Stop Sending Spreadsheet Proposals: A Prompt That Sells the Backyard Before You Build It

I've been designing residential landscapes for sixteen years, and here's the uncomfortable truth: most of us lose the sale before the homeowner ever sees the number. We email a PDF with a plant schedule, a stamped concrete SKU, and a subtotal. Then we wonder why a $48,000 hardscape feels expensive and a $90,000 kitchen renovation feels reasonable.

The kitchen designer didn't sell cabinets. They sold Sunday breakfast with the grandkids. We need to do the same thing — and AI finally makes it fast enough to be worth doing on every bid.

Why your current proposal is killing your close rate

Look at the last proposal you sent. I'll bet it opens with a scope summary, jumps to a materials list, and ends with a price. That document is built for **you**, not the client. It's a build sheet wearing a tie.

The homeowner doesn't know what a 6" compacted base of 3/4" minus costs. They don't know why Techo-Bloc Blu 60 is priced differently than a big-box paver. So when they see the total, their brain does the only thing it can: it compares the number to a number they **do** understand — usually whatever their neighbor paid three years ago, or a Houzz screenshot with no zip code attached.

The 'Flawed Human' framework, applied to proposals

Good proposal writing borrows the same techniques as good copywriting: specificity, sensory detail, and a little bit of opinion. The mistake is assuming AI can't do that. It can — but only if you stop asking it to 'write a professional landscaping proposal' and start feeding it the things a real designer would notice on a site walk.

What to gather before you prompt

• **Site facts**: lot orientation, dominant tree species, prevailing wind, that one neighbor's roofline you're screening.
• **Client tells**: how they actually use the yard now (firepit on Friday? dog with a path worn into the lawn?), and the one phrase they kept repeating during the consult.
• **Material story**: not just SKUs — *why* you chose them. Thermal bluestone because they walk barefoot. Hornbeam over arborvitae because deer pressure is brutal.
• **The honest constraint**: budget ceiling, HOA setback, the septic field you're working around. Naming it builds trust.

If you skip this and just dump a CAD export into ChatGPT, you'll get adjectives — 'lush,' 'serene,' 'tranquil.' Useless. The prompt below forces specificity because it requires you to fill in the texture before the model is allowed to write.

The prompt

You are a senior residential landscape designer with 15+ years of experience writing winning proposals for projects in the $40k–$150k range. You write like a working professional, not a marketing department.

I'm going to give you the technical scope of a project. Your job is to convert it into a 4-section narrative proposal that helps the homeowner feel what it will be like to LIVE in this space — without making any claim the build sheet can't back up.

=== PROJECT INPUTS ===
Client first names: {{CLIENT_NAMES}}
Property location & climate notes: {{LOCATION_CLIMATE}}
Lot orientation & key site facts: {{SITE_FACTS}}
How the family uses the yard today: {{CURRENT_USE}}
The phrase the client repeated during consult: {{CLIENT_PHRASE}}
Primary materials & WHY chosen (not just SKUs): {{MATERIALS_WITH_REASONS}}
Dimensions of major elements: {{DIMENSIONS}}
Plant palette + reason for each anchor species: {{PLANTS}}
Lighting & water features (if any): {{LIGHTING_WATER}}
Key constraint you solved (budget/HOA/grade/etc): {{CONSTRAINT_SOLVED}}
Investment range: {{PRICE_RANGE}}

=== OUTPUT STRUCTURE ===

1) "A Saturday in [Month]" — 150–180 words.
   Open with a specific sensory moment 18 months from now. Name the time of day, the temperature, what they hear, what they smell, where they're sitting, what their hands are touching. Use {{CLIENT_PHRASE}} as a callback once, naturally. Do not use the words 'oasis,' 'retreat,' 'tranquil,' 'lush,' 'transform,' or 'dream.'

2) "What we're building, and why" — 3 short subsections (60–90 words each).
   For each major element (hardscape / planting / lighting-or-water), explain the design decision in plain language. Reference the actual material and dimension, then connect it to a real human behavior from {{CURRENT_USE}}. Use one mildly opinionated sentence per subsection — the kind of thing a designer would actually say out loud on a site walk.

3) "What we considered and ruled out" — 80–110 words.
   Name 2 alternatives you rejected and the honest reason. This is the trust-building section — it proves you thought about this instead of defaulting.

4) "Your investment" — 60–80 words.
   Frame {{PRICE_RANGE}} in terms of cost-per-year-of-use (assume 15-year material life). Do NOT apologize for the number. Close with a single sentence about what happens next, written like a human, not a template.

=== TONE RULES ===
- Vary sentence length aggressively. Some sentences should be four words.
- No corporate softeners ('we are pleased to,' 'thank you for the opportunity').
- No em-dash overuse. Max two per section.
- Reference {{CLIENT_NAMES}} by first name exactly twice across the whole document.
- If any input is missing, write [NEEDS INPUT: x] inline rather than inventing detail.

Notice what the prompt does **not** ask for: a hero adjective list, a 'vision statement,' or a mission paragraph. Those are the tells of AI-generated proposals, and homeowners are getting better at spotting them every month.

How to tweak it for the project in front of you

For a pool surround or full outdoor kitchen

Add a fifth section called **'How this ages'**. High-ticket clients are buying a 15-year asset; tell them what the travertine looks like in year three, when the Japanese maple actually casts useful shade, and when the pergola wisteria will fully knit. Future-pacing closes deals the way nothing else does. If you're newer to writing this kind of long-form spec, our breakdown of [the Flawed Human framework for AI prose](/posts/flawed-human-ai-prose-framework) covers the pacing tricks that keep these sections from reading like a brochure.

For a budget-sensitive or first-time client

Replace section 4's cost-per-year framing with a phased build narrative — Year 1 hardscape and irrigation, Year 2 plantings and lighting. It lowers the activation cost without lowering the design ambition. Pair it with a clear written policy on change orders so the phasing conversation doesn't turn into a margin leak six months in — our piece on [explaining policy changes without losing clients](/posts/explaining-policy-changes-exclusions-premium-increases) covers the script.

For a referral or repeat client

Cut section 3. They already trust your judgment — what they want is the daydream. Lean section 1 longer (220+ words) and add a callback to a detail from their previous project.

Where designers go wrong with AI proposals

A field-tested checklist before you hit send

• Read section 1 out loud. If it sounds like a brochure, the sensory detail is too generic — rewrite with a specific time of day.
• Search the document for the words **'transform,' 'oasis,' 'dream,' 'unique,' 'bespoke.'** Delete every one.
• Confirm every dimension and material in the narrative matches your line-item sheet exactly. AI will round numbers.
• Add one handwritten-feeling sentence the model couldn't have generated — a callback to something the client said, a joke about the dog, a note about the neighbor's wind chimes. That's the line they'll remember.
• Attach the line-item budget as **page 3 or later**, never page 1.

What changed when I started writing proposals this way

My close rate on proposals over $35k went from roughly 1-in-4 to a little better than 1-in-2 across the last two seasons. Average project size went up too, mostly because clients stopped value-engineering the lighting package out — they'd already pictured the boxwoods uplit at dusk.

None of that is the AI. The AI is a drafting tool. The shift is that you're now selling a Saturday morning instead of a stack of pavers, and the prompt above just makes it fast enough to do on every single bid.