Get Real Value from AI: The REASON Method
- Clare Patterson
- Aug 28
- 4 min read
Updated: Sep 8
Stop getting generic AI outputs. Learn the REASON Method to structure prompts with context, clarity, and proof for better results.
I was speaking with a good friend and a person who I have worked with over a number of years who was struggling to get results from ChatGPT. The problem wasn’t effort; it was context. One short line in, generic answer out. Painfully disappointing and time-consuming.
Think real life. If a plumber comes to your house, you don’t say, “Fix the sink.” You say, “The tap started dripping on Tuesday, pressure’s low, and there’s a small puddle under the U-bend.” That context tells them where to start. ChatGPT and all AI tools are the same.
At work, you wouldn’t brief a colleague with, “Make a presentation on sales.” You’d say, “Board next Tuesday, focus on Q3 pipeline quality, three slides, visual, end with two decisions.” Specifics drive better work — human or AI.
And if you’ve ever typed a vague prompt, you’ll know the kind of nonsense you get back: “In the world we live in today, we offer a revolutionary and transformative solution to help unlock your potential.”
Clichés in, clichés out. If you want to get real value from AI follow this list.
First: set up a Project in AI (non-negotiable)
Tone of voice: straightforward, practical, human, slightly irreverent.
Who you are / who you sell to: ICPs, sectors, pains you solve.
Proof: case studies, stats, testimonials, pricing ranges.
House rules: plain English. Banned words: “I’m delighted,” “revolutionary,” “game changer,” “thrilled to announce.”
Style samples: upload examples you like and a short piece you’ve written. Tell ChatGPT to mirror that style.
This gives the tool context, memory, and guardrails — so every prompt starts ahead, not from scratch.
The REASON Method (my framework) to Get Real Value from AI
Use REASON to structure every prompt. It forces clarity.
R — Role: Who should it be? “Act as our Sales Director…”
E — Evidence: Hard facts — metrics, quotes, links, pricing.
A — Audience: Who’s reading? What they value, fear, ignore.
S — Situation: Objective, constraints, deadlines, sensitivities.
O — Output: Exact format, sections, length, CTA.
N — Nuance: Tone, banned words, persuasion tactics. If you’re selling, ask for ethical NLP patterns (future pacing, embedded commands).
Examples you can copy
1) Sales — sending a proposal for a software subscription
R: Senior Account Executive at a B2B SaaS company. E: 21% faster time-to-value; median payback 4.5 months; SOC 2; case study link; £1,250/mo. A: Ops lead at a 120-person manufacturer; hates risk, values predictability. S: Follow-up to discovery; goal is a signed proposal this week. O: Email, 220–260 words, subject line, three bullet reasons, one-paragraph risk reversal, CTA with two time options. N: Plain English, confident, practical. Layer subtle NLP: future pacing (“so your team finishes jobs faster”), embedded commands (“review the proposal today”), social proof.
2) Managing Director — Q4 focus memo
R: MD of a 45-person SME. E: Margin 51%→54%; ship v2 by 15 Nov; meetings baseline 26 hrs/person/month. A: Whole company. S: Competing priorities; need alignment for Q4. O: Under 400 words; headings and bullets; include a four-point checklist and a “Not doing” list. N: Calm, clear, practical; plain English; no clichés.
3) Marketing — sharpening core product messaging
R: Senior B2B marketer and conversion copywriter. E: Product: “Line Monitoring Starter.” Outcomes: downtime –18%, scrap –9%, payback 4.5 months; integrates with [X, Y]; ISO 9001; two customer quotes. A: Ops Directors at mid-market manufacturers; time-poor, sceptical of vague promises. S: Need a clear core message that earns clicks and demo requests. O: 120-word value proposition, 10-word tagline, three-line elevator pitch, one proof-driven headline, and three CTA button options. N: Plain English. No waffle. Lead with outcomes and risk reduction.
REASON skeleton (copy/paste)
R — Role
E — Evidence (facts, links, metrics)
A — Audience (who, what they care about)
S — Situation (goal, constraints, deadlines)
O — Output (format, sections, length, CTA)
N — Nuance (tone, banned words, persuasion tactics/NLP) Real nuggets (small things, big wins)
Outline first: “Draft a bullet outline; wait for my OK before writing.”
Numbers > vibes: “If data is missing, insert [STAT NEEDED]—don’t invent.”
Three variants: “Give me safe / punchy / bold.”
Red-team it: “List five reasons this might fail, then fix each.”
Tighten pass: “Keep meaning, cut 20% of words, remove clichés.”
Audience switch: “Rewrite for a CFO who cares about ROI.”
Single source of truth: “Use only facts in the Project.”
Self-audit: “Before sending, confirm Role, Evidence, Output, and CTA are clear.”
Style lock: Paste 150–200 words in your voice and say “Mirror this style exactly.”
Assumption check: “List the top three assumptions behind this draft.”
Beyond ChatGPT
Here’s the bigger point: this applies to any AI tool you’re using — DeepSeek, Manus AI, Claude, Gemini, whatever comes next. They all work better when you give them context, structure, and constraints.
For a salesperson, that means loading proof points, numbers, and objections so the AI can shape a persuasive proposal. For a marketer, it means feeding it the exact outcomes, banned words, and proof-driven claims you want in your messaging. Whatever the role, the REASON method keeps prompts tight, transferable, and effective across platforms.
Wrap-up
Context is the multiplier. Load it once in a Project. Structure every ask with REASON. Apply the nuggets. Iterate in small, clear steps. That’s how you stop generic responses and start getting useful work product — every time.
Next week, I will head back to lead generation, but in the meantime, I hope this article will provide you with an easier week ahead.
Comments