Prepare a quarterly review, both kinds

The situation

Quarter end. You owe a customer a business review, and you owe yourself an honest look at the quarter you just had. In most teams both are built the same way: somebody exports to a spreadsheet on the Sunday, and the numbers never quite reconcile with the ones from last quarter because the export was filtered slightly differently.

What must be true when you are done

• Every number came from the same place and can be reproduced next quarter.
• The customer review says what happened, not what you sold them.
• You know what you got wrong, not just what you got.
• Both end in commitments with dates rather than slides.

The customer review: what actually happened

Ask for the account's own quarter. What you want is not a usage chart, it is the story of the relationship, which is already on the record if you have been logging calls.

The part to look at hardest is "whether we did it". Completed tasks are a record of your own promises kept, and going into a customer review knowing you dropped two things in June is considerably better than finding out across the table.

Be honest in the deck about the ones you dropped. It costs you almost nothing and it is the difference between a review the customer trusts and a presentation they sit through.

Your own quarter: the four questions

Ask these in this order, because each one makes the next one mean something.

What closed. Won and lost, by value, and the count of each.

Why the losses lost. From a fixed list, so it is countable. If the biggest bucket is "no decision", the problem is qualification rather than product or price, which is a different fix entirely.

What moved and what did not. Deals that changed stage versus deals that sat still all quarter. A pipeline that grew without moving is a pipeline with a stalling problem, and the total hides it.

What you said in January. The uncomfortable one, and the only one that improves you.

That is answerable because the history holds it. Every change to a close date is recorded with who moved it and when, so "we thought this was a Q3 deal in April, and it has moved twice since" is a fact rather than a recollection. Do this for two quarters and you will know your own optimism as a number.

Same source, every time

The reason quarterly numbers never reconcile is that they are rebuilt from scratch each time, by a person making slightly different choices about what counts.

Write the questions down once, as the exact instructions, and re-run them verbatim. The numbers come from the database over every matching record rather than from a page somebody fetched, so the same question genuinely gives the same answer. When a number moves, it moved because the business moved.

Keep the instructions next to the deck. In a year the person doing this will be somebody else, and the most valuable thing you can leave them is the exact wording rather than the spreadsheet.

Where the reviews meet

Do the customer reviews first. Ten of them will tell you something the aggregate cannot: the same objection in three unrelated accounts is a pattern, and it will not appear in any grouping because it lives in the notes.

That question is why the summaries are worth writing properly. It is unanswerable from a stage field and trivially answerable from a quarter of one-paragraph readings.

Verify it

• Reconcile against last quarter. The prior quarter's closed number should still be the same number. If it moved, something is being counted by when you asked rather than by when it happened.
• Check the lost reasons add up. Losses without a reason are the ones that will make next quarter's analysis useless.
• Make sure both reviews end in tasks. A review that ends in a deck ends. One that ends in five dated commitments is the reason to have held it.

Run it again

Quarterly, from the same written questions. The compounding value is entirely in asking the identical thing every time: the first run is a snapshot, the fourth is a trend, and the eighth tells you things about how you sell that no individual quarter could.