What Would Have to Be True

Why this matters

When an owner and a partner, or an owner and a service manager, argue about a big move, the argument usually goes in circles. One person lists reasons it will work, the other lists reasons it will not, and whoever is more stubborn or more senior wins. Nobody changes their mind because nobody is asked to. What would have to be true is a question that ends the circle. Instead of arguing whether an option is good, you list the conditions under which it would be the right choice, find the one you are least sure of, and go test it. The argument turns into a short list of things to check, and the check usually settles it.

Where the question comes from

Roger Martin developed the approach with A.G. Lafley at Procter and Gamble and described it in their 2013 book Playing to Win. Their insight was that strategy debates stall because people argue conclusions. If you ask "what would have to be true for this option to be the best one," people who disagree about the conclusion can usually agree on the conditions, and then the disagreement becomes about a fact you can go find out.

The method is worked here on one decision, start to finish: opening a second location.

The situation

An 11-truck residential electrical shop has grown about as far as its home town allows. Calls are steady but the market feels saturated. A town 45 miles away is growing fast, and the owner has wanted a second location there for a year. His operations manager thinks it is a mistake. They have had the argument four times.

Step 1: Frame the choice as two to four genuinely different options

Not "open the branch or not." Name real alternatives that each solve the underlying problem, which here is "where does growth come from."

Why: with only one option and its absence, the conditions you list are just the case for and against that one plan. With several, you can see which one's conditions are most likely to hold.

The three they settle on:

• A. Branch: open a small office in the new town with 3 trucks and a lead tech who lives there.
• B. Commute: serve the new town from the home shop, with trucks driving out on scheduled days.
• C. Deepen: stay put, raise prices modestly and push generator and EV charger installs in the home market.

Step 2: Suspend the argument

Agree a rule before listing anything: for the next hour, nobody argues for or against any option. The only allowed contribution is a condition that would have to be true.

Why: an owner who has wanted the branch for a year will hear every doubt as an attack. A doubt reframed as "for A to be right, X has to be true," with no claim about whether X is true, is something he can write down himself. It moves the conversation from advocacy to inquiry.

What breaks if you skip it: the list fills up with the operations manager's objections to A and the owner's objections to C, and you are back to the argument with a whiteboard.

Step 3: For each option, list what would have to be true

Work through the areas that decide whether a shop move succeeds: demand, people, operations at home, cash, and any rule or license. Each condition should be specific enough to check, with a number where one belongs.

For the branch, the list they write, and how they mark it in step 4:

OPTION A: branch in the new town, 3 trucks

What would have to be true                Status
1 About 120 jobs a month within 20 miles  DOUBTFUL
  of the site by month 12
2 A lead tech will run it and live there  DOUBTFUL
3 Home shop runs 2 days a week without    PLAUSIBLE
  the owner on site
4 Cash covers the branch's shortfall for  KNOWN
  12 months without the credit line
5 Local contractor registration done      KNOWN
  within 30 days (same state)

Barriers: 1 and 2. Test those first.

The 120 comes from the shop's own numbers: 3 trucks at about 2 jobs a day for about 20 working days a month. A branch that cannot fill its trucks is a branch that loses money every month.

For B, commute:

• enough demand to fill one truck's scheduled days there
• customers in the new town accept a scheduled day rather than same-day service
• the windshield time (about 90 minutes round trip per truck per day) is tolerable to the techs and the margin

For C, deepen:

• the home market absorbs a price rise of about 8 percent without the call count falling more than a few percent
• there is enough local generator and EV charger demand to add a real share of the growth the owner wants

Step 4: Mark each condition, and find the barrier

Mark each condition known (you already have evidence it is true), plausible (you think so, without evidence) or doubtful. A barrier condition is one you are least confident in that would sink the option if false.

Why: you test the barriers first because they are where options die. Testing a condition you are already confident in feels productive and changes nothing.

For A, the barriers are demand and the lead tech. For B, the barrier is customer acceptance of scheduled days. For C, the price-rise condition is doubtful to the owner and plausible to the operations manager, which is itself useful: it is the one they disagree on, so it is the one to test.

