This lesson is being recorded
Every video of this course is being uploaded to my YouTube channel. Subscribe so you do not miss it.
Bruno Ashford @brunoashfordSubscribeModule 3 · Lesson 3.4 · 11 min
From Analysis to Recommendation
Move from numbers to a decision: get a recommendation with evidence, trade-offs and a clear next step.
Use ← → to move. The marker highlights text and the pencil draws on the slide; navigation locks while you annotate. Undo removes the last mark, Esc exits.
1Frame the decision, not the data
Start with the decision you actually face. A decision has options, a deadline and a cost of being wrong. Give Claude those, and it will look at the data through the right lens.
I need to decide where to focus our sales effort next quarter. Options: push our top 5 customers for more volume, win back customers who stopped buying, or target new customers in our best product line. Use the clean sales file to help me decide.2Demand evidence for every claim
Tell Claude that every statement needs a number and a source file. This single instruction removes most vague, confident-sounding filler.
For each option, give me the evidence from the data with numbers, and name the file and columns you used. Clearly separate facts from assumptions.3Ask for the trade-offs and the risks
A recommendation without downsides is a sales pitch. Ask what would make each option fail and what it would cost.
What is the main risk of each option, what would we need to believe for it to work, and what would tell us within 30 days that we were wrong?4Get one page, not a thesis
Ask for a one-page decision memo in a new file: the recommendation first, then evidence, risks and the first three actions with owners. This is the format you can forward to a partner or manager.
Write a one-page decision memo to reports/sales-focus-memo.md. Structure: recommendation in two sentences, the three strongest pieces of evidence, the main risk, and three actions for next week with an owner for each.5Challenge it once
Before accepting, ask Claude to argue the opposite. If the recommendation survives its own best counter-argument, you can move with more confidence.
Now argue against your own recommendation as strongly as you can. Does it still hold?Analysis is cheap now. You can have ten charts in ten minutes. What owners are actually short of is judgment: what does this mean, what should I do on Monday, and what could go wrong if I am wrong?
Claude can do far more than summarise, as long as you ask it to argue like an adviser who has to defend the recommendation in front of you. Numbers attached to every claim, alternatives weighed, and a clear line between fact and assumption.
Key points
- Start with the decision, then bring the data.
- Every claim needs a number and a source.
- Always ask for risks and what would prove it wrong.
- One-page memo: recommendation first.
- Make it argue against itself once.
Avoid
- Asking "what do you think?" with no options or constraints.
- Accepting recommendations that do not cite the data.
- Treating a well-written memo as proof it is right.
- Delegating the final decision to the agent. You decide.
Tom runs a 9-person agency and wanted to raise prices. Claude analysed 18 months of client billing and time logs and recommended raising prices only on the three service lines where hours per client had grown by more than 25 percent, keeping the rest flat. It flagged the risk that two long-standing clients made up most of one line. Tom raised prices on new contracts first, lost nobody and added roughly 8 percent to monthly revenue within a quarter.