Reputation Crisis Monitoring Checklist for the First Week
A reputation crisis monitoring checklist for days one to seven: what to watch, what to log as evidence, who owns each check and when to escalate or stop.
A reputation crisis monitoring checklist for the first week has four parts: a short list of places to watch, a log that records what you saw and when, a named owner for every check, and thresholds that tell you when to escalate. Free tools are enough for all of it. What decides the outcome is whether the checks are written down and repeated at fixed times, not how many dashboards you open.
Most people in a crisis refresh one platform all day, miss the court record or the AI answer that changes next, and cannot say afterwards what the results looked like on day two. A checklist replaces that anxiety with a routine.
This guide is built around seven days and ends with the part most checklists skip: how to taper the checks so that watching does not become the crisis. It assumes the response itself is handled elsewhere. This is the monitoring layer under it.
- Watch seven surfaces on a schedule: search results, autocomplete, news, social, reviews, AI answers and records. Heavy checks on day one, lighter every day after.
- Log every finding with the date, time, exact URL or query, a screenshot and an archive copy where the page allows it. The log is your evidence and your memory.
- Name one watcher, one recorder and one decider. A solo person can hold two roles but never all three at once.
- Escalate on a threshold you wrote on day one, such as a second independent outlet or a false claim reaching a top result, not on how the news feels.
- From day five, cut the checks to one or two a day. Stop daily checks when two quiet days in a row match your baseline.
Seven surfaces to watch, and the free way to check each one
Each surface has one free method, one thing to record and one reason it matters. Start with these seven and add nothing until all seven are running.
| Surface | Free method | Record | Why it matters |
|---|---|---|---|
| Search results for the name | Private window, signed out, name in quotes and without | Top ten results in order, with screenshot | It is the first thing a client, journalist or employer sees |
| Autocomplete | Type the name and common endings, note each prediction | Exact predictions and the typed prefix | Predictions show what people are already asking |
| News | Google Alerts plus the news tab of two search engines | Outlet, headline, URL, publication time | A second outlet changes a story from a post to a record |
| Social mentions | Each platform's own search on name, handle and key phrase | Post URL, author, reach if visible | Origin and spread decide the response |
| Review platforms | Your own profiles and the platforms where you are listed | Rating, count, new reviews with dates | Attention often spills into reviews |
| AI answers | A fixed set of questions asked to each assistant you care about | Question, assistant, date, full answer | Assistants repeat and sometimes distort what is online |
| Court and regulator records | Public dockets and regulator registers for the places you operate | Case or filing number, date, link | Records outlast headlines and rank well |
Search signed out and in a private window, because personalized results skew what you see. Google's autocomplete policies say predictions are not assertions of fact or opinion, though they may be perceived as such, and that they do not prevent anyone from typing out a full query. They also say predictions that associate potentially disparaging or sensitive terms with named individuals are not allowed, nor are predictions that can be read as accusations of serious malevolent acts where well-established or expert evidence is lacking, and that violating predictions are removed once found. Record what you see before deciding whether to report a prediction.
Related readCrisis Communication to Employees: Who to Tell FirstFor AI answers, do not improvise each day. Write four to six questions once, such as "Who is [name]?" and "What happened with [name] and [topic]?", and ask the same ones in the same assistants at each check. Answers can differ from run to run, so one odd answer is a note, while the same wrong claim in two assistants on two days is a finding.
Set up the log and the owners before the first check
Build the log first, because everything else feeds it. One shared spreadsheet or document is enough. Do not use messaging threads as the log: they lose the entries you need.
- Write the name registerList every form of the name: full name, short name, handle, old spellings, company name and the key phrase the story uses. Every check runs on this list.
- Create the log with fixed columnsDate and time with time zone, surface, finding, exact URL or query, severity (note, watch, escalate), evidence file name, owner, decision taken.
- Take the baselineRun all seven checks once and record them as they stand. Without a baseline you cannot tell later what changed.
- Set three Google AlertsOne on the name in quotes, one on the name plus the story's key phrase, one on the company or handle. Choose the most frequent delivery the options offer for the first week.
- Fix check times and escalation contactsPut the windows in calendars, and write down who is called when a threshold is hit: lawyer, communications lead, board or family contact.
The Google Alerts help page lists the options you can set: frequency, sources, language, region, number of results and recipient. It does not say Alerts catches every matching result, so treat it as one input, not as the monitor.
What to log so a screenshot works as evidence
A finding without a date, a source and a copy is a memory. With them it is evidence that a lawyer, a platform or a journalist can act on. Capture first, then react.
Each entry needs: the date and time with time zone; the exact URL, or for search the exact query and the account state (signed out, private window, country); a full-page screenshot showing the address bar and your device clock; the claim copied exactly; the author or account name as displayed; and the file name of the saved evidence.
