The Best Listing Prep You’ve Ever Done

The CMA most agents hand over is forgettable.

It's a stack of three comps and a price. No read on the wider market, no trend, nothing that tells the seller where things are actually headed — a snapshot, handed over while they make a six-figure decision about the future.

You could build the great version: pull the local market data, study the segment, read the momentum, design it beautifully. That's the better part of a day, per appointment. So nobody does it.

But a better tool can't do the whole job, and that's the point. No model knows that the neighbor runs a salvage yard, or that the lake view is the whole ballgame, or that the train tracks read as “walk to the station” in one town and “freight at 2am” in another. The decisive facts live in your head. This system's real job is getting them onto the page.

Step 1. Read this before you start

Most of what we build gets out of your way. This one asks more of you, and here's the honest reason: when you're sitting across the table from a seller, your confidence comes from you knowing the market — not from a tool knowing it for you.

A seller can feel the difference in about ten seconds. The agent reading numbers off a screen sounds like a narrator. The agent who studied the comps, who can say why this one's worth more and that one sold soft — that agent gets the listing.

It matters most exactly where it's hardest. When inventory is rising and homes sit a little longer, the room for a lazy price disappears. You have to be sharper at pricing, which means prepping harder than you ever have.

It's built for the listing appointment. One bonus: at the very end you can turn the same document into an equity review for a past client in a single step — no separate mode to pick.

Step 2. How it works: two sets of comps

Most CMAs dump every comp into one pile and average it. We keep two sets apart — because one tells you the price, and the other tells you the market's behavior, which is how much to trust that price.

The workflow, end to end

1. You — drop the address in and tell Claude to start. It begins researching the home right away.
2. Claude, at the same time — researches the subject's facts and the three portal estimates. Never the comps.
3. You — pull your two comp sets from the MLS while it works.
4. You — confirm or correct the research, then upload both files.
5. Claude — analyzes the data: segment trends in code, your comps checked against the subject.
6. Claude interviews you — you name the threats.
7. Both of you — walk through each comp, one at a time.
8. Out comes a CMA, an agent brief, and a content pack, all in your brand.

Step 3. Install the skill

The skill is the engine. It installs once and works everywhere — it isn't part of any single project.

1. Get best-listing-prep.skill from the email above.
2. In Claude, open Settings → Capabilities → Skills and upload it.
3. That's it. Skills are global, so it's available to every project from now on.

Step 4. Create your project

1. In Cowork, click New project → Start from scratch.
2. Name it Best Listing Prep.
3. Paste the block below into the project's instructions field, then create it.

PROJECT — Best Listing Prep

This project prepares an in-depth CMA for a face-to-face listing appointment.
Run it with the Best Listing Prep skill.

Use my brand skill for who I am (name, brokerage, market) and apply my brand to
the CMA and the Agent Brief. Plan on 3–6 high-fidelity comps.

The comps are mine — I upload both files (comps.csv + segment.csv). You never
search for or invent comparable sales. Your research covers the subject home's
own facts and the threats I name.

Guardrails:
• You gather and analyze; I own the judgment. Never decide the home's value on
  your own, and never decide what matters about the area — ask me.
• Every fact you research must be sourced and shown to me to confirm or correct
  before it appears anywhere. Never guess the subdivision, HOA, or schools.
• No editorializing on schools, "family-friendliness," safety, or neighborhood
  demographics. Factual figures only if I add them. (Fair Housing.)
• Local Market Segment: compute trends in code, segmented by status, outliers
  shown — never deleted. High-Fidelity Comps: flag only, never edit.

Deliver three things, all in my brand: a client-ready CMA (HTML), an internal
Agent Brief (HTML — threats, objections, the hard price conversation), and a
locked content pack for Claude Design. At the very end, offer to convert it into
an equity review by dropping the suggested-list-price page.

Step 5. Kick off the research

Two things happen at once here, and the order matters.

Step 6. Pull your two comp sets

This is the part only you can do. Claude never pulls comps for you — you're making the calls a machine can't.

Set one — high-fidelity comps

Pull three to six: the hand-chosen comps you'd normally present. Your closest matches on size, recency, location and condition — the ones you'd stand behind in front of a seller. Export them on their own.

Set two — local market segment

Now zoom out. You're not picking comps here, you're capturing how your segment is behaving. Widen the criteria and pull enough to see a real trend.

