The core trade-off
Every keyword decision is a trade between coverage and noise:
- Too broad ("AI", "growth", "digital") matches everything, floods the feed, and buries the signals that matter.
- Too narrow ("multi-cloud kubernetes cost optimization initiative") rarely matches real-world phrasing, so signals slip past silently.
The failure modes feel different. Broad keywords annoy you visibly. Narrow keywords fail invisibly: the feed looks quiet and you assume nothing is happening. If your alerts feel sparse, suspect narrow keywords before suspecting the data.
Start from how the signal is written, not how you say it internally
Keywords match against real published material: news articles, job posts, earnings transcripts, filings. Write keywords in the language those sources use.
A practical way to calibrate: find two or three articles you wish Salesmotion had flagged, and look at the exact words in them. If the article says "expands European operations" and your keyword is "EMEA go-to-market investment", you will miss it.
Short, concrete phrases beat long descriptive ones. Two or three words that appear verbatim in text ("plant expansion", "Series B", "chief revenue officer") outperform sentence-length descriptions of a theme.
A setup pattern that works
- Begin with 10 to 20 keywords, not 150. A focused set makes it obvious which keywords produce value and which produce noise.
- Cover each theme from two angles. For a digital-transformation theme: "digital transformation" plus the concrete versions ("ERP migration", "modernization program").
- Turn alerts on. Adding keywords does not enable alerting by itself. Check your alert settings after any keyword change, it is the most commonly skipped step.
- Review weekly for the first month. Prune keywords producing irrelevant matches, and note recurring stories your set missed.
- Expand gradually. Add themes once the current set runs clean.
If you prefer wide-net filtering
Some teams intentionally run broad keywords and filter manually, treating the feed as a wide net rather than a precise alarm. That works if you check the feed daily and would rather skim noise than risk a miss. Choose deliberately: wide net plus daily triage, or precise keywords plus trusted alerts. The frustrating middle is broad keywords with alert notifications, which trains you to ignore alerts.
Common mistakes
- Duplicating built-in signal types. Funding rounds, leadership changes, job postings, and earnings events are detected natively. Keywords are for the themes specific to what you sell.
- Keywords about your product instead of your buyer's problem. Match what target accounts publish about themselves, not your category jargon.
- Set and forget. Territories and messaging change quarterly; keyword sets should be reviewed on the same rhythm.
- One keyword set for two different motions. If half the team hunts infrastructure projects and half hunts compliance pain, split the themes rather than blending everything into one noisy set.
Frequently asked questions
How many keywords should I have?
Most well-tuned workspaces run 15 to 40. Below 10 usually means invisible misses; above 60 to 80 usually means nobody reads the matches. Volume is not the goal, verbatim-matchable phrases are.
Why am I getting alerts every few days instead of daily?
Alert frequency follows match volume. Narrow keyword sets on a small account list legitimately produce a few alerts per week. Widen keywords or add accounts if you want a daily rhythm, and see How Often Salesmotion Data Refreshes for the scan cadence behind alerts.
Can keywords be scoped to a specific account?
Keywords apply across your tracked accounts, and signals are always tied to the account they came from. If one account matters far more than the rest, its account page and alert settings give you a focused view without changing the global keyword set.
Do keywords match exact phrases only?
Matching is rooted in the phrase you define, so "plant expansion" catches "plant expansions" but a long descriptive sentence will not match an article that expresses the same idea in different words. When in doubt, prefer the shorter form.