Media monitoring
- Follows a whole topic.
Brand monitoring without the noise
Brand monitoring means watching one thing very closely, your own name.
That includes the founder who gives interviews and every product with its own name.
The mentions sit in press stories and Reddit threads, and lately in the answers AI assistants give. Finding mentions is easy.
Finding only yours takes work. Pocket PR runs it as part of media monitoring in the Growth plan, from $890 a month.
Illustration
Brand monitoring tools and their mention caps
- 30,000new mentions a month on Awario's $29 Starter plan, across 3 topicsawario.com
- 2,000mentions a month is the cap on Brand24's entry planbrand24.com
What is brand monitoring
Brand monitoring is the narrow cousin of media monitoring, built around your own name.
For a startup its value is early warning, because a first critical review or a wrong AI answer is small when it appears and grows when nobody answers.
Brand monitoring
- Asks who named you and whether the mention needs an answer.
Why tools catch the wrong mentions
Most tools match text, so a name that is also an ordinary word matches every use of that word.
Brands called Mention or Pocket share their names with everyday words.
Left raw, such feeds fill with junk, and on plans that cap monthly mentions the junk uses up the quota before real mentions arrive.
Copy and send
How to build a brand passport
A brand passport is a short document that tells the monitoring system what your name looks like in the wild.
It is the single biggest fix for noise, and it takes about an hour.
Here is an illustrative passport for a fictional company called Gridwise.
Brand passport, Gridwise NAME VARIANTS. Spellings, spacing, transliteration Gridwise, Grid Wise, the name in Cyrillic ANCHOR. A signal that proves it is you gridwise.eu, a founder's full name PRODUCTS. Product and feature names Gridwise Clinic Unit PEOPLE. Founders and spokespeople Olena Marchenko NEIGHBOUR WORDS. Words that often sit near your real mentions backup power, clinic, battery EXCLUDED WORDS. Contexts that are never you smart grid policy, electricity grid news
A match counts only with the name plus an anchor or neighbour word and no excluded word. Review the passport after week one and at each launch or new spokesperson.
How to cut false positives
False positives fall into four groups, each with its own fix.
A person reads every match in the first week and turns each wrong one into a rule.
After that, a weekly spot check is enough for most companies.
Homonyms
- The name is a common word. Fix with anchors and excluded words.
Namesakes
- Another company shares your name. Exclude its domain and city.
Reprints
- One story copied to aggregators. Group by body text, count the original once.
Self-mentions
- Your own posts and job ads. Exclude your domains or list them separately.
How far to trust sentiment scores
Accurate enough to sort, not to decide. Sarcasm reads as praise, and a killer feature reads as hostile.
An outage report that lists several providers can score as an attack on each of them.
Let the model sort the negative pile first and a person write the label that goes into a report.
When should a brand mention trigger an alert
Most mentions can wait for the morning digest.
An alert should interrupt someone only when delay would cost money or trust, and too many alerts teach a team to ignore them.
Four thresholds cover almost every case.
- A story about you in an outlet on your own priority list.
- A negative mention with a factual error, in any outlet with a real audience.
- Your company named, or left out, in an AI answer to one of your key buyer questions.
- A rival's press release on a newswire that names you or your category.
How to respond to a negative brand mention
A short playbook keeps the first answer calm and correct.
If several outlets start asking for comment, it has become a crisis and needs the crisis plan, not this list.
Read the whole thing
Who wrote it, how far it spread.
Sort it
A factual error or a customer complaint needs an answer. An opinion usually does not.
Fix facts privately first
Send the editor the correct figure and a source.
Answer customers in public, once
Acknowledge, then move to a direct channel.
Leave opinions alone
Arguing doubles the audience.
Log it
The same complaint three times is a product issue.
Should you monitor your founder's name as well as the brand
Yes, and often first. At an early company the reporter quotes a person. AI answers then borrow that person's phrasing to describe the product.
How to track a spokesperson
- Full name plus the company as the anchor. Add their personal site and LinkedIn profile as sources.
Keep an on-record list
- What each person said on record, so the next quote does not contradict the last.
How to monitor brand mentions in AI answers
Assistants never alert you when they mention you, so the only way to know is to ask them on a schedule.
The full audit method is on AI search visibility, and the free brand mentions check runs a short version for you and a rival.
Write ten buyer questions
What a buyer in your category would ask.
Run them weekly
In ChatGPT first, then in whichever other assistants your buyers use.
Record three things
Whether you are named, how you are described, which pages are cited.
Fix the source page
Correct the page the answer cites, or earn coverage on sites it trusts.
From brand passport to a clean feed
A brand passport before the first alert
We write down every spelling of your name and the look-alike words to exclude. You approve that list before anything is collected.
A person checks the first week
For the first seven days every match is read by hand, and each false hit becomes a new rule, so the feed is clean by week two.
AI answers read every month
We put your buyers' questions to the AI assistants they use and note whether you are named. If you are, we record which page the answer leaned on.
When the brand name is an ordinary word
A company whose name is a common word
- Situation
- A fictional analytics startup called Signal set up free alerts and received about 400 matches a day, almost all of them about phones and traffic lights.
- What we do
- The passport kept a match only when it also carried the domain or a founder's name. Traffic and messaging contexts were excluded outright.
- Outcome
- Daily matches dropped to a handful, and the one forum thread comparing Signal with a rival was answered the same morning.
Illustrative example, not a client result
Anastasiia Bratkova
Chief Brand, Marketing & Communications Officer, RSE
Signs every pitch
Questions
Before you set up the first alert
What is brand monitoring?
Brand monitoring is tracking where your company and its people are named online, AI answers included, so you can correct a mistake while it is still small.
What is the best free way to monitor brand mentions?
Set a Google Alert for your name in quotes plus one anchor word. Search Reddit once a week, and once a month ask two AI assistants about your category.
How do I monitor my brand in ChatGPT?
Write ten questions your buyers would ask and run them in ChatGPT every week. Note whether you are named and which sources the answer cites, then repeat the set on Perplexity if your buyers use it.
How accurate is sentiment analysis?
It handles clear praise and clear anger well. Sarcasm trips it up, and so does neutral news about a bad event. Treat it as a sorting aid, not a verdict.
Should I reply to every negative mention?
No. Reply when the post has an audience and either gets a fact wrong or describes a real customer problem. Silence is fine for a one-off rant nobody reads.
Read next
Related guides
Next step
Get a 15-minute PR diagnosis
Tell us where you are. A senior PR expert replies within one business day.