Step 5: Design the cheapest test for each barrier, and set the pass mark first

For each barrier, find the least expensive way to get evidence, and write down what result would count as passing before running it.

Why the pass mark comes first: without it, whatever result comes back gets read as support for what each person already believed. See related: Running a Small Experiment Instead of Arguing.

The tests:

• A, demand. The shop's call records already hold an answer: over the past 12 months, 214 calls came from the new town's area, about 18 a month, and the office turned most of them away as too far. To see what active marketing adds, they run six weeks of local ads and a local listing pointing at the shop's number. Pass mark: a pace that plausibly reaches 120 jobs a month within a year, which they set as at least 60 calls a month by the end of the test.
• A, lead tech. Ask the two techs who could run it, directly and privately. Pass mark: one says yes to running it and living within 20 minutes of it.
• B, customer acceptance. During the ad test, the office offers every new-town caller the next scheduled day rather than a same-day slot. Pass mark: at least 7 in 10 accept.
• C, price. Raise prices 8 percent on new quotes only, for one quarter, and watch close rate. Pass mark: close rate falls by no more than 3 points against the same quarter last year.

Step 6: Run the tests and let the conditions decide

The results, about two months later:

• A, demand: calls from the new town rose from about 18 to about 40 a month. Well short of the 60 pass mark. Barrier failed.
• A, lead tech: one tech says yes to running it but will not move; he would commute. The other says no. Barrier failed, in a specific way: the condition was "live there," and nobody will.
• B, acceptance: 31 of 40 new-town callers in the final month took the next scheduled day, about 78 percent. Passed.
• C, price: close rate fell 2 points on the higher quotes. Passed.

Option A is out, and the owner is the one who says so, because it failed on conditions he wrote down himself.

The decision that follows

About 40 calls a month in the new town is far too few for a 3-truck branch but enough for part of one truck. The shop books about three in four calls, so 40 calls is about 30 jobs. At the same 2 jobs a day the branch numbers used, one truck spending about 15 days a month there (three or four days a week) has room for about 30 jobs, matching the demand the test showed. The windshield time comes to about 90 minutes on each of those 15 days, about 22 hours a month (15 times 1.5), which the owner judges acceptable as the cost of building a presence.

So they do B at a small scale plus C: one truck in the new town three to four days a week, and the 8 percent price rise at home. They also write a condition that would reopen A: when new-town calls run above about 100 a month for three straight months, revisit the branch, starting with the lead-tech condition, which is the one most likely to fail again.

What changes how you run it

When the barrier cannot be tested cheaply. Some conditions have no small test: whether a lease in a particular building will turn out well, whether a market will keep growing. Then the move is protection rather than testing, a shorter lease or an exit clause, and the option should carry a visible discount for the condition nobody could check.

When the conditions for every option look doubtful. That is a finding, not a failure. It means the options are wrong, and the right next step is to generate new ones, not to pick the least bad.

When one person controls all the information. If only the owner knows the cash position, the cash condition will be marked known by the person who wants it known. Have someone other than the option's champion mark each condition, or at least confirm the evidence behind each "known."

How the method goes wrong

Conditions too vague to test. "Customers will want us there" cannot be checked; "60 calls a month by the end of a six-week ad test" can. If a condition has no number and no observable result, rewrite it until it does.

Testing the comfortable conditions. The temptation is to confirm the conditions you already believe (cash, licensing) and call the option validated. Barriers first, every time.

Moving the pass mark after the result. Calls reaching 40 against a pass mark of 60 is a fail. The moment someone says "40 is pretty good considering," the method has turned back into the argument it was meant to replace.

References

• Lafley, A. G., and Martin, R. L., Playing to Win: How Strategy Really Works (2013), the origin of the "what would have to be true" question
• See related: Running a Small Experiment Instead of Arguing; The Second Location That Looked Certain; Running a Premortem Before a Big Commitment; Making a Decision With Missing Information