• At least a few hundred records. Fewer than that and the trend is just noise.
• All statuses — active, pending, closed, expired, cancelled. Each one says something different.
• Two years back or more, so you see direction rather than a single snapshot.

We don't define your segment — you do. A segment is a judgment call and it's yours. The two below are reference frames; the move is always the same: tight criteria for your comps, then loosen every dial for the segment.

Two worked examples

Step 7. Confirm what Claude found, then upload

Come back to your project. Claude has the home's facts and the three portal estimates waiting. Your first job is to confirm or correct them.

If you're correcting a public record — telling Claude the home is really 3,850 sqft, not the tax record's 3,629 — good. But it won't print that on the client CMA until you confirm you have the appraisal or measured floor plan to back it. Unverified corrections stay in your internal brief.

Upload both comp files

Drag in comps.csv and segment.csv. Claude reads them and works the numbers before it asks you anything.

From the same data it calls the market: buyer's or seller's, and how strongly. That read sets how aggressive or conservative to price, and it drives the visuals in both documents.

Step 8. The interview: who you're with, then the threats

With the numbers on the table, Claude turns to the part it can't compute. Two things: first the people, then the threats.

First — who are the sellers?

• Names, for the cover, so the report is never addressed to “the seller”.
• What you already know — what they do, how they came to you, their timeline, how they like to communicate.
• A social link if you have one (LinkedIn, Facebook, Instagram) and Claude pulls what's public and relevant.
• No link but you want a look? Give a name plus one detail — Claude searches, then asks “is this them?” before it uses a single word.

Then — name the threats

This is the salvage-yard moment. Claude has the data, but it can't know the neighbor's yard is a mess, or that the lake view is the whole sale, or that the busy road only bites on the patio side. You can. So instead of guessing from a checklist, it asks you, with the numbers already in hand. You're getting ahead of issues to position and disclose them — never to hide them.

1. “What would a sharp buyer's agent point to when they're arguing for a lower price?”
2. “What will you want to be ready to speak to — the things a buyer asks about that the photos won't show?”
3. The anchor question: “If this sits on the market, what'll be the reason?”

Six prompts to jog your memory — skip any that don't apply

• The lot and what's around it — busy road, a commercial use behind, power lines, the neighbor's property, the view or what kills it.
• Noise and nuisance — train, flight path, freeway hum, a nearby venue.
• Condition and systems — roof, HVAC, pool, foundation, anything dated against the comps, anything deferred.
• The layout's quirks — a walk-through bedroom, no true primary, an addition that doesn't flow.
• HOA, legal and money — special assessments, litigation, rental caps, a solar lease, an insurance or flood-zone issue.
• The intangibles only you know — why the last buyer walked, a local stigma, the thing locals know that no portal ever will.

Then Claude does its half: it researches what you named, with sources for you to confirm, and raises the structural threats every sale has — the price gap, the active competition, the momentum, the condition gap, time-on-market risk — as questions for your local read, never as conclusions.

One more: market timing. Data can lie about timing. If your comps mostly sold in a hot spring and you're listing into a cooling autumn, Claude flags that the comps cluster in one season, and your read turns that flag into a more or less aggressive price.

Step 9. Walk through each comp

Claude takes your handpicked comps one by one, lays each beside the subject, and asks the only question that matters: what's different?

One honest sentence each — the upgrades, the lot, the condition. The kitchen that's remodelled and yours isn't. The premium cul-de-sac lot. The one that sold to family under market. You're not filling in a form; you're making the call only the agent who has seen these homes can make.

If a comp looks off — too small, too stale — Claude flags it and asks; it never removes it. Your notes ride along to that comp's page in the final CMA, in your words.

Step 10. Polish it in Claude Design (optional)

The HTML CMA is genuinely good. But if presentation matters for this meeting, the content pack goes into Claude Design and comes out as a branded, multi-page CMA you'd be proud to hand across the table. The pack says what goes on each page and where the emphasis sits; Design decides colors, type and layout, because your design system already holds all that.

1. Cover — your hero photo of the home, addressed to your sellers by name. No photo? A clean branded cover from your design system instead.
2. About me — pulled straight from your brand, no rework.
3. Market overview — the segment story: headwinds, tailwinds, the data, the direction.
4. Comp comparisons — one comp per page, side by side with the subject, plus your condition notes. Readable for a seller, not a 30-row grid.
5. The closing page — your recommended pricing strategy.