Table of contents
When online sentiment monitoring goes live, the hard part is not the tool but the list. A default setup watches the brand name only, and returns competitor ads and your own press releases while missing what matters. To make monitoring useful, layer the keywords, fix the frequency, and turn alert rules into something a person can actually follow.
1. The List Comes Before the Tool
Start with a blank sheet and write out every way a customer might refer to you. There are three patterns: naming you directly, describing the problem, and describing the scenario. Direct mentions are few and precise; problem descriptions are numerous and scattered; scenario descriptions attract irrelevant noise.
Then rank by business importance and keep the top thirty. Too many keywords blur the alerts — two hundred notifications a day and nobody reads them. The value of monitoring is filtering noise, not capturing everything.
2. A Four-Layer Matrix
| Layer | Examples | Purpose | Suggested count |
|---|---|---|---|
| Brand | Name, abbreviation, products, founder | Direct mentions | 5–8 |
| Category | Industry terms plus region | Demand and discussion | 8–12 |
| Competitor | Rival brand names | Share and reputation shifts | 3–5 |
| Risk | Complaint, lawsuit, accident, exposé | Early negative signals | 5–8 |
The four layers need different handling. Brand alerts push instantly, category alerts summarise daily, competitor alerts summarise weekly, risk alerts escalate the moment they appear. Treating all four with one rule is the common design error.
3. Frequency and Thresholds
Set frequency by business volatility. Consumer, food and education categories with heavy discussion should crawl core terms hourly; industrial and B2B services can run every two to four hours. Night-time frequency can drop, but risk-layer terms should run around the clock.
Combine two dimensions rather than relying on volume alone. Volume spikes are measured in absolute counts; sentiment shifts in negative share. Trigger when either holds — for example a daily mention count at three times the normal average, or negative share above thirty percent. Review thresholds quarterly, since seasonality changes the baseline.
4. Who Reads the Daily Report
The structure depends on the reader. If the report goes to communications only, detail is fine. If it goes to management, it must fit one screen. Use a fixed four-part format: volume and change, items needing attention, items resolved, decisions required.
A fixed format builds reading habits; a report that changes layout daily stops being opened by day three. Describe items as facts with sources, without emotional framing — that belongs to the assessment step.
5. FAQ
How many people does this need? One or two in a mid-sized company: one owns the list and report, one owns assessment and external communication.
Are free tools enough? They work to start, with limits on coverage and retention. Once you need historical retrieval, consider a fuller setup.
Must every negative post be actioned? Judge reach and accuracy first. If the claim is false, use platform appeals. If it is true, focus on explanation and correction rather than removal.
Should comments on short video be included? Yes. Comment sections concentrate emotion and are frequently missed.
Related See also our notes on sentiment handling processes and crisis communications.
GrowLab focuses on integrated content marketing, offering brand strategy, website development and search marketing (SEO/GEO). Contact us for an assessment of your own plan.
Need a professional growth plan?
Add our WeChat for a tailored strategy and annual quote.
Contact Us