Add an archive copy where it works. The Internet Archive's Save Page Now saves a single page, including its images and style sheets, and returns a permanent address. Its help page says it saves only that one page, that some sites prohibit crawling and that some security settings make it fail. A missing archive copy proves nothing. Never archive intimate or sensitive material: keep a screenshot offline in a protected folder.
Related readProduct Recall Reputation: What Regulators PublishKeep the log factual: "post says X, 14:20 UTC, 38 replies", not "people are furious". Anyone reading it later should be able to separate what was observed from what was feared.
Who owns each check
A check with no name next to it does not happen on day four. Assign three roles, even if one person fills two. The watcher runs the checks and writes the raw finding. The recorder captures the evidence and keeps the log consistent. The decider reads the log at fixed times and decides on escalation. The person at the center of the crisis is a poor watcher and recorder, because every finding lands personally.
| Check | Watcher | Backup | Output to the log |
|---|---|---|---|
| Search results and autocomplete | Trusted colleague or assistant | The affected person, once a day only | Ordered list plus screenshot |
| News and Alerts | Communications lead | Assistant | New outlets with times |
| Social mentions | Social or community manager | Colleague | Post URLs, authors, spread notes |
| Reviews | Whoever already runs the profiles | Assistant | Rating and count at each check |
| AI answers | Anyone following the fixed question set | Colleague | Full answer text per assistant |
| Records | Counsel or paralegal | Company secretary | Case or filing numbers |
A private individual can compress this: a calm friend as watcher and recorder, you as decider, a lawyer as records owner if a dispute exists. Give the friend the register and the question set, not your passwords.
Day by day: what to check from day one to day seven
Heavy early, lighter every day, with a written reason for each reduction. These windows are a starting design from experience, not a rule.
Day 1: baseline and three heavy windows
Run all seven checks, record the baseline before anything is published or reported, set the Alerts and write the thresholds. Hold a decider review in the evening and write the decision, even if it is "no escalation yet".
Day 2: confirm the shape of the story
Four windows. Is it one post or a spreading narrative? Check whether new outlets report independently or cite the first, and whether the phrase is moving into reviews, autocomplete or assistant answers.
Day 3: where second-order effects show up
Three windows. Search results and AI answers often lag the news. Repeat the full AI question set, compare the top ten with the baseline, and check the relevant public register if a dispute or authority is involved.
Related readReputation Crisis First Hour: A Checklist in OrderDay 4: look for the stall or the second wave
Three windows. Compare new mentions with day two. A story that has stopped producing outlets is a different case from one reaching new communities. Read the replies, not just the post.
Day 5: reduce, with a reason
Cut to two windows plus Alerts and write why, for example no new independent outlet for 36 hours. If the reason is only tiredness, hand a window to the backup instead.
Day 6: test the quiet
Two windows, with a fresh signed-out search and the AI questions again.
Day 7: write the week-one review
One window, then an hour for the review: what the log shows, which thresholds were hit, what was decided and the week-two schedule. Mark which items need legal, platform or publisher action, which sits outside this checklist.
Thresholds that trigger escalation: a decision table
Write thresholds on day one, when you are calm, and apply them without negotiating on day three. The values below are starting points from experience, not statistics; a private person will usually set them lower than a large company.
| Signal | Starting threshold | Action | Owner |
|---|---|---|---|
| News | A second independent outlet reports the claim | Brief counsel and communications lead; confirm facts | Decider |
| Search results | The story enters the top five for the plain name | Capture, plan what accurate content can sit beside it | Decider |
| Autocomplete | A prediction ties the name to a serious accusation | Capture, then check it against the policy before reporting | Recorder |
| Social | A false factual claim appears in three or more unconnected places | Capture, prepare a factual correction, review platform rules | Decider |
| Reviews | A cluster of new reviews from accounts with no history within a day | Capture, report each to the platform with the reason | Review owner |
| AI answers | The same wrong claim in two assistants on two checks | Capture, use each provider's reporting route | Watcher then decider |
| Records | A filing, complaint or case number naming you or the company | Counsel reads it the same day | Counsel |
| Safety | Home address, family names, threats or intimate images appear | Stop routine monitoring on that item; protect first, involve counsel and police where appropriate | Decider |
The safety row overrides everything else. If a check turns up a home address, a child's school or threats, protect the people involved, keep a minimal record and do not circulate it.
Write downward thresholds too, the signals that let you relax, or a team can only escalate.
Court and regulator records: what free access gives you
Records are the surface people forget, because nothing pings you. In the United States, the federal courts' PACER system charges per page, and its fee page sets out three figures worth knowing before you search.
Source: the PACER fee page of the United States Courts, as read for this guide. Fees can change; state courts and other countries work differently.
The same page says searches are billed by the pages they generate, even when nothing matches, and have no maximum, so a few targeted searches are cheap but an open-ended trawl is not. Court opinions are always free, and spending of $30 or less in a quarter is waived. State courts, regulators and other countries publish their own registers, and counsel will know which apply. Your job is to look at the right register on days three and seven, log the case or filing number and hand it to counsel, not to read a filing and decide what it means.
