PushOwl

A merchant installed PushOwl and started the trial. Before making a campaign, they still had questions: Are the features I need included? Will I be charged if I forget to cancel? What should I make first?

There was a 60% drop-off between starting a trial and creating a first campaign. Interviews put words to that gap. Merchants worried about unexpected charges and feature access; unclear billing led to refunds. The trial offered little help when someone hesitated.

Access to the product was only the beginning. We needed to make the next step clear enough to take.

“I installed the trial but I'm scared to get charged if I forget to cancel…”

“Does the trial include all features or just basic ones?”

Shopify order count predicted trial behavior better than store age or traffic. We already had that information. I used it to decide which help to put first.

The four paths changed the sequence of the trial: a starting point for a first campaign, room to explore independently, or an offer to handle setup. Each path began with the help most relevant to that store's order range.

• 0–100 orders: Start with an AI bundle to help make the first campaign.
• 101–2k orders: Start with the standard trial. Offer a discount if the merchant stalls.
• 2k–10k orders: Start with self-serve and AI help in the campaign editor. Introduce managed help after the first useful result.
• 10k+ orders: Lead with managed setup, while keeping self-serve available.

For the smallest stores, opening the campaign editor exposed another job: come up with the idea, write the words, and make the creative. A blank canvas can look simple while asking a lot. AI templates gave merchants something to start from.

In the smallest cohort, those templates increased day-two campaign sends by 42%. More merchants reached an actual send within the first 48 hours.

The invitation needed the same care. Copy tailored to each order range increased trial starts by 19%. The invitation made the trial feel relevant; the template helped someone use it. Both moments mattered, and each needed its own design decision.

We tested putting the discount upfront. Trial value fell 18%. The offer was easier to accept, but it weakened the measure we needed to protect.

We rolled that version back. Merchants began with the standard trial, and the discounted offer appeared after signs that they had stalled. That gave them a chance to explore before we introduced a second offer.

The discount needed a place in the sequence. Moving it to the beginning had changed what the trial was worth.

For larger stores, the next useful step could be handing setup to the team. Merchants were already clicking through from the dashboard to the managed-service page. But that page opened with features, leaving the reason to trust us further down.

I changed the order: the merchant's problem, proof we could help, then how the service worked. We removed pricing from the opening pitch and changed "Book a call" to "Schedule consultation." The page now explained what the conversation was for before asking someone to make time for it.

Among qualified visitors from stores with 10,000 or more orders, bookings rose from 3.2% to 12%. Managed customers also showed about $350 higher average revenue per user and roughly 40% lower churn. Those differences describe the managed customer group; they do not establish that the page redesign caused the retention difference.

Across the tested merchant cohorts, the combined trial changes improved revenue and paid conversion while holding trial value per signup steady. That last measure mattered. The upfront discount had already shown us how an easier entry could weaken the trial.

These are results across the changed trial paths. They should be read together with the individual experiments, rather than attributed to a single template, sentence, or offer.

• 35% revenue lift across the tested merchant cohorts.
• 15% higher trial-to-paid conversion, with trial value per signup steady.
• Fewer refund questions after the trial copy made billing expectations explicit.

The useful question became "What does this store need to do next?" It led to specific changes: a template inside the editor, a clearer explanation of managed setup, a discount offered after hesitation. Each removed a different reason to stop.

Next, I would test whether the order ranges still predict where merchants need help. I would track first campaigns and paid conversion within each range, alongside trial value per signup. If more trial starts came with weaker value again, I would revisit the offer.

A useful trial gives a merchant something worth trying, and enough confidence to try it.