How to tell if your PR is working

How to tell if your PR is working

PR is working when it changes something the business cares about. For a startup, that may mean more qualified sales conversations, stronger recruiting, greater trust among investors, or better recognition within a category.

Coverage is evidence that the work reached the market. It is not, by itself, evidence of a business result. Measure the chain from outreach to coverage to audience response to business outcome, and decide what success means before the campaign begins.

Measure outcomes, not just coverage

AMEC's Integrated Evaluation Framework separates communications measurement into distinct stages:

1. Activities: The work performed, such as pitches, briefings, and media follow-up.
2. Outputs: What the work produced, including articles, interviews, and other coverage.
3. Out-takes: What the target audience noticed, understood, or considered.
4. Outcomes: Changes in trust, preference, intent, or behavior.
5. Impact: Contributions to organizational goals such as revenue, hiring, fundraising, or reputation.

Each stage answers a different question. Outreach volume shows whether a program is active. Relevant coverage shows whether reporters found the story useful. Branded search, direct traffic, and prospect feedback can show audience response. Qualified leads, stronger recruiting, and influenced deals connect the work to the business.

Do not collapse these stages into one score. A placement can be strong even if it sends little referral traffic, and a search increase after coverage does not prove that coverage caused the increase. Use several forms of evidence and state what each one can support.

Track a small set of useful PR metrics

Choose metrics that match the objective and target audience. A fundraising campaign and a product launch should not use the same scorecard.

Coverage quality needs a written standard. Before outreach starts, name the publications and audiences that matter, the claims you want the coverage to carry, and the action you hope readers take. A mention in a relevant trade publication may be more valuable than a larger outlet with the wrong audience.

Treat impressions, clip counts, and AVE carefully

Estimated reach, clip counts, and social engagement can describe distribution. They become vanity metrics when they are presented without audience relevance or a connection to an objective.

Advertising value equivalency, or AVE, assigns a hypothetical advertising price to earned coverage. Barcelona Principles 4.0 says AVE and similar measures should not be used; communications should instead be evaluated through outcomes and impact.

Be cautious with numbers such as total impressions or "potential reach." Ask how the number was calculated, whether the audiences overlap, and whether it measures an observed action. A large estimate can provide context, but it cannot establish awareness, trust, or revenue.

Measure AI visibility by platform and question

AI visibility is measurable, but no single score represents every answer engine. Record the platform, question set, location or account conditions when relevant, date, brand appearance, cited source, and answer position. Repeat the same protocol on a regular cadence so changes are comparable.

Official tools now provide part of this evidence. Bing Webmaster Tools' AI Performance report reports citations, cited pages, and grounding-query themes across supported Microsoft experiences. Bing states that these metrics do not measure ranking, authority, or the role of a page within an answer. Google Search Console's generative AI report reports visibility in Google's generative Search features, including AI Overviews and AI Mode, for sites that have access to the report.

Keep four measures separate:

• Appearance: The company or product is named in an answer.
• Citation: A company-owned or third-party page is shown as a source.
• Referral: A visitor arrives from an AI product.
• Outcome: The visit or appearance contributes to a lead, applicant, partnership, or other goal.

A citation is not a click, and a mention is not a recommendation. Platform updates, changes in user demand, and changes elsewhere on the web can also move citation counts. Report AI visibility as an observed trend, not as proof that one placement caused an answer.

Build attribution before coverage lands

Attribution improves when several simple records agree. Set a baseline before a campaign, annotate important coverage dates, and review the following sources together:

• Analytics for referral traffic, direct traffic, engaged sessions, and conversions.
• Search data for branded queries and landing-page visibility.
• CRM fields for lead source and influenced opportunities.
• "How did you hear about us?" responses on forms and sales calls.
• Recruiting, investor, and partner notes that name a publication or story.
• A coverage log with publication, date, link, audience, message pull-through, and campaign.

Use tagged links when an outlet permits them, but do not depend on last-click attribution. Someone may read coverage, search the brand later, and convert through another channel. Record direct evidence, such as a prospect naming the article, separately from directional evidence, such as branded search increasing during the same period.

Use 30-, 60-, and 90-day checkpoints

Visible activity should exist before business outcomes do. A provider may reasonably explain that a specific placement or outcome needs more time. It should still be able to show the work completed, the response from the market, and what it is changing as a result.

Require an auditable PR report

• The objective, audience, baseline, target, and reporting period.
• Outreach activity and reporter response, with enough detail to verify progress.
• Coverage links and a qualitative review of relevance, accuracy, prominence, and message pull-through.
• Audience-response and business-outcome data from agreed sources.
• A separate AI-visibility section that names the platforms and query set measured.
• The method, limitations, and next action for every conclusion.