Making a Decision With Missing Information

Why this matters

The decisions that matter most in a shop almost never arrive with all the facts attached. A builder wants an answer on a subdivision by Friday. A good tech wants a raise or he walks. A supplier is discontinuing the line you stock. Owners handle the gap in one of two bad ways: they decide on gut and hope, or they keep gathering information until the decision gets made for them by the deadline passing. This card is the middle path: sort the unknowns into the few that would change your answer and the many that would not, buy the cheap information that matters, decide on time, and protect yourself against what you still could not learn.

Step 1: Write down the decision and its real deadline

State the decision as a choice between named options, and write the date by which it is actually forced, not the date you would like to have it settled.

Why: "Should we do more new construction" is not a decision; "accept, decline or counter the builder's offer on 40 homes by the 18th" is. The real deadline matters because it sets your information budget. Ten days allows a few phone calls. Two hours allows none.

What breaks if you skip it: without a deadline, information gathering has no end, and the decision drifts until someone else makes it.

The case used throughout: a 12-person HVAC shop with a two-person install crew is offered the heating and cooling installs for a 40-home subdivision by a builder it has never worked for. The builder wants an answer in 10 days. The options: accept, decline, or counter with conditions.

Step 2: List what you know, with where each fact came from

Write down the facts you have, each with its source. Source matters because a fact from your own records is worth more than one from the builder's sales pitch.

The case:

• 40 homes over roughly 8 months, so about 5 a month (builder's schedule)
• each home takes about 1.5 crew-days for rough-in and trim (the shop's own install times on similar tract homes)
• the install crew has about 21 working days a month (shop calendar)
• so the builder work would take about 7.5 crew-days a month (5 times 1.5), about 36 percent of the crew's month (7.5 of 21)
• the crew is booked about 85 percent in summer and about 55 percent in fall and winter on replacement work (dispatch records, last two years)
• the builder's price works out to about 60 percent of the margin per crew-day the shop earns on residential replacements (the shop's own estimate against the builder's price sheet)

Step 3: List what you do not know

Write every unknown, without judging yet which matter.

The case:

• whether the builder pays on time
• whether the builder's 8-month schedule will hold
• whether equipment prices will rise during the job
• whether the builder will bring more subdivisions if this goes well
• how the crew feels about tract work
• whether the inspector in that town is slow

Step 4: Sort the unknowns into swing and background

For each unknown, ask: if I learned the bad value of this, would I choose differently? If yes, it is a swing unknown. If the answer would hold either way, it is background: interesting, possibly worth knowing later, not worth spending the deadline on.

Why: most of the anxiety in a decision comes from the background list, which is long. The decision usually turns on two or three items. Separating them is what makes a deadline survivable.

The case:

Three swing unknowns. Note that the schedule only became a swing unknown once the capacity numbers were run. Before that it looked like background.

Step 5: For each swing unknown, decide whether to buy the information

For each swing unknown, ask two questions: can it be learned before the deadline, and what does it cost to learn? Information that is cheap and swing gets bought immediately. Information that is expensive or unavailable gets handled in step 7 instead.

The case:

• Payment: cheap and available. The owner calls two subcontractors who worked the builder's last project (about an hour) and asks his equipment supplier's credit manager, who sees how builders pay their accounts. Both subs say the builder pays reliably at about 45 days. The credit manager confirms no trouble on file.
• Schedule: partly available. One of the subs mentions the last project slipped about two months. That is history, not a forecast, but it makes a summer overlap likely rather than possible.
• Equipment prices: not knowable. No phone call reveals what a manufacturer will announce in month 5. This one gets protected, not researched.

What breaks if you skip it: owners either buy nothing (deciding blind on the payment question that took one hour to answer) or try to buy everything, including the unknowable one, and run out the clock.

Step 6: Write the default decision before the information comes in

Before the calls come back, write what you would decide if you learned nothing more. This is your default.

