An alert is only useful if what it sends you is actually relevant. Too broad, and you're drowning in policies that don't matter. Too narrow, and you're missing things you should be catching. This article walks through how to build alerts that deliver the right results consistently by using an EV charging infrastructure alert as a worked example throughout both approaches.
The core principle: one topic per alert
The most common mistake is trying to capture everything about a broad issue in a single alert. An alert on "Electric Vehicles" that tries to cover every angle (incentives, infrastructure, battery sourcing, fleet policy, emissions standards) will return hundreds of loosely related policies and quickly become noise.
The better approach is to break a broad issue into focused sub-topics and build a separate alert for each. For EV policy, that might look like:
- EV purchase incentives and tax credits
- EV charging infrastructure
- Battery component sourcing and supply chain
- Fleet electrification requirements
- Zero-emission vehicle standards
Each alert stays tight. Together, they give you complete coverage without overwhelming your inbox.
Step 1: Define your scope
Before you open PolicyNote, get clear on exactly what one slice of your issue looks like. Write it down in plain language: What specific type of policy am I trying to catch?
For our example: policies about building, funding, or regulating electric vehicle charging stations and infrastructure.
Step 2: Choose your approach
PolicyNote supports two search modes for alerts: AI-Assisted and Keyword (Boolean). They work differently, and the right choice depends on your situation.
AI-Assisted is the recommended starting point for most alerts. It understands meaning and context, so it finds relevant policies even when bill text uses different language than you do. It requires less upfront work and is easier to maintain over time.
Boolean is the right tool when you need exact, predictable control. For example, when you're tracking a very specific regulatory term, when AI results are consistently returning noise you can't filter out, or when you need to be certain your alert will only fire on precise language.
When in doubt, start with AI-Assisted. You can always rebuild an alert as a Boolean if the results aren't tight enough.
Path A: Building an AI-Assisted Alert
Describe what you're looking for
In the Discover search bar, switch to AI-Assisted and describe your topic the way you'd explain it to a policy colleague. Include context like geography, issue framing, and who's affected if it helps narrow the focus.
For our example:
State and federal bills related to building or funding electric vehicle charging stations and infrastructure
The more specific your description, the more targeted your results. A vague prompt like electric vehicle charging will work, but a fuller description will surface better matches.
Review the results
Scan the results before saving. Ask yourself:
- Are these the kinds of policies I'd actually want to be alerted on?
- Are there obvious irrelevant results appearing repeatedly? If so, add context to your description to exclude them — for example,
...not including utility rate proceedings. - Is the volume reasonable? A well-scoped AI alert should feel manageable, not overwhelming.
Save as an alert
When the results look right, click Save as Alert, give it a descriptive name, and assign it to the appropriate project.
Name it specifically. "EV Charging Infrastructure" is more useful than "Electric Vehicles", especially six months from now when you're managing a full list of alerts.
Path B: Building a Boolean Alert
Identify the phrases policymakers actually use
Legislators and regulators don't always use the same terminology you do. Brainstorm the specific phrases most likely to appear in bill text for your topic.
For EV charging infrastructure:
"electric vehicle charging station""EV charging station""public EV charging""EV charging infrastructure"
Keep phrases specific (two or more words in quotes). Single words and very short phrases tend to appear in too many unrelated contexts to be useful.
Test each phrase individually
Before combining anything, run a Keyword search for each phrase on its own. Filter by active sessions and note how many results each returns.
What you're looking for:
- Too broad (1,000+ results): The phrase is appearing in unrelated contexts. Make it more specific by adding a qualifying word, or pair it with AND to scope it further.
- Too narrow (under 10 results): The phrase may be rarely used in legislation. Consider whether it's worth including.
- In range (roughly 100–600 results across US state and federal): A good signal the phrase is specific enough to be useful.
For higher-volume issue areas, up to 1,000 results can be workable. Over that, you'll be triaging more noise than signal.
Combine into a Boolean string
Once you've validated your phrases, combine them using OR. Since you're capturing synonyms for the same concept, OR is almost always the right operator at this stage.
"electric vehicle charging station" OR "EV charging station" OR "public EV charging" OR "EV charging infrastructure"
If the combined count is still too high, add an AND requirement to anchor results to the right context:
("electric vehicle charging station" OR "EV charging station" OR "public EV charging") AND ("funding" OR "grant" OR "infrastructure")
If specific types of irrelevant results keep appearing, use NOT to exclude them:
"EV charging station" OR "public EV charging" NOT "rate case"
Review and save
Scan a sample of the combined results. If relevancy looks good and volume is reasonable, click Save as Alert, name it specifically, and assign it to the appropriate project.
Maintaining your alerts over time
A saved alert isn't set-and-forget. Legislative language evolves, and an alert that worked well last session may start drifting.
Check your Marked Irrelevant list periodically. If you're consistently dismissing the same kinds of policies, something in your alert is pulling them in. For Boolean alerts, tighten the offending phrase. For AI alerts, add context to your description to exclude that type of result.
If you're missing policies you'd expect to see, for Boolean alerts check whether a phrase has become more common in recent legislation and add it with OR. For AI alerts, try broadening or reframing your description.
If results spike suddenly, a news cycle or session opening can temporarily inflate volume. If it's sustained, consider splitting the alert into two more focused ones.
Edit alert criteria from the Alerts page by clicking the dropdown on the alert name and selecting "Edit Search Criteria." This takes you to Discover with your current query loaded, ready to adjust.
Comments
Article is closed for comments.