Quick Comps

Statistical prep for every hot client, from your phone.

Three outputs, one client file. Ask for what you need and the same data answers a different question each time.

Step 1. Pre-flight checklist

Step 2. Install the skill

The skill knows how to read your comp data, anchor on a defensible value, measure leverage and produce the memo the same way every time — so the number is reproducible.

1. Get the skill file from the email above and save it somewhere you can find again.
2. On a desktop or laptop browser, open Customize → Skills in claude.ai.
3. Click +, choose to create or upload a skill, and select the file as-is. No need to unzip.
4. Find it in the list and make sure the toggle is on.

Step 3. Tell it who you are

Your Brain Doc is what makes Quick Comps yours. The skill reads it on every run to know who you are, what market you're in, and how you build a comp set. Start with the identity that prints on every memo footer: name, brokerage, license number, phone, and the state you're licensed in.

Step 4. Your client files folder

Quick Comps works on a folder-of-clients pattern. You create one folder. Each hot buyer client gets their own MLS export inside it, scoped to their must-have criteria. When you ask Claude to prep you for the Johnsons, it finds the Johnson file automatically.

Create the folder

1. In your browser, open Google Drive.
2. Create a new folder — Quick Comps Client Files is a clean default.
3. Copy the folder's link. Claude extracts the folder ID from it automatically, and you'll paste that link into your Brain Doc in Step 6.

Build a good client export

The discipline is filters first, geography second. Your export should be wide enough geographically to cover their search area, but tight on the criteria they won't bend on. The criteria filters are what keep the file small.

1. Set the hard filters first — single-level only? Pool required? Bedroom count? Price range? These are what shrink the result set.
2. Then set the geography — the cities or ZIPs they'll consider. Going wide here is fine, because the filters above already control the size.
3. Set the time window — the last six to twelve months, statuses active, closed, expired, cancelled and under contract.
4. Drop the heavy text columns if your MLS lets you pick an export template: features, public remarks, private remarks. That's about 80% of the bytes and none of it is used.
5. Paste it into a new Sheet in your folder and name it after the client. A tight, criteria-scoped export usually clears 500 KB.

Name the files so retrieval is one-shot

Claude is good at fuzzy matching — “Mike Johnson” still finds the file when you say “the Johnsons” — but consistent naming makes it reliable. Good: Johnson Family, Mike Johnson — Chandler, Williams Spring Search. Avoid cryptic abbreviations only you understand; those make Claude ask “which file?” more often. When a client closes or pauses, drag their file into an Archive subfolder — the folder listing is your active-client roster.

Step 5. How you run comps

The data shows price, size, year and location. It can't see condition, layout or finish. So this step is where your judgment gets written down: read the starting methodology and change the lines that don't match how you actually build a comp set.

What disqualifies a property from being a comp

• A different number of stories — never compare a two-storey to a single-level.
• Septic against sewer.
• A different lot type — acreage against a subdivision lot, rural against urban.
• A significantly different bedroom count. A three-bed and a five-bed are not the same comp.
• A different property type — single-family, patio home and townhome are three different things.

What you weight comps by, in order

1. Square footage — closest size first.
2. Year built — within five years if possible.
3. Upgrade and finish level — builder grade against renovated against luxury.
4. Pool: presence and quality.
5. Lot size and premium features — view, corner, oversized, cul-de-sac.

What the data never sees

These become “confirm at the property” prompts in every read: condition, finish quality, floor-plan flow, kitchen layout, lot quality and orientation, smell and sound and the neighbor situation — and anything that would scare off a buyer in the first thirty seconds.

There's room for two more things: anything about your market the data hasn't caught up to yet, and any personal rule you want respected — “I never recommend going above list”, or “on rural acreage, widen the comp radius to five miles”.

Step 6. Wire it up

Your answers become a Brain Doc that lives in a project's custom instructions, so every chat in that project loads it automatically. Paste the block below with your own answers filled in.

QUICK COMPS — BRAIN DOC

WHO I AM
- [first] [last] · [brokerage] · license [number]
- [phone] · licensed in [state]      (this prints on every memo footer)

WHERE MY CLIENT FILES LIVE
- Folder: [paste your Google Drive folder link]
- One Sheet per hot client, named after the client.
- Refresh a client's Sheet by re-running their saved MLS search and pasting
  over it. Replace, never append.

HARD FILTERS — that's not a comp
- Different number of stories.
- Septic vs sewer mismatch.
- Different lot type (acreage vs subdivision, rural vs urban).
- Significantly different bedroom count.
- Different property type (single-family vs patio home vs townhome).
- [your own]

WEIGHTED FACTORS, IN ORDER
1) Square footage, closest first.
2) Year built, within 5 years where possible.
3) Upgrade and finish level.
4) Pool: presence and quality.
5) Lot size and premium features.

CONFIRM AT THE PROPERTY — the data never sees these
Condition · finish quality · floor-plan flow · kitchen layout · lot quality and
orientation · smell, sound, neighbors · anything that loses a buyer in the
first 30 seconds.

MY RULES
- [e.g. I never recommend going above list.]
- [e.g. On rural acreage, widen the comp radius to 5 miles.]
- Market temperature is derived from my closed comps on every run — don't ask
  me to declare it.

1. In claude.ai, go to Projects in the sidebar — the regular Projects section, not Cowork. Quick Comps runs on demand in normal chats.
2. Click + New Project and name it Quick Comps.
3. Paste the Brain Doc into the project's custom instructions field and save.
4. Inside the project, confirm the skill is available and the Drive connector shows as connected.

Step 7. The first test — pick a hot client

The whole point is real client prep, so test it on a real one. Build their saved search in your MLS, export, paste into a new Sheet in your folder, name it after them — then start a chat inside your Quick Comps project and try all three modes.

1. Segment report: “Prep me for the Johnsons this weekend.” No addresses needed — Claude finds the file by name and gives you a segment briefing with three talking points.
2. Showing prep: “I'm showing the Johnsons these three homes Saturday…” and it builds a card per home, color-coded by pricing health, with watch-outs and walk-through prompts.
3. Single property: “Run quick comps on [address] for the Johnsons.” The three-tier ladder memo, with pricing-health gating.
4. Read the outputs on your phone — that's what they're designed for. Check the market-temperature stats against your gut. If anything feels off, tune the methodology in your Brain Doc and re-run; the skill picks up changes immediately.

Step 8. Keep the client files fresh

Quick Comps is only as smart as the data behind it. Refresh a client's Sheet whenever you're about to prep for them: re-run the saved search and paste over the existing data. Replace, don't append — old comps lose relevance fast, and the skill derives your market temperature from this data on every run.