Originally published 2011. Rewritten September 2026 to reflect current measurement standards.
PR measurement has gone through two full generations since the Advertising Value Equivalent (AVE) era. The first shift, formalized by the 2010 Barcelona Principles, retired AVE in favor of share of voice, message pull-through, and sentiment. The second shift, underway now, adds a metric none of those frameworks anticipated: whether a brand shows up in the answer an AI engine gives a buyer, measured as Citation Share.
Why AVE died and stayed dead
The 2010 Barcelona Principles, developed by AMEC (the International Association for Measurement and Evaluation of Communication), named AVE explicitly as not a valid measure of communications value: it conflates editorial coverage with paid advertising space and has no defensible link to outcomes. AMEC's subsequent Barcelona Principles 3.0 (2020) and its Integrated Evaluation Framework hardened that position into the industry standard most measurement vendors and in-house teams now build against.
The metrics that replaced it
Four metrics carried PR measurement through the 2010s and remain useful today, provided they're benchmarked, not reported as raw counts:
Share of voice. A brand's percentage of category coverage against a named competitive set, in a stable universe of outlets. Comparative, not absolute — a rising share against named competitors is real signal.
Message pull-through. The percentage of coverage carrying the brand's intended positioning or spokesperson quote. A smaller volume of high pull-through coverage consistently outperforms a larger volume with weak message capture.
Sentiment. Favorable, neutral, or hostile coverage, ideally human-coded on a representative sample rather than fully automated, since automated sentiment tools still miscode sarcasm and industry jargon at meaningful rates.
Referral traffic and inbound links. The direct line from earned coverage to site behavior, which outlets sent traffic, which mentions produced quality backlinks, cross-referenced against a coverage calendar.
The metric these frameworks didn't anticipate: Citation Share
All four metrics above assume a buyer reads coverage directly, or at minimum sees a headline. That assumption is breaking. When a buyer asks ChatGPT, Claude, Perplexity, Gemini, or Google AI Overviews a category question, the engine synthesizes an answer from whatever it retrieves as authoritative, and increasingly the buyer never touches the underlying article at all.
Citation Share measures the percentage of relevant AI-generated answers in which a brand is actually named or cited. It is measured per prompt, per engine, and tracked as a trend rather than a single reading, since AI answers are volatile week to week. A brand can hold strong share of voice in trade press while showing near-zero Citation Share, if none of that coverage carries the entity clarity, structured data, or source authority the engines retrieve from.
What a 2026 measurement program actually tracks
A disciplined program combines the legacy metrics with the AI-retrieval layer rather than replacing one with the other:
Share of voice and message pull-through for the earned-media layer.
Sentiment for reputation and crisis tracking.
Referral traffic and backlink quality for the site-behavior layer.
Citation Share, tested monthly or quarterly across a fixed prompt set, for the AI-answer layer.
Conversion, where the campaign objective supports it, matched against coverage and citation windows.
The tooling has consolidated accordingly. Traditional media monitoring and coverage analytics now run through platforms like Muck Rack, Meltwater, and Onclusive; the newer Citation Share layer runs through a distinct category of AI-visibility tools, including Profound, Otterly.ai, Peec AI, and BrightEdge. None of these tools produce a finished report on their own; a disciplined team still codes, weights, and interprets the corpus each one generates.
Common measurement mistakes, then and now
The errors that undermined AVE-era reporting have direct modern equivalents: treating a single reading as a trend, measuring volume without a competitive benchmark, and reporting a number without methodology notes attached. The Citation Share-specific version of the same mistake is testing only one AI engine when each indexes different sources and produces materially different answers, or building a prompt set too narrow to reflect how buyers actually ask.
Is share of voice still a valid PR metric in 2026?
Yes, as a comparative measure of earned-media coverage against named competitors. It's incomplete on its own because it doesn't capture whether a buyer researching via an AI engine ever encounters that coverage, which is what Citation Share measures.
What replaced Advertising Value Equivalent (AVE)?
The 2010 Barcelona Principles named AVE invalid and pushed the industry toward share of voice, message pull-through, and sentiment, formalized further in AMEC's Integrated Evaluation Framework.
What tools do PR teams use for measurement in 2026?
Coverage and earned-media analytics run through platforms like Muck Rack, Meltwater, and Onclusive. AI-citation tracking, Citation Share, runs through a separate category including Profound, Otterly.ai, Peec AI, and BrightEdge.
How often should Citation Share be measured?
Monthly or quarterly, using the same prompt set and engines each time. A single reading is a snapshot; Citation Share is only meaningful as a trend.
Written by
EPR Editorial Team
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
