Annual Retrospective Guide
A year review that changes next year's plan
The problem
Most year-end reviews are highlight reels: revenue went up or down, a few wins get toasted, and the same plan gets written again with bigger numbers. What never gets examined is the decisions. Which calls were skill and which were luck, what you believed in January that turned out false, and which habit quietly cost you the most. A review that skips those questions can't change anything, which is why so many businesses have the same year twice. This prompt audits the decisions, not just the results, and refuses to end until something in next year's plan is different because of it.
The tool
You are a retrospective facilitator with a firm rule: a review that
doesn't change the next plan was entertainment. You care less about
what happened than about why the owner's predictions missed, because
that gap is where next year's plan goes wrong too.
MY YEAR IN RAW MATERIAL: [REVENUE VS LAST YEAR, WHAT YOU PLANNED IN
JANUARY IF ANYTHING, BIG DECISIONS MADE, THINGS THAT SURPRISED YOU,
TEAM CHANGES, ANYTHING ELSE YOU HAVE]
NEXT YEAR'S DRAFT PLAN, IF ONE EXISTS: [PASTE IT, OR SAY "NONE YET"]
STEP 1, PREDICTION AUDIT: Extract every belief my January plan rested
on (or reconstruct what I evidently believed, if I had no written
plan) and grade each: CAME TRUE, FALSE, or UNTESTED. For the false
ones, ask me whether the miss was information I couldn't have had or
information I didn't gather. Be blunt about which. This calibration
matters more than the revenue number.
STEP 2, DECISION LEDGER: Have me name the year's 5 biggest decisions,
including the passive ones (kept the same prices, didn't replace the
weak hire, stayed out of that market). For each: what I knew at the
time, the outcome, and your honest sort into GOOD CALL, BAD CALL, GOT
LUCKY, or GOT UNLUCKY. Judge the decision by what was knowable then,
not by the outcome. Then name any repeating pattern across the bad
calls in one sentence.
STEP 3, FORCE THE PLAN TO CHANGE: Deliver the retrospective's output
as edits, not reflections: 2 things next year's plan must ADD, 2 it
must DROP, and 1 belief it must state explicitly so we can grade it
next December. Tie every edit to a specific finding from Step 1 or 2.
If I gave you a draft plan, mark exactly where each edit lands in it.
Close with the January action list: 3 items, owners, dates.
RULES: Do not produce a summary of the year; produce changes to the
plan. Refuse "work harder" or "stay focused" as edits; every edit
names a number, an offer, a role, or a process. If my raw material
is all wins, ask me what I would call my biggest mistake if I had to
name one, and don't proceed until I do.How to use it
- 1Book 90 quiet minutes in December or early January, and pull three things first: this year's monthly revenue, last January's plan or goals if they exist, and your calendar.
- 2Skim the calendar before you start; it surfaces forgotten decisions and surprises that memory smooths over.
- 3Resist defending yourself in Step 2. The sort into good call versus got lucky only works if you report what you knew at the time, not what you know now.
- 4Take the Step 3 edits and physically change the plan document the same day. An edit list saved in a chat log changes nothing.
- 5Save the "belief stated explicitly" line somewhere you'll find it next December. Grading it becomes the first input to next year's retrospective.
- 6If the retrospective surfaces a serious directional question, run it through a pivot framework separately rather than answering it inside a year review.
Example
Input: a 10-person renovation contractor, revenue flat at $1.4M against a planned $1.7M, hired two crew members in spring, passed on buying a competitor's book of business in June.
Sample output excerpt:
PREDICTION AUDIT: January belief: "Kitchen remodel demand holds at 2023 levels." FALSE, and knowable: three suppliers mentioned slowing orders in Q1 and it went unexamined. January belief: "Two new hires reach full billing by month two." FALSE: they reached it in month five, which alone explains roughly $80K of the plan miss. Show the math: 2 crew x 3 extra months x ~$13K monthly billing target.
DECISION LEDGER: Passing on the competitor's book of business: GOOD CALL. At the asking price, you'd have paid 1.5x annual profit for customers with no contracts in a softening market. The pattern across the two BAD CALLs: both were staffing decisions made without a ramp-up estimate.
PLAN EDITS: ADD: a written 90-day ramp plan and billing target for any hire before the offer goes out (from the hiring miss). DROP: the blanket 20% growth target; replace with a bathrooms-and-aging-in-place target where your quote win rate ran 55%. STATE THE BELIEF: "Remodel demand recovers by Q3." Put it in the plan, dated, so next year grades it.
Pro tip
Do the decision ledger for the year's passive decisions first, the things you didn't change. Owners audit their actions but almost never their defaults, and a default that survived four quarters unexamined, like pricing, is usually carrying the biggest hidden cost in the business.
Related tools
Want this customized and automated for your business?
We take the tools in this toolbox and wire them into your business: your data, your brand voice, running on autopilot.
Talk to Vexlo