Customer Value Added: Two Meanings, and the Number That Splits the Budget

An average is biased, and it is biased the same way every time. Knowing which way is what lets you read one safely.

The bias comes from the top of the list. An average is dragged upward by a customer's best events, and the best events are the ones that were funded first and would be funded under any sensible plan. They are not in question. The events actually under discussion sit at the bottom of the list, and their contribution to the average is diluted by everything above them.

So a big program flatters the case for making it bigger. A customer with twelve events and one brilliant January feature will show a comfortable average long after events nine through twelve have stopped paying.

The same trap catches anyone who reads a track record as a forecast. A fund manager's lifetime return is a poor guide to next quarter for the same reason, and the fix is the same one. Judge the decision you are actually taking, which is the next increment, using the number that describes the next increment.

None of this is a promotions idea. Marginal analysis is the standard method for allocating any budget across competing uses, and it appears in the retailing canon in exactly this form: keep spending while each additional dollar generates more than a dollar of incremental contribution.

Take one customer's promotional year. Every event has a cost and a return, measured the way Lesson 1 measured it: net profit over the money that left the business, with break‑even sitting at zero percent.

The average return pools them all together.

Average return = total net profit from all funded events / total money spent

The marginal return looks at the edge of the plan, where the money ran out. It has two sides, and the allocation decision uses both.

Looking back, the last event the budget paid for. That prices a cut: take money away and this is the return you lose first.

Looking forward, the best event the budget could not reach. That prices an addition: put money in and this is what it earns. This forward reading is what the word marginal means in economics, where the margin is always the next unit rather than the last one.

Marginal return, backward = the return of the last event funded
Marginal return, forward = the return of the best event not funded

Worked, on one account

MegaMart runs five events. Their returns are 75, 55, 30, 15 and 5 percent, and they cost $130k, $150k, $180k, $150k and $160k. Total spend is $770k and total net profit is $265k.

Average return = $265k / $770k = 34 percent. That is a perfectly respectable number and it is the one that appears in the review pack.

Its last funded event returns 5 percent, so cutting MegaMart's budget costs you 5 percent first. Its best unfunded event returns minus 5 percent, so adding to MegaMart's budget destroys money.

All three numbers are true. The 34 describes what already happened. The other two describe the decision in front of you, and they point the same way: this account should not get more.

You have one more $100k to place. Two accounts want it.

BudgetBarn is your largest customer by volume. Its promotional program this year averaged an 11 percent return across $720k of spend.

QuickStop is small, a convenience chain, and its program averaged a 101 percent return across $220k.

The intuition in the room will be QuickStop, and the intuition is right, but not for the reason anyone will give. It is not right because 101 beats 11.

Look at what the $100k would actually buy. At BudgetBarn, the next event on the list is a three‑week volume drive costing $270k and returning minus 5 percent. The $100k cannot even fund it, and funding it would destroy money.

At QuickStop, the next event is a weekend top‑up offer costing $80k and returning 45 percent. That is what the money buys, and it is an excellent use of it.

Now change one fact. Suppose QuickStop has already committed every meter of shelf and every hour of staff time it has, and physically cannot run another event. The 101 percent average has not changed. The marginal return is now undefined, because there is no next event, and the right answer flips to a third account you had not considered.

An average could not have told you any of that. A ranked list of what the money would actually fund told you all three answers in about a minute.

Three places this distinction shows up in the working week, and what to do about each.

The customer performance pack. Almost every trade review reports promotional return per customer as a single blended percentage. That number is an average. It is the right number for asking "was last year's money well spent at this account" and the wrong number for asking "should this account get more". Ask for the event list sorted by return, and read the bottom of it.

The customer who argues from their average. A large account will tell you its promotions return well above your portfolio average, and it will often be true. The reply is not to dispute the figure. It is to ask which specific events the extra money would fund, because the answer is always events that are not currently on the plan, and those are by definition the ones that did not make the cut.

The efficiency target. A blended return target set at the portfolio level can be hit by cutting the worst customers entirely and changing nothing anywhere else. That improves the average and can easily reduce total profit, because a customer returning 8 percent on $400k contributes more money than one returning 60 percent on $40k. Targets on averages reward the wrong move.

The handoff from Trade Terms

Trade Terms Lesson 4 builds a ceiling on what a customer's whole investment can return, and says explicitly that the ceiling bounds the average of the pot rather than the next dollar into it. This lesson is the other half of that sentence.