Seventeen Weeks Left on the Agreement
January. An account signs for 10,000 units at tier pricing and releases against a blanket PO through the year. Good deal, well negotiated. Somebody got credit for it.
It is September. They have taken 4,100.
Nobody knows that. The account orders every month, pays on time, calls when something ships late. Every signal anyone actually watches says the relationship is fine. The one number that says otherwise is the number nobody is computing, because computing it means joining a quantity that lives in a signed PDF to a quantity that lives in the ERP, and no report in your building does that
And it is not one deadline. It is three, stacked on the same date.
ONE DATE. THREE THINGS THAT HAPPEN ON IT.
The commitment: they hit the volume or they did not. If they did not, the price you gave them was priced on volume that never showed up.
The tier: twelve months of shipments went out at a rate that assumed a number. Short of it, you funded a discount you never earned back.
The renewal: most of these roll over unless somebody gives notice. Ninety days out from December 31 is October 2. Sixty days is November 1. Both are closer than they sound.
Here is the arithmetic, and it is the only arithmetic in this issue. September 3 to December 31 is 119 days. Seventeen weeks. Today is day 246 of the year, so 67 percent of the year is gone. That is your benchmark and it costs nothing to compute.
Put every account with a commitment on one chart against that line and the whole year resolves in about four seconds.
Two ways to be off, and they fail differently.
Behind pace. They are buying somewhere else. Not all of it, some of it, and the part they moved is the part you never see. The account still reorders its stock items, so it never trips a churn flag, never lands in an at-risk report, never generates a phone call. Corbett up there is 26 points down with four months to make it up, which is not going to happen, and no one has said the word out loud yet.
Ahead of pace. This one is conditional, and it turns on what your cost has done since January. Trumbull exhausts the committed quantity in October and keeps releasing at a price you set against last winter’s replacement cost. If your buy side has moved since then, every unit past the commitment ships on old economics and the agreement says you keep doing it until December. The bigger bill arrives in January. They come to the renewal having bought 15,000 against a 10,000 commitment, and they want the 15,000 price. They have the volume history, you never raised it once all year, and that is a hard argument to answer in the room.
This Week, Try This: Count the Commitments
Three parts. The first one is the part nobody has ever done, and it is the reason no report exists.
1. The tracker, and where it lives
The committed quantity is in a signed agreement. Sometimes a PDF, sometimes an email confirming terms, sometimes a Word file in a folder named after a rep who left in 2023. The shipped quantity is in the ERP. Two systems, no join, so the subtraction has never once been run.
Point Claude at the folder. Account, dates, committed quantity, tier breaks, whether it auto-renews and how many days of notice it takes to stop that. And for the three numbers that matter, the exact sentence each one came from, so a person can check it in four seconds instead of reopening the file.
Read the output before it goes anywhere. The first run is the one most likely to be wrong, and a wrong committed quantity poisons every number downstream of it. Fix what it got wrong, then paste it in. After that the commitments do not change, so nobody ever has to do this again.
Read every agreement in this folder. Return one row
per agreement.
These are annual pricing agreements, blanket purchase
orders and volume commitments between us and our
customers. Some are signed PDFs. Some are emails
confirming terms. Some are Word files.
For each one return:
- account: the customer company name
- customer_number: if it appears anywhere, else blank
- effective_date
- expiration_date
- committed_quantity: the number and the unit (units,
pounds, cases, dollars)
- tier_breaks: any quantity threshold that changes the
price, listed
- auto_renews: yes / no / unclear
- notice_days: days of written notice required to stop
a renewal, else blank
- source_quote: for committed_quantity, expiration_date
and notice_days, the exact sentence you took each from
Rules:
- Quote, do not paraphrase, in source_quote. If you
cannot find a sentence to quote for a field, leave the
field blank and say so. Never infer a number from
context.
- If a document is scanned or unreadable, return the
filename with status unreadable and move on. Do not
guess at its contents.
- A quantity stated as a range, a forecast or an
estimate is not a commitment. Return it and flag it
non_binding: yes.