Why: it gives the new information something to move. Without a written default, each new fact gets absorbed into whatever you were leaning toward that day. With one, you can see exactly which fact changed your mind and whether it should have.

The case: the owner's default, written on day 2, is "decline unless payment is confirmed reliable, because thin margin plus slow payment is the one combination the shop cannot absorb."

Step 7: Decide on time, and protect against what you still do not know

When the deadline arrives, decide with what you have. Then, for each swing unknown you could not resolve, build in a protection: a contract term, a staged commitment, a trigger to revisit.

The protection is the point. You cannot make the unknown known, but you can change the decision so a bad value hurts less. The general tool is converting an all-in commitment into one you can step back from. See related: Reversible and Irreversible Decisions in a Small Shop.

The case: payment came back good, which clears the default's condition. The schedule and price risks remain. The owner counters rather than accepting as offered:

• Staged commitment: the shop takes phase 1, the first 12 homes, scheduled for fall and winter when the crew has room, with an option on the remaining 28 that either side can decline after home 12.
• Price protection: a clause passing through documented manufacturer price increases on equipment above an agreed threshold. Before signing, the owner has his attorney read the clause and the builder's standard subcontract, because the builder's paper will contain terms about payment timing, backcharges and lien rights that he should not interpret alone.
• Summer limit: in writing, the shop commits to no more than 2 homes a month from June through August unless it adds a third installer. At 2 homes, that is 3 crew-days, about 14 percent of the month (3 of 21), which brings summer to about 99 percent: tight but workable.

The builder accepts phase 1 and the summer limit, and negotiates the price clause to a higher threshold.

Step 8: Set a review date and the signals that would reopen the decision

Write a date, and the specific observations that would make you reconsider before that date.

The case: review after home 12 or at 90 days, whichever comes first. Signals that reopen it sooner: any invoice unpaid past 60 days; the builder pushing more than 2 homes a month into summer; trim-out running over 2 crew-days per home on three homes in a row.

What changes how much information you need

How reversible the decision is. A decision you can undo cheaply deserves less information-gathering, and a fast call is usually right. A one-way decision (signing a long lease, a large equipment purchase, ending a key relationship) deserves more of the deadline spent on swing unknowns, and more protection built in at step 7.

How much the swing unknowns matter relative to each other. If one unknown dominates (in the case, payment risk could have sunk the shop on its own), spend nearly the whole budget on it. If three unknowns are roughly equal, spread the effort and lean harder on protections.

Whether waiting has a cost. Sometimes the missing information will simply arrive if you wait a week, and waiting costs nothing. Then waiting is a legitimate option and belongs on the list in step 1. When the builder will give the job to someone else on day 11, waiting is not free, and the cost of not deciding has to be counted against it. See related: The Cost of Not Deciding.

Two traps this procedure is built to catch

Researching the background list. It feels productive to learn about the inspector, the builder's other projects, the subdivision's sales pace. None of it would have changed the answer. Hours spent there come straight out of the time available for the payment question.

Treating "not yet known" as "unknowable" or the reverse. The payment question felt unanswerable until the owner realized two subs and a credit manager could answer it in an hour. The price question felt answerable ("I will ask my rep") until it was clear no rep knows what the manufacturer will announce in five months. Getting this sorting right is most of the skill.

How to verify you got it right

At the review date, look at each swing unknown and write down how it actually resolved, next to what you assumed. In the case, at home 12: payment averaged 41 days (better than the 45 the subs reported), the schedule had already slipped 3 weeks, and one price increase had passed through under the clause. The protections earned their keep on the two unknowns that went the wrong way. If, at review, the unknowns that hurt you were ones you had marked as background, your sorting test in step 4 is too loose. See related: Keeping a Decision Journal.

References

• Howard, R. A. (1966), "Information Value Theory," IEEE Transactions on Systems Science and Cybernetics, the origin of valuing information by whether it can change the decision
• See related: Reversible and Irreversible Decisions in a Small Shop; The Cost of Not Deciding; What Would Have to Be True; Keeping a Decision Journal