What You Need to Know
Research gathers and synthesises outside material, so you start from what is already out there rather than a blank screen. Planning turns a goal into ordered steps. Process optimisation finds and fixes the step that slows a process down. All three share one shape: Claude produces the first pass, and you own the judgement that follows.
Take an estate agency whose manager uses Claude for all three: other agencies' valuation practice, an open-house weekend plan, and slow lettings.
Research at Two Scales
The product supports research at two scales, and the Help Center draws the line. Web search suits straightforward factual queries that one or two lookups can answer. Research mode runs multiple searches that build on each other. It explores different angles of a question and works through the open questions systematically. It returns cited findings and suits bigger multi-source jobs. At the agency, one lettings rule is a web search; a survey of valuation practice is Research.
The value is coverage and speed. The limit is that anything important gets checked before anyone repeats it, citations or not: a citation shows where a finding came from, not that it is right.
Plans Are Built from Stated Constraints
A first-pass plan beats a blank page. Claude is unusually good at completeness. Ask what a plan is missing and you get dependencies, sequencing problems and the step everyone forgets.
Claude will also plan confidently around constraints you never mentioned, since it does not know them: the one accessible entrance, the volunteer who cannot work Sundays, the double-booked room. Nothing in the output warns you the plan was built without them. An unconstrained open-house plan might staff Sunday viewings with a negotiator who never works Sundays. That plan is fiction that reads like competence, and more dangerous than an obviously bad plan because it invites trust.
The countermeasure is cheap. Anthropic's business prompt-engineering guidance says to treat Claude like an intern on its first day, with clear, explicit instructions and all the necessary detail. For a useful plan, put the real constraints in the brief, then ask for the assumptions it rests on. An assumption you can read is one you can correct.
The Optimisation Loop
Claude generates hypotheses about where a process is slow, fragile or duplicated. It cannot tell you which one is true, because it has never watched your process run. Data decides. The loop has four steps:
- Describe the process and its constraints.
- Ask for candidate bottlenecks, and for the evidence that would confirm each one.
- Go and get that evidence.
- Change the step the evidence indicts.
Evidence comes after the hypotheses are named, so you know what you are looking for. It comes before anything changes, so the change has a target. For the agency's lettings, Claude might suggest slow referencing or late landlord sign-off; the stage dates in its own files decide.
The same guidance advises approaching the work like a scientist, testing and iterating often, and pairing prompt work with the organisation's own subject-matter experts rather than trusting either alone.
The Wrong Bottleneck
The most expensive mistake is optimising a step that was never the constraint. It spends the budget and the appetite for change on something that was fine. It usually comes from treating the most plausible hypothesis as the confirmed one. Plausible and real causes coincide often enough to tempt you, and rarely enough to be expensive. The evidence that would have separated them was usually a day's work to collect. Buying faster referencing when the dates blame landlord sign-off spends the agency's money and leaves the delay.
In the exam
The exam tests the judgement, not the first pass. Every keyed answer in this objective either supplies the facts Claude could not know or checks something before acting. Every recurring distractor treats a fluent output as a finished one. Domain 3 decides which feature carries research; this objective tests the practice around it. Checking findings before repeating them is Domain 2's discipline.
The professional tell
In this objective's stems, the keyed answer checks the evidence before changing anything: last season's timings, the usage records, the complaint log. Distractors rebuild the process on fluent reasoning, or reach for a bigger feature when local facts were missing.
Exam Traps
Treating a plausible, well-structured plan as a checked one
Plausibility is a property of the prose; validity comes from the constraints and facts the planner supplied. A plan built without the real constraints fails on contact with the real event, so the keyed answer states them up front or validates the draft against them before relying on it.
Changing the process on the strength of the best-sounding hypothesis
Claude's candidate bottlenecks are hypotheses, not diagnoses; it has never watched the process run. Acting on the most plausible one spends money on a step no evidence has indicted, and the keyed answer gathers the evidence first.
Substituting more research for local evidence
Citations from elsewhere cannot say which cause is true in your process. When the question is why this process behaves this way, the deciding evidence is local data, and no amount of external synthesis replaces it; escalating the feature is not the same as answering the question.
Practice Scenario
A city bike-hire scheme's coordinator keeps hearing that commuter-belt docking stations are empty by eight in the morning. She describes the scheme to Claude, which suggests four plausible causes: the overnight rebalancing round finishes too early, two residential stations feed the same office cluster, weekend maintenance keeps bikes off the road on Mondays, and riders arriving at full stations cycle on with the bikes. Only her own scheme's figures will settle which cause is at work, and she will act on none of them until they do. What is the professional next step?
Build Exercise
Plan It Twice, Then Find the Real Bottleneck
What you'll learn
- How stated constraints change the usefulness of a first-pass plan
- How to spot the fiction-that-reads-like-competence failure
- How to run the describe, hypothesise, evidence, change loop
- How to test whether the data points at the step you would have guessed
- Pick a real upcoming piece of work with genuine constraints: an event, a rota, a launch, anything that is actually yours
Why: The exercise only works when you know the real constraints well enough to catch a plan that ignores them.
You should see: One named piece of work and, privately, its three most binding constraints.
- Ask Claude for a first-pass plan twice: once with no constraints stated, once with the three most binding real constraints written in
Why: The comparison makes the objective’s core failure visible: the unconstrained plan reads just as competently as the constrained one.
You should see: Two fluent plans that differ in exactly the places your constraints bite.
- List what the unconstrained plan got wrong that nonetheless reads as perfectly sensible
Why: Naming the specific fictions builds the reflex of checking plans against facts rather than against fluency.
You should see: Three or four concrete errors, each invisible to someone who did not know the constraints.
- Now pick a process that feels slow and describe it to Claude end to end; ask for four candidate bottlenecks and, for each, the evidence that would confirm or rule it out
Why: This is the optimisation loop’s first half: hypotheses plus the evidence that would decide between them.
You should see: Four candidate causes, each paired with a named, gettable piece of evidence.
- Collect one of those pieces of evidence this week, and note whether the data pointed at the step you would have guessed
Why: The gap between the guessed bottleneck and the evidenced one is the lesson; most people are surprised at least once.
You should see: One piece of real data and a one-line verdict: guess confirmed or guess wrong.
Sources
- Use research on Claude — Anthropic
- When should I use web search, extended thinking, and research? — Anthropic
- Prompt engineering for business performance — Anthropic
- Claude Certified Associate – Foundations Exam Guide — Anthropic. No Anthropic document covers process optimisation as a business skill. This lesson draws on the exam guide's objective and standard practice; the articles above support only the points they are cited for.