The 20-minute proposal retro

Most proposal post-mortems do not happen. Leulu & Co. described the failure mode plainly — the debrief ends, the document is published, people move on, and nobody asks at the next sprint planning what happened to last week’s action items. The retro is on the calendar, then the calendar reschedules it, then the team submits the next bid and the previous bid’s lessons are gone.

We run a 20-minute retro after every submit. Same ritual every time. It works because it is short enough that it actually happens.

The five questions

Run them in this order. Twenty minutes total — about four minutes per question.

1. What did we submit on time? A factual question. Did the bid go out at the right hour, in the right format, to the right portal, with the right attachments? Yes or no. If no, what stopped it.
2. What did we get wrong in the response? Not what the buyer thought — we usually do not know that yet. What did we ourselves see, on the read-through right before submission, that we wished we had caught earlier?
3. What did the KB miss? Which claims did we have to draft from scratch because no block existed? Which blocks were stale? Which blocks did the engine surface that we ended up rewriting?
4. What did we learn about this buyer? Capture intelligence we did not have at kickoff. Names, roles, evaluation patterns, win-theme reactions during the Q&A window. Whatever the bid surfaced about how this buyer evaluates.
5. What changes for the next bid? One named action per category, with one owner and one deadline. Not a list of ideas; a list of commitments.

Why these five and not a longer list

Longer post-mortem templates exist. They produce more comprehensive notes and they happen less often. The short list happens.

The five questions are the minimum that captures the four post-mortem categories worth tracking — execution (1), quality (2), KB (3), and intelligence (4) — plus the question (5) that turns the retro from a discussion into a commitment. Drop any of the five and the retro stops paying for itself.

The “one owner per action item” rule

The single rule that makes the retro stick is the owner rule. Every action item has one owner — not a team, not a “we should.” If no owner is named, the action item is not real and gets struck from the list.

This is uncomfortable in the meeting. It is what makes the action items happen. The Shipley canon’s white-team discipline folds the same rule into the formal review framework — every issue has an owner and a deadline before the meeting ends. Our 20-minute retro is the small-shop version of the same discipline.

What we do with the output

The retro notes go into three places:

• The bid record. Tied to the submission, viewable by anyone who works on a similar bid in the future.
• The KB. Action items in category 3 (what the KB missed) become specific block-update tickets. Stale blocks get a freshness flag; missing blocks get a creation ticket assigned to the most-likely SME.
• The capture file for this buyer. Action items in category 4 (intelligence) update the buyer’s capture record. The next bid we run for the same buyer starts with the new intelligence baked in.

When to run it

Day-after-submit, not the day of submit. The team is too tired on submission day to do a useful retro. Twenty-four hours later they have perspective; thirty-six hours later they are already writing the next bid and the lessons are starting to fade. The 24-hour window is the right slot.

If the buyer has not announced an award yet — and most haven’t, that comes weeks later — the retro is what we run anyway. Once the award announcement lands, we run a second, smaller retro keyed to the debrief: what changed between what we thought happened in our retro and what the buyer told us happened in theirs. That second retro is fifteen minutes and it usually surfaces one or two corrections worth saving.

Closing

The retro is not glamorous and it does not produce headlines. It is the loop that compounds. Every bid that runs it makes the next bid one increment easier; every bid that skips it leaves the next bid starting from zero on whatever the skipped bid would have learned.

Twenty minutes. Five questions. One owner per action item. Run it tomorrow on the bid you shipped this week.