Intent-Layered Dormant User Reactivation
Score dormant users by why-now signals before sending re-engagement messages
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 7
- Confidence
- 98%
Treat dormant users as heterogeneous rather than as one email list. Gather signals explaining why a person might return now: recent website behavior, external product research, activity from colleagues, or patterns shared with highly active users. Turn those observations into an intent score and separate high-intent users from people who have shown no recent interest. Begin with the lowest-hanging cohort, then match outreach to the behavior that produced its score. A user researching a related category should receive a different message from one whose teammates recently adopted a relevant feature. The team can also map the content and discovery paths followed by its best users and invite dormant users into those journeys. Cohort experiments reveal which combination of signals and messages produces genuine reactivation.
Origin
Extracted from Marketing Against the Grain in response to MongoDB's question about millions of inactive users who sometimes return months or years later.
Core principles
- 01Past sign-up intent does not imply present buying intent
- 02Prioritize users showing a credible reason to act now
- 03Combine first-party behavior with external and account-level signals
- 04Match the message to each cohort's current context
- 05Use successful active-user journeys as reactivation templates
How to run it
- 1
Segment dormancy
Separate users by inactivity duration, prior product depth, role, account, and original use case.
Pro tip Keep the first segmentation simple enough to explain.
Watch out A single dormant list hides large differences in intent.
- 2
Collect why-now signals
Combine website activity, product events, external research, and relevant account behavior.
Pro tip Include recent actions by peers at the same company.
Watch out Respect privacy, consent, and data-use obligations.
- 3
Score current intent
Weight the signals according to how strongly they predict renewed product need.
Pro tip Favor recency and combinations of behavior over isolated weak events.
Watch out Do not mistake any activity for purchase intent.
- 4
Prioritize the easiest cohort
Start with users showing the clearest and most recent evidence of renewed interest.
Pro tip Use this cohort to validate the model before scaling.
- 5
Match the message
Create outreach that reflects the specific behavior, role, or use case behind each cohort's score.
Pro tip Explain the relevant next step instead of merely announcing product features.
Watch out Generic nudges discard the value of intent segmentation.
- 6
Replay successful journeys
Identify the content and discovery paths common among active users and route suitable dormant users through them.
Watch out Correlation between a content path and activation does not automatically prove causation.
- 7
Test cohort lift
Measure return, activation, and production usage against control groups and lower-intent cohorts.
Pro tip Optimize for meaningful product use rather than email clicks.
In the wild
A PLG database company has millions of dormant accounts. Its model identifies 5,000 users who recently visited implementation pages, researched adjacent tools, or work at accounts where colleagues resumed product activity. The lifecycle team divides them by signal and sends use-case-specific messages while withholding generic outreach from users with no recent behavior.
→ The company concentrates effort on users most likely to return and can attribute reactivation lift to specific intent cohorts.
Common mistakes
Blasting the entire dormant base
Uniform outreach treats a user active today like someone who has ignored the product for a year.
Optimizing for clicks
A catchy email may earn engagement without restoring meaningful product usage.
Overcomplicating the first score
An opaque model is difficult to validate; begin with a few interpretable signals and improve it with evidence.
Is it for you?
Best for
It is best for product-led companies with large dormant audiences and enough behavioral data to prioritize outreach.
Not ideal for
It is not ideal for products with tiny user bases, sparse telemetry, or no meaningful repeat-use case.
From the transcript
“the best solution I think here is how do you get the right intent data on your users”
“the why now data basically like why would this person be why should I reach out to this person now”
“let's come up with some type of score and figure out who we should go reach out to”
From the episode
6 Marketing Problems Solved In 53 Minutes With Dave Gerhardt
Dave Gerhardt