- If two agreements cover the same account, return both
as separate rows. Do not merge them.
- Do not calculate anything. No percentages, no days
remaining, no pace. Extraction only.
2. The burndown sheet
Cloud sheet, shared drive. Same rule as the tracker two weeks ago and the same two reasons: an automation has to be able to write to it while somebody has it open, and a sales manager has to be able to open it without asking anyone.
Account, expires, days left, notice deadline, committed, shipped, percent consumed, percent of term elapsed, variance, status, owner. Two columns do the work.
Variance is percent consumed minus percent elapsed. Negative is behind. Past twelve points either direction, the row colors itself, which is what draws the chart you saw above.
Notice deadline is the expiration date minus the notice period from part one. This is the column that will surprise people. Some of those dates are inside the next four weeks.
Then the rule that keeps it from being a spreadsheet nobody opens, and it is the same one as last time. Every row gets an owner and a next date, including the healthy ones. A row reading “on pace, checked Sept 8, next check Oct 6” is finished. A blank one is the defect.
We built the template. Two tabs, the burndown itself and a setup tab with both prompts sitting on it. Download it, drop it in Drive or SharePoint. Nothing to sign up for.
Get the Commitment Burndown Sheet
3. The weekly recount
The ERP export refreshes. The commitments do not. So part one runs once and this runs every Monday.
Scheduled task in Claude Cowork, working folder pointed at wherever the sheet lives. Setup is the same path as the mailbox watch two weeks ago, so it is six minutes if you have done it before. Weekly is right here. Hourly would be theater.
Every Monday, do this.
1. Open the Commitment Burndown sheet in the working
folder and read the Burndown tab.
2. Open the newest sales-by-customer export in the same
folder. If there is nothing newer than your last run,
say so in one line and stop.
3. For each account already on the Burndown tab, update
Shipped qty from the export, for the agreement's term
to date. Do not add accounts. Do not touch Committed
qty, Expires, Notice deadline or Owner. A person
enters those.
4. Recalculate Percent consumed and Variance.
5. Post a summary with three lists and nothing else:
- Accounts whose Variance moved more than 5 points
since last week, either direction, with the old
number and the new one
- Any Notice deadline inside the next 30 days
- Accounts more than 12 points behind pace with fewer
than 90 days to expiration
6. If a list is empty, write one line saying so. Do not
pad it.
Run part one this week, on whatever agreements you can actually find. A partial list is still a list. Right now the number is zero.
An Honest Take
Here is what I actually think about volume commitments, and it is not the flattering version. Most of them are theater, and both sides know it.
The buyer commits to 10,000 to get tier pricing. Comes in at 6,000. And nothing happens, because no distributor is going to invoice a customer for a shortfall or sue them over one, and the buyer has watched that not happen at four other suppliers already. So the commitment is not a promise. It is a price-extraction device that works because suppliers do not count.
Which does not make the number worthless. It just does not do what the contract says it does. Its real value is that it gives you standing to start a conversation you otherwise have no reason to start. “You’re at 41 percent with four months left, what changed?” is a fair question in September and an accusation in December. You only get to ask it if you counted.
So the fix is not tighter contract language. Nobody enforces the language you already have. It is counting out loud, in front of the customer, twice a year. The supplier who turns up in September with the number is a different vendor than the one who turns up in December with a renewal.
One more, and this one connects to what we sell, so weigh it accordingly. The reason nobody counts is not that the math is hard. The math is subtraction. It is that the releases arrive as POs and get typed by hand, one at a time, by the same person doing everything else, and a number that is only ever typed is a number that never gets totaled. That is what we sell against. The logic holds regardless of who you buy it from.
The Bottom Line
There is a folder of signed agreements somewhere in your building. Every one of them holds a number somebody negotiated hard for, an expiration date, and probably a notice clause nobody has read since the day it was signed.
Seventeen weeks. The buyer on the other side already knows exactly where they stand, because their system counts it for them and it is the number their own boss asks about. The only question is whether you walk into that conversation holding the same sheet they are.