, and viral events too, or the model becomes a fortune teller wearing an analytics badge.
The proof comes from triangulation: self-reported AI discovery, rising visibility on valuable prompts, and branded search lift moving together. No single gauge owns the truth, but three independent needles pointing the same way can keep a useful program from being executed by a dashboard with poor eyesight.
