LONDON: To design a chatbot, global snacking giant Mondelez had to rethink its approach to investing in innovation, by moving away from long-term investment in one ‘silver bullet’ idea, toward smaller pots of funding and a test-and-learn process of continual iteration.

This is according to Kunal Lohia, ITS Marketing Capability Deployment Lead at Mondelez who told an audience at the Festival of Marketing (London, October) that business must detoxify failure, provided that teams do it fast and learn from it. (For more, subscribers can read WARC’s exclusive report: How Mondelez learned to fail).

"Failure," Lohia said, "gives an invaluable feedback loop." It allows development to progress around a hypothesis, which it then tested to destruction – if it does not work at all, the idea is proved to be wrong, if there is room for improvement, then the idea develops, the team adapts, the product is improved.

This lesson emerged from Lohia’s work on a retailer-facing chatbot that would allow business customers to access content more easily. The restraints of a narrow user base and an abundance of content proved useful.

"Can we implement on this site a chatbot that improves the user experience, and helps those retailers engage with content on the site, and in turn, can it lead to better performance for the retailer and hence us? Can we then build further to get good insights for ourselves and see how people are interacting with our content?"

Using the agile test-and-learn methodology, the development team created a minimum viable product in just two weeks, which was available for both internal and external stakeholders to use.

Firstly, this gave the team huge amounts of insight on how their customers phrased questions and how the terminology retailers used differed from Mondelez. The process also shifted KPIs: originally, the purpose of the chatbot was to reduce call-centre phone time, but this measure was flipped as the team realised the importance of engagement to the consumer conversation.

Agile working brought a lot to Mondelez's development team, lessons applicable to other areas of the business, Lohia posited. Crucially, being agile requires you to back your own ideas, not only with data, but through experimentation to prove the hypothesis and to test the destruction.

In addition, the fact of the minimum viable product was crucial for sparking the imaginations of the business team, who became far more engaged with an extant product rather than ensuring no money was being wasted in the development of a larger project. The result, Lohia explained, was a better product for consumers and for Mondelez.

Sourced from WARC