Scale Smarter
Case study · AI product & growth

Building an AI product capability that improved customer outcomes and revenue

I helped Tidio turn ecommerce product data into a scalable capability for its AI Agent — connecting customer discovery, integrations, product experience and monetisation.

  • +25%

    Metered AI revenue

  • +25%

    AI conversation resolution

  • +12%

    WordPress and WooCommerce activation

  • AI product
  • Product strategy
  • Monetisation
  • Integrations
  • Ecommerce
Company
Tidio
Product
Customer service platform spanning live chat, helpdesk and an AI Agent
Market
B2B and B2B2C SaaS
My role
Lead Product Manager, responsible for the Connect product area across two squads
Scope
AI product discovery, integrations, APIs, automation workflows, monetisation and Product Operations
The challenge

Tidio was becoming AI-first, but the AI Agent needed deeper product knowledge

The AI Agent was already handling service conversations well. The harder problem was commercial: ecommerce shoppers were arriving with buying questions, and the assistant had no dependable way to understand a merchant’s catalogue or recommend the right product inside the conversation.

  • The AI Agent could answer service questions, but could not reliably recommend the right product.
  • Ecommerce product data sat outside the conversation, in catalogues the assistant could not read.
  • Shopify was the obvious first surface, but the opportunity was far wider than one platform.
  • AI usage was metered, so better answers and commercial performance were directly connected.
  • Adoption depended on integrations quality as much as on model quality.
My diagnosis

This was not simply an AI feature or a Shopify integration

Four problems had to be solved together. Solving any one of them alone would have produced a demo rather than a capability.

  1. 01

    Customer intent

    Shoppers were not asking service questions. They were asking for help choosing — size, fit, availability, comparison and suitability.

  2. 02

    Data ingestion

    Recommendations were only as good as the product data reaching the assistant, so ingestion had to be dependable, current and structured.

  3. 03

    Experience design

    The assistant needed to present products inside the conversation in a way that felt useful rather than promotional.

  4. 04

    Commercial design

    Because usage was metered, deeper capability had to translate into measurable revenue rather than uncosted consumption.

What I did

From conversation research to a scalable recommendation capability

01

Mapped real product-discovery conversations

I reviewed live conversations across ecommerce accounts to separate service intent from buying intent, and grouped the recurring discovery questions the assistant was failing to answer. That research set the definition of a good recommendation before a line of it was built.

02

Shaped the end-to-end product experience

I defined how products were ingested, matched and surfaced inside the conversation — including how the assistant handled partial data, out-of-stock items and low-confidence matches, so it stayed trustworthy at the edges.

03

Launched through Shopify, then designed for broader compatibility

Shopify gave the fastest route to real usage and feedback. In parallel I specified an API-based ingestion path so WordPress, WooCommerce and non-standard catalogues could reach the same capability without a bespoke build per platform.

04

Connected adoption to monetisation

I worked with commercial and analytics partners to align capability, metering and packaging — so that stronger resolution drove metered AI revenue instead of unpriced usage, and pricing conversations followed real value delivered.

05

Strengthened the wider integrations strategy

I prioritised the integrations roadmap around platforms that unlocked data the AI Agent needed, and set up Product Operations practices so two squads could ship against a shared definition of quality.

How the work connected

Four decisions treated as one system

  • Customer understanding

    Conversation research defined what a useful recommendation actually looked like for shoppers and merchants.

  • Product capability

    Ingestion, matching and in-conversation presentation turned that understanding into a repeatable capability.

  • Ecosystem strategy

    Platform and API decisions widened reach beyond Shopify without multiplying the build.

  • Commercial model

    Metering and packaging tied improved outcomes directly to revenue rather than cost.

Results

Better answers, wider reach and measurable revenue

  • +25%

    AI conversation resolution

  • +25%

    Metered AI revenue

  • +12%

    WordPress and WooCommerce activation

  • 8

    Languages supported

  • API

    Broader platform reach through API ingestion

Additional product impact

Work that supported the same outcome

  • Delivered three messaging channels within five months alongside the AI work.
  • Improved integration setup completion by removing avoidable steps in connection flows.
  • Reduced support load on integration issues through clearer error and state handling.
  • Established shared discovery and prioritisation rituals across two squads.
  • Created reporting that made AI quality and AI revenue visible in the same view.
Relevant capabilities
  • AI product strategy
  • Product discovery
  • Ecommerce integrations
  • APIs and data ingestion
  • Product-led growth
  • Monetisation
  • Platform strategy
  • Product Operations
  • Cross-functional leadership

Need to turn an AI or integration opportunity into measurable product value?

I help SaaS businesses connect customer needs, product strategy and commercial outcomes — without treating them as separate workstreams.

Discuss a projectExplore Fractional Product & Growth