Related readArrested: What to Do for Your Reputation in Week OneA worked scenario: an invented company, seven days
This is an invented scenario with generic parties. A small online retailer's former customer posts a long thread claiming the founder misled buyers about a recall. Assumptions: a three-person team plus part-time counsel, signed-out checks, thresholds as in the table above.
| Day | What the log shows | Decision |
|---|---|---|
| 1 | Thread on one platform, 40 replies, no news, search results unchanged, assistants silent | Baseline taken, alerts set, no escalation |
| 2 | A niche blog repeats the claim, citing the thread; two new reviews mention the recall | Watch; counsel checks the recall facts |
| 3 | Blog post enters the top ten for the brand name; one assistant paraphrases the claim as fact | Capture; assistant threshold needs two checks, so repeat tomorrow |
| 4 | No new outlet; the same assistant repeats it, a second assistant does not; reviews steady | AI threshold not met in two assistants; prepare a factual notice about the recall on the company site |
| 5 | Thread quiet for a day; blog still ranks eighth | Cut to two windows |
| 6 | No change in top ten; both assistants answer neutrally after the notice is indexed | Hold two windows |
| 7 | Stable on all seven; records show nothing | Week review; one check daily for week two |
The team did not escalate on day three because the threshold asked for two assistants on two checks. The blog was captured before anyone contacted its owner. And the day-five cut was tied to a written signal, not fatigue. None of the numbers describe real cases.
Where this checklist stops and other routines begin
This checklist leaves out jobs with their own routines. The first-hours response, meaning statements, silence and who to tell, starts before this log and the log runs underneath it. Standing brand monitoring is a long-term program a crisis week is the wrong time to design. Detailed Alerts configuration is its own topic. Responding to a backlash is a decision process that monitoring only feeds. Measuring reputation over months starts once there is a quiet baseline.
The Results about you help page says the tool regularly checks whether your name and personal information match results in Search, with optional email notifications. It also says removal hides a result from Search without deleting the information from the web, and that requests involving newspapers and government or educational sites are often denied. It suits contact details and is the wrong tool for a news story. If your address or phone number could be targeted, add it on day one and treat any hit as a safety item.
Related readContent Creator Crisis Response: A Plan That Protects IncomeWhen to stop checking obsessively
Monitoring can become the crisis. Checking every ten minutes degrades decisions and keeps the affected person inside the worst part of the story. The cure is a schedule that shrinks on purpose.
The windows come from the day-by-day plan above, an example from experience and not a measured standard. Day one excludes the baseline run.
Keep Alerts running to cover the hours when no one is looking, let the backup take the evening window, and skip the check before sleep. Stop daily checks only when your downward thresholds are met on two consecutive days: no new independent outlet for 36 hours, the top ten unchanged, and AI answers stable across two runs.
After that, check weekly for a month: search results, the fixed AI questions and records. Reputation Partners offers monthly monitoring of search and AI answers as a separate service, but a calendar reminder and the same log will cover a quiet month. If a threshold fires again, restart from day two, not day one: you already have a baseline.
Next step: copy the log, name three people and set the thresholds
Create the log, write the name register, take the baseline in a private window and set three alerts. Then name a watcher, a recorder and a decider and write your thresholds. If the story is already spreading faster than you can log it, the crisis desk can take the monitoring and the first decisions off your hands. See crisis management.
Questions people ask
How often should I check during a reputation crisis?
Check heavily on days one and two, three or four fixed windows a day, then taper, with a reason written for each reduction. Constant refreshing mostly raises stress, since a story rarely changes in ten minutes. Keep Google Alerts on throughout so the gaps between windows are covered by something that costs you no attention.
What should I screenshot, and what should I not?
Screenshot public posts, articles, search results, reviews and AI answers, with the address bar and the date and time visible, and save the exact URL beside each image. Do not circulate images showing intimate material, a child or a home address. Keep those minimal, offline and protected, and let counsel or the police advise on handling them.
Do Google Alerts catch everything about my name?
No. Google's help page lists the options you can set but does not claim complete coverage, and Alerts only reflects what the service finds. Private groups, closed platforms and unindexed content will not appear. Use Alerts as one input beside manual searches, platform searches and a fixed set of AI questions.
When is it safe to stop monitoring daily?
When your written downward thresholds are met on two consecutive days: for example no new independent outlet for around a day and a half, the top ten unchanged against the previous check, and AI answers stable across two runs. Then check weekly for a month. If an upward threshold fires again, restart the heavier schedule.
Who should do the monitoring if the crisis is about me?
Someone else, where possible. The person at the center sees every finding as personal, which slows decisions and distorts the log. Give the watcher and recorder roles to a calm colleague, assistant or relative, and read their daily summary. If you are alone, set two windows a day and write the finding before you react.
