Zicy for PR & Communications Teams
What this guide is for, in one sentence: How a PR or comms team uses Zicy to monitor, correct, and prove the narrative AI tells about a brand — the measurement layer earned media has been missing since AVE stopped being defensible.
Who this is for: In-house comms leads and PR agency teams whose job is the story — and who’ve realized the most influential re-teller of that story is no longer a journalist or a search page, but the AI answer itself.
The problem you’re living
Section titled “The problem you’re living”Reputation is now shaped less by what people search than by what AI answers — and every article, interview, and expert quote you place becomes input to those answers. Guides by role Earned media isn’t just coverage anymore; it’s model food. That cuts both ways: your best placements compound, and so do the errors. An outdated claim or an outright hallucination, once absorbed, is sticky — it resurfaces confidently, engine after engine, and most teams discover it only when it walks into a crisis. The cost of an uncorrected error compounds with time.
And the measurement problem sits underneath all of it: the industry agrees AI answers are PR’s new mandate, but most teams have no way to demonstrate that a campaign changed what AI says. AVE is dead; nothing standard has replaced it. That replacement is, concretely, what the workflows below produce.
How the loop maps to your job
Section titled “How the loop maps to your job”- Measure — reputation monitoring per engine: the exact wording AI uses, and which sources it leans on.
- Diagnose — the reputation-risk audit: hallucinations, misaligned sentiment, missing associations.
- Act — corrections at the source, plus source-backed thought leadership.
- Prove — before/after AI-answer snapshots per engine: the earned-media proof that replaces AVE.
The worked example through the module articles is MenuPilot, whose PR lead Priya runs exactly this loop — including catching AI confidently asserting a wrong founding year to anyone who asked.
Your four core workflows
Section titled “Your four core workflows”1. Monitor the questions that carry risk (Measure). Build a reputational prompt set alongside the commercial one — “is X reliable,” “X controversy,” “X vs alternatives” — via custom prompts (How to Write Tracked Prompts That Actually Work). The per-prompt view (Per-Prompt Results) shows each engine’s full answer and the sources it drew on. That sources view is your targeting map: it tells you which outlets move which engine, which turns your pitch list from habit into strategy.
2. Audit the record (Diagnose). Brand Intelligence is your reputation-risk audit: what four AI engines say about you, checked against your profile and website, with what is wrong turned into fixes you can add to your fix list or dismiss. Read it with the Brand Sentiment panel on the dashboard (the words AI reaches for, positive and negative) and Competitor Sentiment Profiles (the framing every rival earns — core comms intelligence you used to assemble by hand). A negative descriptor that’s really a wrong fact — “expensive” built on phantom pricing — is a PR problem with an engineering-simple fix.
3. Correct and author (Act). Fix at the source: confirmed gaps flow to a fix list; Take Action → Schema Generator and LLMs Generator generate the schema and llms.txt that state your verified facts where machines look, and Content → Content Generator drafts thought leadership with every claim cited — pitchable precisely because the receipts are attached. The discipline that keeps you safe: nothing publishes unreviewed; the tool drafts, your team owns the byline.
4. Prove the shift (Prove). Your before/after is native here: Brand Intelligence results (engines correct, facts right, fixes done) and sentiment by period, per-engine answer snapshots, citation mix (Citation Analysis — did the placements you earned become sources AI actually cites?), and downstream movement in Site Traffic. “Our campaign moved AI’s answer on these five questions, on these engines, and here’s the facts-right count before and after” — that’s the sentence the industry has been trying to buy since AVE died. Claim it.
Honest guidance — cadence, not crisis alarm. Tracking runs on a daily cadence now — the dashboard’s Performance Trends panel lets you view it daily, weekly, or monthly, with event markers you can drop on the day a placement ran — but it’s the strategic layer, not a real-time siren. For crisis-adjacent accounts, pair it with your live monitoring stack and use Zicy to verify — after the fire drill — what actually stuck in the answers.
Honest guidance — the stickiness clock is real. Because errors embed deeper the longer they circulate, the audit isn’t a someday task. Run it first, fix the confirmed facts first, and only then spend on new narrative. Correcting the record is the cheapest campaign you’ll run this year.
Your first month
Section titled “Your first month”Week 1 — Audit. Profile + reputational prompt set live; first analyses complete. Review your Brand Intelligence check-up and triage every fix — add it to the fix list or dismiss it. This is your before-state; save it. Week 2 — Map. Baseline the dashboard’s Brand Sentiment panel and Competitor Sentiment Profiles; read Citation Analysis to learn which sources each engine trusts in your category — your new media list. Week 3 — Correct and author. Ship the fact corrections (Take Action → Schema Generator and LLMs Generator); draft and review one source-backed thought-leadership piece aimed at your weakest engine’s trusted outlets. Week 4 — The new report. Re-run the audit, snapshot the deltas, and build your first AI-narrative report: accuracy score, sentiment movement, citation mix, answers before/after. Name it what it is — your AVE replacement — and make it monthly.
What to read next
Section titled “What to read next”- Brand Intelligence — the audit; start here
- The dashboard’s Brand Sentiment panel and Competitor Sentiment Profiles — your tone and framing map
- Citation Analysis — which sources move which engine
- Take Action and Content — corrections and citable authorship (Schema Generator, LLMs Generator, Content Generator)