The PR Measurement Problem
PR has been measurement-hard for decades. The exposure is hard to enumerate (you cannot count "people who saw the New York Times article" reliably), the conversion path runs through tracked channels, and AVE (advertising value equivalent) was always a flawed metric. The AI search era adds a new dimension: PR placements increasingly feed AI training data and AI assistant retrieval, which means PR's value compounds through AI visibility long after the original placement.
Step 1: Build a PR Exposure Series
Weekly aggregate of PR placements weighted by reach and authority. Tools like Cision, Meltwater, and Muck Rack produce weekly placement counts with circulation estimates. The aggregate exposure series is the MMM input; tier-weighted versions (tier-1 publication mentions count more than tier-3) improve identification.
Important: PR placements also affect AI visibility with a lag. A New York Times feature today often shows up in AI training data and citation patterns weeks to months later. The MMM should capture both the immediate effect on branded search and the longer-tail effect through AI visibility.
Step 2: Enter PR as an MMM Channel
Adstock prior: geometric half-life four to twelve weeks (PR has long carryover). Saturation prior: Hill with gentle aggressiveness; PR rarely saturates within the operational range of most brands. Use a tier-weighted exposure series rather than raw placement counts; tier-1 placements drive most of the channel effect.
Step 3: Use AI Visibility as a Leading Indicator
A high-tier PR placement increases the brand's presence in AI training data and immediately in RAG-based platforms like Perplexity. AI visibility (LLM share of voice) is a measurable leading indicator of PR's downstream effect on consideration and conversion. Track AI visibility movement following major placements; the size of the AI lift is a proxy for the placement's value.
Step 4: Distinguish Direct PR Effect From AI-Mediated PR Effect
PR has two pathways to outcome. Direct: people read the article and behave differently. AI-mediated: AI assistants absorb the article and recommend the brand more often. Both are real; the MMM separates them when AI visibility is included as a separate variable. PR drives AI visibility, AI visibility drives conversion. The two coefficients describe different parts of the same causal chain.
Step 5: Calibrate With Geographic Lift Testing
Pause regional PR distribution in matched markets for eight to twelve weeks. Measure the lift in branded search, direct traffic, AI visibility, and converted revenue in test versus control. Most PR is national, but regionally-targetable PR (regional press tours, regional thought leadership distribution) produces clean geo-test designs. National-only PR can be evaluated through synthetic control around major placements.
Step 6: Build the PR Dashboard
PR teams should report: weekly placement count and tier mix, weekly AI visibility movement attributable to placements (via Presenc AI), MMM-derived PR contribution to revenue (quarterly), and downstream business outcomes (branded search lift, conversion rate). The dashboard replaces AVE with metrics that connect to revenue.
How Presenc AI Helps
Presenc AI provides the AI visibility data that turns PR from an unmeasurable channel into a leading-indicator-supported channel. Weekly LLM share of voice tracks the AI consequence of PR investment; rolling event-study analysis around major placements estimates the placement-specific AI lift. The data lets PR teams move past AVE and report against the AI signal that increasingly mediates PR's downstream effect.