edgeacademyCUSTOMER ENGINEERING
MODULE 08 · LESSON 1 OF 5

Opening discovery and following the evidence

12 min read · Sources linked below

Theory and industry practice are explained below. Cloudflare-specific behavior links to official documentation. Customer examples are hypothetical.

Learning objectives

  • Structure a focused discovery call
  • Adapt questions to observed evidence
  • Close with a decision-oriented recap

The mental model

A discovery call needs structure without feeling like an interrogation. Begin by confirming why people joined, what decision they need to make, and how much time you have. Ask an open question about the problem, then follow the answer with specific diagnostic questions. If the customer says latency is affecting sales, ask which journey and what evidence links the two. If they say security is blocking a launch, ask what requirement remains unmet. The conversation should narrow uncertainty rather than jump through a memorized list. A short recap during the call lets the customer correct your model before a mistaken assumption drives the rest of the discussion.

How it works

Use a funnel from impact to system detail. Establish affected users, frequency, severity, and the consequence of doing nothing. Then reconstruct the request path and recent changes. Ask what still works, because a working comparison often isolates a boundary faster than a broad inventory. Record evidence with source and confidence: observed metric, customer report, interpretation, or untested hypothesis. This protects the later recommendation from quietly converting guesses into facts.

Branch deliberately. If the issue affects only one region, investigate network path and regional dependencies. If only authenticated requests fail, inspect identity, policy, and personalized processing. If no baseline exists, the first deliverable may be measurement rather than migration. Keep a parking lot for valuable topics that do not affect the current decision. When a customer requests a specific product, acknowledge it and ask what outcome they expect from it. Close with the agreed problem, constraints, unknowns, and a named next action. This is a general discovery method, not a prescribed Cloudflare sales script. Its effectiveness depends on listening and adapting, rather than asking every possible question regardless of the answer.

Applied example · hypothetical

Hypothetical branching scenario: the customer says checkout is slow. Weak response: We should turn on caching everywhere. Strong response: Which checkout step, which users, and what timing do you have? If the customer identifies payment authorization, investigate the external provider and retry behavior. If the delay is image loading before checkout, inspect asset delivery and browser timings. If they cannot identify a stage, agree a measurement session. Each branch leads to different evidence, showing why the first product suggestion would have been premature and potentially unsafe for personalized data.

In a customer engagement

In a Cloudflare discovery call, use product knowledge to ask sharper questions without turning the meeting into a feature tour. Ask whether the hostname is proxied only when that affects the request path under discussion. Invite the customer's application and operations context. End with a written recap that labels hypotheses and confirms which observation will determine the next architectural choice.

Interview relevance

For a discovery simulation, the interviewer may intentionally provide incomplete information. Resist filling gaps with confident assumptions. Ask one useful question at a time, reflect the answer, and explain why the next question matters. Demonstrate that the conversation can produce a decision and a next step within the allotted time.

Common mistakes

  • Reading a fixed checklist without listening
  • Converting reported symptoms into proven causes
  • Promising improvement before a baseline

Key terminology

Discovery agenda — A shared outline of the problem, desired decision, participants, and available time.

Diagnostic question — A question designed to distinguish plausible explanations using evidence.

Working comparison — A similar case that succeeds and can help isolate the failing boundary.

Explain it without the page

  • Which answer would make you change your next question?
  • What would you put in a discovery recap versus a parking lot?

Further reading

Sources explain the underlying concepts; examples and practice answers are original course material. Recheck current limits and plan eligibility before a real proposal.