What ShopPad learnt from Yedric support conversations
A vendor case study suggests auditing the task behind each support question. How app teams can separate repeat usage from successful resolution.
ShopPad’s Yedric case study offers a useful question for Shopify app teams: are merchants asking to understand a feature, or simply to finish a task?
The Infinite Options case study says ShopPad’s customer-success director, Annette Powell, reviewed more than 3,000 merchant conversations over six months after adding Yedric. The assistant provides multilingual help within the app. The account is published by the vendor; it is not an independent evaluation of answer accuracy.
Look for the job behind the question
ShopPad describes a shift from teaching concepts such as conditional logic towards helping a merchant configure the result they wanted. Its published screenshot shows the assistant alongside the options editor, with prompts for adding a field or changing where options apply.
For an app team, our recommendation is to audit support conversations by intended outcome. A request about gift wrapping might need a price adjustment, a visible choice or a conditional field. Those are different jobs. Label what the merchant was trying to accomplish, whether the assistant completed it, and whether a person later had to repair the configuration.

Return visits are not the same as resolution
The case study reports that 38% of merchants who spoke to Yedric returned for more than one session over its stated 30-day window. That is a vendor-reported engagement measure. It does not, by itself, establish successful resolution, time saved or a reduction in support costs.
Measure repeat contacts alongside unresolved requests and escalation reasons. Someone returning to ask about a different task is different from someone repeatedly trying to fix the same problem. Sample conversations in the languages your merchants use, and keep the original request visible during review.
Make changes reviewable
Where an assistant can change app settings, show what it proposes to change and give merchants a way to inspect the result. Preserve a human escalation route for unclear requests and unusual configurations. For a practical pilot, choose one common task, compare it with your existing support process, and expand only after checking both completion and recovery from mistakes.
More from AI
See all →
