Most communications teams think they have crisis communication software.
What they usually have is a monitoring dashboard — or a ChatGPT thread.
Both are useful. Neither stops the sentence that creates the crisis.
The pattern: a mention spikes overnight; Legal wants a line by noon; the CEO drafts LinkedIn in the taxi; someone pastes last year’s holding statement. The post goes live. An employee screenshots a contradiction with a 2024 earnings call. Now you manage two problems: the original issue, and what you just published.
Crisis communication software worth buying should cover more than “see the fire” and “draft something fast.” It should cover the moment before publish — when a claim, a tone, or a forgotten commitment can still be fixed.
This guide defines three layers teams confuse: monitoring, response, and pre-publish review. It explains why pre-publish is the missing layer for in-house Comms, who needs to sit in that workflow, and a checklist you can run this week.
The real problem: tools that help everywhere except the publish button
In-house teams typically stitch together:
- Listening tools (Meltwater, Cision, Brandwatch and peers) — excellent at volume, sentiment, and media pickup.
- Alert / incident platforms (Everbridge-class systems) — built for mass notification and operational continuity, not narrative consistency.
- Generic AI — ChatGPT or Copilot for a first draft under time pressure.
What still fails is the last gate:
- Does this claim match what we already said publicly?
- Would Legal, HR, or Investor Relations flag this wording?
- How will employees, customers, journalists, and critics read the same sentence?
- Is this in the CEO’s voice — or a polished generic that will sound fake by lunchtime?
Monitoring tells you something is happening. Response tools help you write faster. Pre-publish review is the layer that asks whether you should send this version.
That is the category gap most “crisis communication software” pages skip.
Layer 1 — Monitoring: see the fire early
What it is
Monitoring (media intelligence, social listening, news alerts) tracks mentions, sentiment shifts, and narrative momentum across press, social, forums, and the open web.
What it is good for
- Early signal when a thread or story is building
- Executive morning briefs with less manual scanning
- Competitive and sector context
- Proof points for post-mortems and board reporting
What it does not do
Monitoring does not validate your draft. It does not know your March interview promise, your HR red lines, or that “no impact on jobs” is a legal landmine. A spike alert at 07:15 does not stop a careless CEO post at 07:40.
Buy monitoring for visibility. Do not expect it to secure the sentence.
Layer 2 — Response: organize the first hour and the draft
What it is
Response software covers crisis cells, playbooks, holding statements, stakeholder maps, message libraries, and increasingly AI-assisted drafting. Some platforms focus on the first hour; others on dark sites and approval trees.
What it is good for
- Roles, thresholds, and “who does what by when”
- Faster first drafts when the clock is running
- Keeping multiple audiences (internal, press, customers) in one plan
- Reducing blank-page panic
What it still misses
Speed without memory creates new risk. An AI draft that sounds confident but contradicts last quarter’s guidance is worse than a slow, accurate line. Response layers often optimize for output — words on a page — not for consistency with company history.
If your stack ends at “generate a statement,” you have accelerated the most dangerous step: publishing under pressure.
Layer 3 — Pre-publish review: stop the wrong sentence
What it is
Pre-publish review is a structured check of outbound language before it leaves the building. Done well, it reads a draft the way a journalist, a lawyer, an employee, and a critic will — then surfaces contradictions, unprovable claims, tone risks, and safer rewordings.
It is not auto-publish. It is not “AI speaks for the brand.” It is a decision aid for the humans who still own the risk.
Why this is the missing layer for in-house Comms
Agencies can still second-pair a draft. Mid-market and corporate in-house teams often cannot — especially nights, weekends, and “CEO posted from their phone” moments.
Pre-publish review closes four gaps monitoring and response leave open:
- Contradiction risk — today’s line vs past statements, filings, interviews, and commitments.
- Claim risk — absolute language (“never,” “no impact,” “undisputed leader”) that Legal cannot defend.
- Perception risk — how different audiences will hear the same words.
- Voice risk — whether the draft sounds like the spokesperson or like a template.
When your CEO is about to post, this layer matters more than another dashboard tile.
How the three layers work together (and where teams over-invest)
| Layer | Primary question | Typical owners | Failure mode if alone |
|---|---|---|---|
| Monitoring | What is being said? | Comms, digital, agencies | You see the fire; you still publish badly |
| Response | What do we say and do first? | Comms, crisis lead, Legal | Fast draft, weak consistency |
| Pre-publish review | Should this version go out? | Comms + Legal + CEO office | Rarely staffed; risk concentrates here |
Many teams over-invest in Layer 1 because it is measurable (alerts, volume, sentiment). Layer 2 gets budget after a scare. Layer 3 stays informal — a Slack ping to Legal if someone remembers.
