A vehicle parts and sales dealership running an online parts basket was losing customers at the final step: shoppers filled baskets, then left without paying. The sales team was already improvising with WhatsApp on personal phones, but no one owned the abandoned-basket problem, and lead distribution between departments was causing internal friction. Working with Stitch-AI, the dealership scoped an integration between Stitch-AI's open API and its Go High Level CRM to trigger automated WhatsApp recovery messages the moment a basket was abandoned, while resolving who on the sales floor was accountable for the resulting lead.
The dealership's parts website was generating baskets that never converted, and nobody inside the business had a structured way to follow up on them. At the same time, the sales team's use of WhatsApp was informal: staff had gravitated to messaging customers on their own initiative, not because a manager told them to, but because it was the format customers replied to on their own time. That grassroots adoption solved a communication problem but created two new ones. First, conversations and customer relationships were sitting on personal devices the business did not control, meaning that if a staff member left, the relationship history left with them. Second, there was no shared system for deciding who followed up an abandoned basket, which sales team member owned it, or how a recovered lead got logged, and this surfaced directly in planning discussions as "challenges with lead distribution and departmental conflicts".
The business was also frustrated with the limitations of previous WhatsApp automation attempts. Existing tools offered branded accounts but came with weak delivery performance or risked account bans, and the team specifically flagged that tools like eSendX delivered lower delivery and open rates than the Go High Level system they already relied on for other marketing. That left a gap: the dealership had a CRM it trusted, a channel its own staff had already proven customers would respond to, and no bridge connecting the two around its highest-intent, highest-drop-off event, the abandoned basket.
The agreed approach centered on integrating Stitch-AI's open API directly with the dealership's Go High Level CRM, rather than replacing either system. Paul, from Stitch-AI, detailed the technical feasibility of building this custom integration for approximately £500, with ongoing maintenance built in to keep the two systems synchronized. This meant abandoned-basket events captured in the CRM could trigger outbound WhatsApp messages automatically, without requiring a staff member to notice and act on every drop-off manually.
Central to the design was Stitch-AI's AI user feature, which allows a business to create multiple specialized bots, each with its own prompt, so that different customer scenarios are handled with different logic. For a parts and vehicle sales operation, this meant one flow could be built specifically for basket recovery, distinct from flows handling new inquiries or existing customer support, so a shopper who abandoned a basket received a message tuned to that scenario rather than a generic response. Outbound WhatsApp messaging templates, triggered via the API, were the mechanism for this: reminders and offers could go out proactively to customers who had left items unpurchased, engaging them before the intent to buy cooled.
The approach also had to solve the lead-distribution conflict alongside the technical build. Bots can be configured to automatically assign chats to a human agent based on what the customer says, which gave the dealership a way to route a recovered basket conversation to the right person, rather than leaving it to land wherever a staff member happened to check WhatsApp first. This directly addressed the department-level disagreement over who should own inbound leads generated by the recovery messages.
Both parties agreed the technology alone would not fix the problem. The plan explicitly recognized that success depended heavily on sales team adoption, and specifically on increasing WhatsApp usage as a critical communication channel across the team, not just within the automation itself. That is consistent with the broader pattern the dealership had already lived through: staff had adopted WhatsApp informally because it worked, and the fix now needed to formalize that instinct rather than fight it.
Implementation followed a structured review-then-build sequence. The two parties committed to a thorough system review before finalizing automation scenarios, ensuring the recovery flow was built around how the dealership's sales process actually worked, rather than a generic template.
The build itself had several concrete components:
This logging layer mattered beyond record-keeping. Because every automated and human message sat inside the CRM, the dealership gained data it could use for analytics and performance monitoring, meaning it could see which recovery messages were converting and which sales staff were following through on assigned conversations.
Getting staff to actually work inside this system, rather than reverting to personal WhatsApp, was treated as a first-class implementation task, not an afterthought. The project's own framing was direct: while the technology was critical, the human element, specifically sales team adoption of WhatsApp messaging, was essential to realizing the benefit of the integration. That is why the review process included planning for sales team communication habits explicitly, alongside the technical integration work.
The stated purpose of the build was to reduce manual workload on the sales team, increase customer engagement, and recover sales opportunities that would otherwise be lost when a basket was abandoned. By automating the reminder and offer messages that previously required a staff member to notice a drop-off and manually reach out, the dealership removed the dependency on individual initiative that had previously made basket recovery inconsistent.
The conditional chat-assignment logic gave the business a repeatable answer to the lead-distribution conflict that had been an explicit agenda item going into the project. Rather than an abandoned-basket lead landing with whichever team member happened to be free, or triggering disputes over ownership, the system routed based on the customer's own response, tying the lead to a defined workflow rather than to informal habit.
Centralized logging inside the CRM gave the dealership a durable, business-owned record of every recovery conversation. That is a meaningful shift from the earlier state, where customer relationships and conversation history lived on personal devices and left with staff when they moved on. With interactions captured in the CRM, the dealership could monitor performance and treat basket-recovery messaging as a measurable part of its sales process, not an ad hoc effort.
The project also positioned the dealership for further automation beyond basket recovery. The same integration architecture, scenario-specific bots and API-triggered outbound messaging, was built to be extensible, opening the door to broader use of AI-driven chatbots and automations across customer engagement and sales efficiency more generally. For example, dealerships can now handle Auto Trader leads and integrate with Keyloop or Pinewood DMS using similar logic to ensure no high-intent inquiry is missed.
For automotive parts and sales operations looking at their own abandoned-basket problem, several lessons from this integration carry directly across:
Stitch, the WhatsApp business communication platform, built this kind of integration around a simple principle: WhatsApp should work for a dealership the way email and phone extensions already do, as owned infrastructure the business controls, not a workaround individual staff invented to get around a problem the business hadn't solved yet. This same infrastructure can be used to send automated MOT and service reminders to keep the service department busy. For a parts and sales operation, that means an abandoned basket becomes a structured, assigned, logged conversation, not a lost sale nobody noticed.
