The boot is gone. Its profit is still in your ad report.

Follow a sold-out boot through the baskets behind two still-running ads. Then test how much replacement profit each would need.

Skip to the investigation

An illustrated stockout investigation

The boot is gone.
Its profit is still in your ad report.

A stocked product can lose the sale that made its advertising worth paying for. Follow the orders before you decide what to do with the ad.

An ochre boot and a teal shoe stand on either side of an empty shop plinth. Paper receipts curl from both occupied displays toward the space where a boot has left a dusty outline.

Monday’s stock email is admirably short: Wide-fit boot sold out. No backorders.

You update its availability and stop its direct advertising. The standard-fit boot is still on the shelf. So is the city shoe. Their Shopping ads can keep running, and their old numbers give you a reason to let them: each brought in $8,400 of attributed gross profit on $4,000 of spend.

Then you open the orders.

Some people who clicked the standard boot bought the wide boot instead. So did some of the city-shoe visitors. Both ads had been taking credit for a product the shop can no longer sell.

This is an invented shop and ledger, built to make the decision visible. All amounts are US dollars. Gross profit here means item revenue minus cost of goods; ad spend, fulfillment and other operating costs have not been deducted.

Before the stockout

Two doors. The same missing sale.

Each ad bought 10,000 clicks for $4,000. Each received credit for 160 complete orders. For clarity, every order in this ledger belongs to just one ad.

Clicked ad

Standard-fit boot

Still in stock

100 complete orders

Standard boot1 pair per order · $48 gross profit$4,800

60 complete orders

Wide boot + socks$50 + $10 = $60 per order$3,600

Attributed gross profit $8,400

Clicked ad

City shoe

Still in stock

100 complete orders

City shoe1 pair per order · $48 gross profit$4,800

60 complete orders

Wide boot + socks$50 + $10 = $60 per order$3,600

Attributed gross profit $8,400

The arrows group orders by the advertised product. They do not claim the ad caused the purchase. Open the complete constructed ledger (CSV).

The boot appears on two ads’ receipts. Each receipt also contains a pair of socks. Keep both purchases in sight as you change the return.

Account for each purchase

Remove the boot. What stays?

Start with one ad’s old $8,400. Imagine another 10,000 clicks for $4,000, with the other orders unchanged. First remove the item the shop cannot sell.

$4,800Other orders

$3,00060 wide boots

$600Their 60 socks

Old return: $8,400 ÷ $4,000 = 2.10 gross-profit dollars per ad dollar.

Every dollar left in the return needs a purchase behind it.

If the socks still sell, that is a basket to defend. Now put back a purchase somebody might actually want.

The small catalogue

What can a replacement order earn?

Standard boot$120 − $72 cost$48 GP

City shoe$100 − $52 cost$48 GP

Wide boot$150 − $100 cost$50 GP · Sold out

Roomy trail boot · Trail 2$150 − $85 cost$65 GP

Clearance wide shoe$90 − $85 cost$5 GP

Socks$20 − $10 cost$10 GP

One unit means one pair. These prices and costs are invented. All products except the wide boot remain available in the example’s relevant sizes.

Who will actually wear the boots?

Our invented merchandising lead has something better than a margin to offer. During an earlier shortage, eight customers who asked for the wide boot tried the trail boot with staff help. They tried the current Trail 2 at the full $150 price, in sizes 8–11—the sizes the shop still has. Six kept it through the return window; two returned it because the fit was wrong.

One of those six has a useful note in the fitting log: room across the toes without the heel lifting on the shop’s short ramp. That buyer preferred the trail boot’s hold to their familiar wide boot’s. The two returns mention pressure across the instep after longer wear.

I would investigate a replacement someone preferred for its fit. The returns show whose problem it may fail to solve. Six kept pairs cannot become a 75% recovery rate for unassisted Shopping visitors.

I would put that fit difference on the landing page beside Trail 2, the sizes still available and the $150 price. The store kept those sales without a discount. If online visitors need one, or the needed size is gone, the same story no longer earns the same basket.

The retailer thinks standard-boot visitors might accept the roomy trail boot with socks: $75 of gross profit per replacement basket. For the city-shoe visitors, the retailer expects the clearance wide shoe with socks: just $15.

Those are two forecasts, not discoveries in the old ledger. The catalogue is available to both groups. