An install gave us a signup. The merchant still needed a campaign.
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?”
One useful signal: how many orders has this store had?
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.
Four starting paths
Path 01 · 0–100 orders
A new store still has to decide what its first campaign should say.
Start with an AI bundle and a first-campaign checklist.
A starting point gives the merchant something to edit before asking them to make a campaign from a blank page.
In this smallest cohort, AI templates increased day-two campaign sends by 42%.
The editor was empty. So was the campaign.
For the smallest stores, opening the editor still meant inventing an idea, writing the words, and making the creative. A blank canvas asked them to do all three at once.
The decision
Give the merchant an AI template to edit, and make the trial invitation speak to their order range.
Why this changed
The template offered a first campaign inside the editor. The tailored invitation helped the trial feel relevant before the merchant entered it. These were different moments in the journey, so their results should be read separately.
Observed in separate tests
In the 0–100 order cohort, AI templates increased day-two campaign sends by 42%. Order-tailored invitation copy increased trial starts by 19%.
The cheaper start cost us 18% in trial value
We tried making the trial easier to accept by putting the discount first. That also changed what each new trial was worth.
The decision
Keep the standard trial at entry. Introduce the discount only after signs that a merchant has stalled.
Upfront discount: trial value fell 18% in the test.
Standard trial first; the discount moves later in the sequence. We rolled back the upfront version.
Why this changed
A merchant gets time to explore the product before seeing another offer. The failed test showed why a higher trial-start count could not be the only measure of success.
What the evidence can say
The 18% drop belongs to the upfront-discount test. No separate lift is claimed for the delayed offer.
Before “Book a call,” give someone a reason to trust you
Larger stores were already reaching the managed-service offer. The earlier page gave prominent space to features and price before clarifying what a conversation would do for the merchant.
The decision
Lead with the store's problem and proof, then explain the service and the purpose of a consultation.
These artifacts show the direction of the offer work, not matched production captures of the booking experiment.
Why this changed
The live booking route also removed price from the opening pitch and changed “Book a call” to “Schedule consultation.” The exploration above shows the same effort to make proof and the conversation's purpose easier to find; its CTA is not the final booking copy.
Observed result and limit
Among qualified visitors from stores with 10,000 or more orders, bookings rose from 3.2% to 12%. Managed customers also had about $350 higher average revenue per user and roughly 40% lower churn. Those group differences do not establish that the page redesign caused the retention difference.
More paid conversions, with the value of each signup intact
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.
Design the next step around the store taking it
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.