If you're reading this, you probably work in product, marketing, or growth and want to actually figure out what users care about—without wasting days wading through vague docs or sales fluff. Maybe your team just got access to Propensity, or maybe you're cleaning up a messy setup someone else left behind. Either way, you want simple, actionable steps to set up intent signals so you can stop guessing and start seeing which users might actually convert.
Here’s how to do it, with clear steps, some pitfalls to avoid, and a few shortcuts I wish someone had told me sooner.
What’s an intent signal, really?
Let’s keep it simple: intent signals are the actions or behaviors that hint a user is likely to do something you care about—like buy, upgrade, or churn. In Propensity, you set these up so the system can help you spot high-potential users (or those at risk). Not every click or scroll is an intent signal. You’re looking for meaningful stuff—think “started checkout,” not “visited homepage.”
Step 1: Get clear on your goals first
Before you touch any settings, stop and ask: What do I actually want to predict or influence?
- Are you trying to spot users likely to buy soon?
- Do you want to find those about to churn?
- Is product adoption your main headache?
Pick one. If you try to track everything, you’ll end up with noise and confusion. Write down your goal. Share it with anyone else involved. Don’t skip this—it’ll save you hours later.
Pro tip: If your team can’t agree on what an “intent” is, you’re not ready to set up signals.
Step 2: Map out meaningful user actions
Next, make a quick list (yes, on paper or in a doc) of the user actions that probably matter for your goal. These are the things you want Propensity to pay attention to.
- For conversion: “Added to cart,” “Started checkout,” “Viewed pricing.”
- For retention: “Used feature X three times,” “Upgraded account,” “Contacted support.”
- For churn: “Canceled subscription,” “Downgraded plan,” “Stopped logging in.”
Don’t overthink it. Start with 3-5 strong candidates. You can always tweak later.
What to ignore: Don’t bother with vanity actions like “Scrolled 50% down the homepage” unless you know it correlates with your goal. Most of these are digital noise.
Step 3: Check your existing event tracking
Here’s where most setups fall apart. If you’re not already tracking the actions you care about, Propensity can’t pick them up.
- Look in your analytics tool (like Segment, Mixpanel, or GA4): Are your key actions tracked as events already?
- If not, talk to your dev or analytics person. Ask them to set up tracking for the specific actions you mapped in Step 2.
Reality check: Don’t trust the event names you think you set up. Actually check your event stream. I’ve wasted hours on typos and “legacy” events that meant nothing.
Step 4: Connect your data source to Propensity
Assuming your event tracking is sorted, you need to get that data into Propensity.
- Head to Propensity’s Integrations section.
- Connect your data source (Segment, Mixpanel, etc.) by following the prompts. Usually this means pasting in an API key or authenticating with OAuth.
- Pick the specific events you want to sync—don’t just dump everything in.
Pro tip: If you pipe in every event “just in case,” you’ll drown in useless data. Be ruthless about what you import.
Watch out for: Time zone mismatches and weird property names. Propensity’s support docs are decent if you get stuck, but don’t be afraid to ping their support chat if something doesn’t look right.
Step 5: Define your intent signals in Propensity
Now you’re actually in Propensity. Here’s how to set up the signals:
- Go to the "Intent Signals" (or similarly named) section in the dashboard.
- Click “Create New Signal.”
- Choose your event trigger—pick one of the key user actions you mapped and confirmed you’re tracking.
- Set up any filters or conditions. For example:
- “Event = Started Checkout” AND “Plan = Pro”
- “Event = Canceled Subscription” within last 7 days
- Give your intent signal a clear, honest name. No one wants to see “Intent_Signal_3” in a report.
- Save it.
Don’t overcomplicate: You don’t need endless nested logic or wild combinations to start. If you’re not sure, keep it simple—one event, maybe with a date filter.
Step 6: Test your signals (don’t skip this!)
Testing sounds boring, but it’s where you catch the dumb mistakes.
- Pull up a few real user profiles in Propensity.
- Check: Are the signals firing when they should?
- Look for false positives (signals firing on the wrong people) or false negatives (missing likely users).
If something feels off, dig into the event data. Nine times out of ten, it’s a tracking or naming issue, not a Propensity bug.
Pro tip: Ask a peer who didn’t set it up to sanity-check your signals. Fresh eyes catch weird logic or obvious gaps.
Step 7: Put your signals to work
Once your signals are live and tested, use them to drive actual actions.
- Set up notifications or alerts for high-intent users.
- Feed lists to your sales, success, or marketing teams.
- Trigger automated emails or in-app messages if Propensity supports it.
- Make sure someone owns following up—signals are useless if they just sit in a dashboard.
What not to do: Don’t try to automate everything from day one. Start with a simple workflow or one team.
Step 8: Review and iterate (seriously)
No setup is perfect out of the gate. Every product, audience, and team is different.
- Check results weekly. Are the signals surfacing the right people?
- Are you seeing more conversions, fewer churns, or other real outcomes?
- If not, tweak your events or logic. Drop useless signals. Add new ones if you spot patterns in the data.
Pro tip: Document what you change and why. Future you (or your replacement) will thank you.
What works, what doesn’t, and what to ignore
Works: - Focusing on a handful of high-quality signals tied to real outcomes. - Testing with real user data before letting other teams in. - Keeping everything as simple as possible to start.
Doesn’t work: - Tracking every possible event “just in case.” - Relying on out-of-the-box templates without checking if they make sense for your product. - Ignoring feedback from sales or support (they’ll know if the signals are garbage).
Ignore: - Vanity metrics and actions that sound important but don’t move the needle. - Fancy features you don’t have bandwidth to use. Stick to the basics until you prove value.
Keep it simple and keep iterating
Setting up intent signals in Propensity isn’t rocket science, but it’s easy to overthink or overengineer. Start small, pick the actions that actually matter, and focus on testing and real results. You can always tweak and add complexity later—don’t let “perfect” get in the way of “good enough to start.”
If you run into a wall, don’t be shy about asking for help or scrapping what doesn’t work. The best setups are the ones that get used—not the ones that look fancy in a deck.