Getting More Referrals for Interior Design Using AI

Referral partners are the highest-leverage lead source an interior design firm can have. Property management companies, builders, real estate investors. One relationship can generate clients for years. We built an AI-assisted marketing asset to activate those partnerships, and what made the difference had nothing to do with the prompt.

Strategy before AI

This build was for 1584 Design, a firm we support as their fractional AI officer. Earlier in the engagement they were supply-constrained, with more projects than they could handle. We focused there first: operational efficiency, more capacity. Once that was solved, the pipeline was underfull relative to what they could take on. Time to work on lead generation.

The target wasn't leads directly. It was referral partners. The property management companies and builders who already work with the short-term rental investors 1584 wants as clients. One referral partner can generate multiple clients over multiple years for roughly the same effort it takes to land one lead directly. The math is obvious.

The play was a printed short-term rental investor guide that partners can hand to their clients. Something that builds 1584's credibility and converts the referral partner's trust into inbound leads. It would plug into a CRM-backed system with a call-request form, automations, and attribution tracking. The piece we had to build was the guide itself.

The prompt was 17 words

AI made this fast. The prompt we used was about 17 words. "Create a guide" with a small amount of elaboration around it.

The first version was a problem. Budget estimates were off by 5x, the kind of numbers that attract the wrong clients and set expectations you can't meet. The audience framing was wrong: it addressed all property owners instead of short-term rental investors. The design advice was DIY-level. String lights. Board games. Nothing that reflects the work 1584 actually does.

The format was a plain document with placeholder text where the branding should have been. You can't hand that to a referral partner.

Why AI produces generic output

AI isn't bad by default. It's a blank slate. Every time you open a new conversation with Claude or ChatGPT, the model knows nothing about your business unless you tell it. The prompt is one piece. The context you provide alongside the prompt is the piece most people skip.

Think of it as a capable employee who forgets everything at the end of each day. Tell them "write a guide" on Monday morning and they produce something generic that sounds plausible. Give them your brand voice, pricing, ideal client, past work, and copy rules, and they produce something you'd actually put your name on.

What 10 context files changed

For the second attempt we attached 10 context files we'd already created as part of 1584's business documentation: brand voice, copy principles, services and pricing, website copy, print layout preferences, and section guidance for guides. Same 17-word prompt. Completely different output.

The AI pulled statistics from sources referenced elsewhere in 1584's business folder. It drew on blog posts they'd written. It incorporated a case study. It built a guide that was branded, written for the right audience, priced correctly, and formatted for print. It rivals what they'd paid professionals to produce in the past.

The prompt didn't change. The context did everything.

The point

Codifying your business isn't overhead. It's infrastructure. Brand voice, pricing, ideal client, SOPs, copy rules. Once that documentation exists, AI can reference it for any task: a guide, a proposal, a follow-up email, a marketing asset. The investment in writing it down compounds every time you use it.