What is a Monday scorecard?
It's the one report every agent sends their coach about last week: appointments set, agreements signed, contracts written, closings. Plus what they're committing to this week. Every 1:1 and every team meeting starts from it.
Most teams have some version of it. Ours used to be a form agents filled in from memory. Now it opens with last week's numbers already pulled from the CRM and the transaction team's records. The agent's job is to confirm they're right, not to type them.
Why team numbers stop matching
Two things broke on Ken Pozek's team, and they're the same problem.
The report lied by accident. One agent closed two sides in one week. Her report said zero. The deals were closed in real life, but in the CRM they still sat at Under Contract, and the report counted the CRM stage. The report wasn't wrong about the CRM. The CRM was wrong about her week.
The same number meant two things. Speed to lead showed hours for an agent on one screen and minutes on another. One took the average on an every-day clock. The other took the median on working hours. Both were called "speed to lead." Neither said which one it was.
The minute two screens disagree, nobody trusts either one.
Once that happens, the 1:1 turns into an argument about the number instead of a conversation about the work.
How Ken Pozek's team runs it now
- Every number has one written definition. One page. Every screen, report and reminder reads it. Change it once, it changes everywhere.
- Every number has one source of truth. Closings come from the transaction team's records, because the TCs are always right about what closed. Appointments and agreements come from the CRM.
- The report shows what's wrong before it sends. An appointment with no outcome or a deal on the wrong stage shows up with a button to fix it. The report won't send until it's fixed.
- Send asks one question. "Are these numbers right?" Yes sends it. "Something is missing" opens the fix.
The template
You don't need software for the part that matters most. You need the page. Here's the shape of ours. Write your own definitions, then make every report and every meeting use them.
| Number | What counts | Source of truth | Who fixes it |
|---|---|---|---|
| Speed to lead | Working minutes from a lead being assigned to the first call or text. The median of the last 30 days, not the average. | CRM call and text log | Agent |
| Appointment | A meeting booked with a client, with a date, a type and an outcome once it happens. | CRM | Agent |
| Agreement signed | A buyer agreement or listing agreement signed, counted the day it's signed. | CRM deal stage | Agent |
| Contract | A deal under contract, counted the day it goes under contract. | CRM deal stage, checked against the contract | Agent |
| Closing | One closed side, counted on its close date. | Transaction team's records | TC team |
Then the report itself. Ours has four parts:
- Last week's numbers, each labelled with its source.
- Anything that makes a number wrong, with what to do about it.
- This week's commitment: calls, conversations and appointments.
- One question before send: are these numbers right?
Run it by hand for a month. You'll find out fast which numbers your team has been arguing about.
Build your own with AI
The template tells you what to count. The kit builds you a tool that does the counting from your own data, wherever it lives today: your CRM, your transaction system, the sheets your coaches keep.
A scorecard tool in Claude. Drop in last week's files from your CRM, your transaction system and your tracking sheets. Get every agent's numbers on your definitions, every place two systems disagree, the records making each number wrong, and a Monday message per agent ready to send.
- Write your definitions.Use the template above. The tool counts only what your page says counts.
- Export last week from your CRM.Activity, deals and appointments. In Follow Up Boss that's Reporting, then Agent Activity, Deals and Appointment, then the export arrow (account owner, or Can Export turned on). In Lofty or another CRM, use its report export.
- Get closings from where they really live.Your transaction system or your TC team's sheet. That's the record that knows what actually closed.
- Add any sheet you track by hand.A Google Sheet downloads as a file from File, then Download, then CSV. Name each file after where it came from.
- Trim the files.Delete phone and email columns before you upload. The tool only needs client names, agents, stages and dates.
- Paste the prompt into Claude.Attach the files, fill in the brackets, and send. Answer its mapping questions. Next Monday, upload new files and run it again.
Build me a weekly scorecard tool for my real estate team as an interactive artifact. About my team: - [NUMBER] agents
Build me a weekly scorecard tool for my real estate team as an interactive artifact. About my team: - [NUMBER] agents - Our CRM: [Follow Up Boss / Lofty / other] - Our transaction system: [SkySlope / Dotloop / a TC spreadsheet / other] - Other places we track numbers: [e.g. "a Google Sheet the coaches update"] - Our week runs [Monday to Sunday] - Coaches: [NAMES, or "none"] I'm attaching last week's data. Each file name says where it came from: 1. Agent activity from our CRM (calls, texts, appointments) 2. Deals or pipeline from our CRM 3. Closings from our transaction system or TC sheet 4. Any tracking sheet we keep by hand Our number definitions (use these, nothing else): [PASTE YOUR DEFINITIONS TABLE] Build a tool that does this: 1. Map before you count. Show me the columns you found in each file and which column you'll use for each number. Ask me which stages or statuses mean "agreement signed," "under contract" and "closed" in our systems. Agent names may be spelled differently across files; show me how you matched them. Don't guess. 2. Count last week for every agent: appointments, agreements signed, contracts, closings. Use the source of truth from our definitions for each number. Label every number with the file it came from. 3. When two files disagree about the same thing (the CRM says under contract, the transaction system says closed), don't pick one quietly. Use our source of truth and list the disagreement. 4. List what makes a number wrong, per agent: - an appointment in the past with no outcome - a deal past its close date that's still under contract - a closing in the transaction system with no closed deal in the CRM - a deal with no price For each one: the client's name, the problem, and where to fix it. 5. Add a row per agent for this week's commitment (calls, conversations, appointments) that I can type into. 6. Write a short Monday message per agent: last week's numbers, anything to fix, this week's commitment. One button copies it. 7. Put team totals at the top. Sort agents by how many records need fixing, most first. Rules: - Never invent or estimate a number. If a column is missing, say which number can't be counted and why. - Keep the data inside the artifact. - I'll upload new files every Monday, so make it easy to run again.
Unlock the full prompt. Free. You'll also get every new blueprint Monday at 3 PM, a day before it's public.
What it won't do: Run itself. It's a snapshot of the files you give it, so it's only as current as your last export. It can't watch your systems between Mondays, and it can't stop a report from sending. That part is what the platform does.
What a sheet can't do
A sheet, or the kit, can hold the definitions and count a week. It can't pull the numbers on its own, catch the wrong record the day it happens, or hold the report until it's fixed. Doing that for every agent every week is the job we built the platform to take off a leader's plate. That part is in the beta.
Questions owners ask
What should be on a real estate agent's weekly scorecard?
Last week's appointments, agreements signed, contracts and closings, each with one written definition and one source. Then this week's commitment: calls, conversations and appointments. Keep it to numbers the agent controls and the coach can check.
Should closings come from the CRM or the transaction coordinator?
The transaction team's records. On Ken Pozek's team the CRM stage lagged reality, so an agent who closed two sides showed zero. The TC team's records are always right about what closed.
Should speed to lead be an average or a median?
Pick one and write it down. We use the median of the last 30 days on working hours, because one lead that lands on a Saturday night shouldn't wreck an agent's number. What matters most is that every screen uses the same one.
What happens when an agent's numbers are wrong?
The report shows the record that makes it wrong, with what to fix, and doesn't send until it's fixed. Without software, the coach checks the fix list before the 1:1 instead.
Do I need software to run this?
No. The definitions page and the four-part report work on paper or in a sheet. The AI build kit gets you a working tool in Claude from your own files, whether they come from Follow Up Boss, Lofty, your transaction system or a Google Sheet. The platform does it automatically every week.