Before this engagement, the subscription service was facing a genuinely urgent conversion problem:
With paid acquisition costs rising steadily, leadership recognized that improving trial-to-paid conversion had become genuinely more valuable than acquiring additional trial volume at the same weak conversion rate -- the real opportunity was extracting more value from users the company was already paying to acquire. Previous attempts to address this had focused primarily on tweaking the discount offered at trial's end, without meaningfully examining what was actually happening -- or not happening -- earlier in the trial experience itself.
We designed and implemented a trial-lifecycle Iterable program built around genuine activation and engagement signals rather than a fixed trial-length countdown:
Rather than treating every trial user identically on a fixed countdown to trial expiration, this structure let the messaging genuinely adapt to what each specific user had actually done -- meeting an unengaged user with simplified activation guidance, and meeting an already-engaged user with deeper value reinforcement and a well-timed conversion prompt, rather than the same generic "trial ending soon" message sent to everyone regardless of their actual trial experience.
This distinction between an uninterested user and one who simply hadn't found the right entry point turned out to be genuinely important. Historical data review revealed that a meaningful share of users who never activated had actually opened and engaged with early onboarding emails, suggesting real initial interest that stalled at a specific friction point rather than genuine disinterest from the start. Identifying this distinction changed how the team approached non-activated users -- shifting from an assumption that they simply weren't a good fit toward specifically investigating and addressing the concrete friction point causing stall-out.
This targeting discipline mattered directly to program economics. Offering the deepest discount to every trial user regardless of engagement would have meant subsidizing conversions that would likely have happened anyway from highly engaged users, while potentially still failing to convert genuinely unengaged users for whom price was never the actual barrier. Concentrating incentive spend specifically on the engaged-but-hesitant segment meant the program's cost per incremental conversion stayed meaningfully lower than a blanket incentive approach would have produced.
Within five months of full implementation, once the engagement scoring model had enough historical data to reliably predict conversion likelihood and the coordinated email-push journey had been refined through several testing cycles:
Analysis of historical trial data revealed a pattern the company hadn't previously acted on systematically: trial users who selected and customized their first meal plan within 24 hours of signup converted to paid at a substantially higher rate than those who took longer or never completed that step. This made the onboarding journey's decision-split logic genuinely consequential -- rather than treating every new trial user identically, the journey specifically detected whether that critical first activation had happened, and responded with more direct, simplified guidance for users who hadn't yet completed it. This early intervention, targeted precisely at the highest-leverage moment in the trial lifecycle, was directly responsible for a meaningful share of the overall conversion improvement, and gave the team clear, evidence-based justification for where to focus limited engineering and content resources going forward.
Before this engagement, marketing and product teams held genuinely different theories about why trial conversion was weak -- marketing believed pricing and incentive structure were the primary barrier, while product believed the onboarding experience itself was too complicated, causing users to disengage before ever reaching genuine value. The engagement scoring analysis, built on actual historical behavior data rather than either team's assumption, showed both were partially right but in different segments -- users who activated quickly and engaged deeply were price-sensitive at the conversion moment specifically, while users who never activated were experiencing a genuine onboarding friction problem unrelated to price. This data-driven resolution let both teams stop debating from assumption and instead collaborate on two genuinely distinct interventions matched to two genuinely distinct problems, rather than either team pushing a single company-wide fix based on an incomplete picture.
Prior to this engagement, push notifications and email operated as genuinely separate programs, often owned by different team members with no shared calendar or coordination. This meant a trial user could plausibly receive an email and a push notification covering similar content within the same day, or conversely, go several days without any meaningful touchpoint if both channel owners assumed the other was handling outreach for that period. Bringing both channels into a single coordinated Iterable journey, with explicit logic determining which channel handled which type of message, closed this coordination gap directly. Push handled genuinely time-sensitive, action-oriented moments; email carried the richer value-reinforcement content trial users needed to understand why continuing past the trial was worth paying for. This division of labor, decided deliberately rather than falling out of whichever team happened to own which channel, was a meaningful structural improvement independent of any specific message content.
A predictive engagement scoring model is only as valuable as the team's confidence in acting on it, and confidence depends on the model being genuinely explainable, not a black box. Rather than building an opaque scoring system, the team specifically documented which behaviors contributed to the score and why, based on the historical correlation analysis, giving both marketing and product stakeholders a clear, shared understanding of what the score actually represented. This mattered practically: when a specific trial user's score seemed to disagree with intuition, the team could trace back to the specific contributing behaviors rather than simply trusting an unexplainable number, and this transparency made stakeholders genuinely willing to act on the scoring in ways they likely wouldn't have trusted a fully opaque system to justify.
The broader outcome leadership took from this engagement extended beyond the specific conversion numbers -- it was confidence that trial performance could be genuinely understood and deliberately improved through data, rather than remaining a source of ongoing cross-team disagreement resolved by whoever argued most persuasively in a given meeting.
Our team of experts is here to help you achieve measurable, lasting results.