A consultant is building a custom Copilot Studio topic that lets prospects on a self-service portal ask configuration-dependent pricing questions, such as how price changes when they add optional modules. The pricing logic depends on a multi-step calculation involving base price, module surcharges, and volume tiers that live in a connected pricing engine, not in a simple lookup table. What is the most effective way to structure this topic?
Select an answer to reveal the explanation.
Short Explanation
When the real math lives somewhere else, in this case a whole pricing engine doing multi-step calculations, the smart move is to have the conversation simply ask that engine for the answer rather than trying to recreate its logic from scratch inside the chat flow. Rebuilding the calculation a second time inside the topic means now there are two places that could disagree with each other the moment pricing rules change, and one of them will quietly go stale. Telling every prospect to go find a human rep the moment modules come up throws away the entire reason for building a self-service pricing topic. And baking in a handful of preset combinations works fine right up until a prospect picks a configuration nobody thought to hard-code, at which point it just gives a wrong or missing answer. Calling out to the system that already owns the logic keeps one source of truth and lets the conversation stay accurate as pricing evolves.
Full Explanation
The correct answer is B. Since the pricing logic is genuinely multi-step and lives in a connected pricing engine, the topic should call that engine through an action and surface its result, keeping the calculation logic in the single system designed to own it and ensuring answers stay accurate as pricing rules change. Option A is incorrect because reimplementing base price, surcharge, and volume tier logic inside the topic duplicates business logic in a second place, creating a maintenance burden and a real risk that the two calculations drift out of sync over time. Option C is incorrect because it abandons the stated goal of self-service pricing answers, pushing every configuration question back to a human rep and defeating the purpose of building the topic at all. Option D is incorrect because hard-coding a limited set of combinations cannot scale to the full range of module and volume-tier permutations, and it will silently produce wrong or missing answers as soon as a prospect's configuration falls outside the hard-coded set.