That informal gate is where reputation incidents are born.
Who needs pre-publish review (not only Comms)
Heads of Communications / Corporate Affairs
Own narrative risk. Need a repeatable gate that works when they are not in the room.
Legal / Compliance
Need claim hygiene and auditability — what was checked, against what sources, before go-live.
Executive assistants and CEO office
Often the last people to see a post. Need a way to run a review without becoming unofficial lawyers.
CEO / executive spokespeople
Need speed and continuity with prior public language. One careless sentence becomes the story.
HR / Internal Comms
Employee reactions often move faster than press. Internal notes leak; external posts land inside first.
A useful rule: anyone who can publish under the company name — or the CEO’s name — needs access to the same review standard.
What “AI crisis communication” should mean (and what it should not)
The phrase AI crisis communication now covers everything from sentiment bots to auto-generated releases. For corporate reputation, three rules apply:
- Humans decide what publishes. AI drafts, flags, and proposes — it should not ship alone. See Cadence’s human-in-the-loop security model.
- Company memory beats generic fluency. A model that does not know your past statements invents confident nonsense.
- Training reflexes matter as much as tooling. Continuous micro-drills beat a yearly tabletop — see why annual crisis drills are not enough (in French).
If a vendor promises to “handle the crisis for you,” treat that as a red flag. Liability still sits with the company.
Cadence’s wedge (without the brochure tone)
Cadence is an AI team for corporate reputation: night watch, drafts in your voice, crisis cell, Slack drills (Popeye), and a Company Brain of what you have said, promised, and banned.
Versus listening suites and alert platforms, the wedge is: pre-publish risk, CEO/exec voice, approval under pressure, company memory, and trained reflexes. See how Cadence works as an AI communications team and how to prepare while nothing is happening. Cadence is early-stage — judge it against your process, not invented benchmarks.
Practical checklist: run this with your team this week
Use this as a 45–60 minute working session. No new software required to start.
1. Map your stack to the three layers
- List every tool used for mentions, alerts, drafting, approval, and publishing.
- Label each: Monitoring / Response / Pre-publish / None.
- Highlight anything that claims to “do crisis” but never touches the draft.
2. Define the publish gate
Write one sentence your leadership will defend:
“No external statement (press, social, CEO channel, customer email on contested topics) goes live without a pre-publish check against known public commitments.”
If that sentence is politically impossible, document why — that gap is your risk.
3. Inventory the five most dangerous claim types
Typical traps: headcount (“no impact on jobs”), pricing promises, safety assurances, market-leadership absolutes, and hard timelines. For each, name the owner who can confirm facts under pressure.
4. Build a one-page contradiction log
Pull 10–20 public statements from the last 24 months (earnings Q&A, CEO interviews, LinkedIn, press room). Tag commitments and red lines. This is the seed of company memory — whether you keep it in a doc or in a product Brain.
5. Name the human chain for after-hours
- Who can approve a CEO post at 22:00 on Sunday?
- Who can freeze scheduled social?
- Who briefs sales / support within the hour?
If the answer is “whoever is awake,” fix that before the next spike.
6. Run one 10-minute micro-drill
Scenario: a journalist DMs a junior colleague for off-the-record confirmation of a rumor. Observe first reflexes; debrief without naming-and-shaming; schedule the next drill within 30 days.
7. Run one real draft through a full pre-publish pass
Take a CEO post, press reply, or announcement. Check contradictions, claims, perception, and voice. Time it — that is your tooling baseline.
Buying criteria for crisis communication software
Ask vendors questions that separate layers:
- Does it check drafts against past statements, or only a style guide?
- Who retains publish authority — humans, or the model?
- Can Legal and Comms share one review trail?
- Does it cover CEO / exec channels, not only the brand account?
- Is training (drills) included, or only dashboards?
- Where does data live, and is it excluded from model training?
- What happens at 2 a.m. — night report, escalation, or silence?
If the demo only shows charts and a draft button, you are still shopping Layers 1–2.
What to do next
If your team already has monitoring and still gets surprised at publish time, you do not need another alert. You need a stronger gate.
- See the product narrative on how Cadence works
- Walk the CEO scenario: when your CEO is about to post
- Build readiness while calm: prepare while nothing is happening
- Subscribe to the Cadence newsletter for lessons from public reputation events
When you want to test the layer that usually sits in someone’s head: try a pre-publish review, or check Cadence pricing if you are mapping budget to the three layers above.
The goal is not louder software. It is fewer sentences you wish you had never sent.