Here's what 

A global telecoms operator wanted to know whether it could sell AI agents to small businesses, and if so at what price. We ran the programme in two phases. The first sized the opportunity: 1,703 technology decision-makers across the UK, Germany, Portugal and Turkey, testing candidate agents including the route to purchase and the operator's credibility in a category it had no history in. The second priced what the first had shortlisted: large sample of decision-makers in Germany, Portugal and Greece, with price ladders run across multiple product tiers and levels of service wrap for each of the agents. 

The opportunity was real, but not where the client expected. A curated marketplace scored far above category norms, and the barrier was not scepticism about agents but not knowing where to start. Tasks businesses said cost them most time did also not predict which agents they would buy: appeal followed how specifically an agent named the job it did, not the size of the underlying problem. Pricing then found demand concentrated in the lower tiers, with the premium sitting in the managed service around the product rather than in the product itself. 

So what 

The two phases answered different halves of one question. The first said the space was open and that the operator's brand made a marketplace more credible rather than less, but that use cases had to be described in the customer's own words to be recognised. The second turned that into a commercial shape: a mid-tier volume anchor, a top tier that only earns its premium when fully managed, and onboarding folded into the monthly fee.

Now what 

The portfolio is due to be sequenced, led by the agent with the broadest relevance and the strongest brand permission. Tiers are being rebuilt around jobs to be done rather than feature lists with the mid tier priced as the volume anchor.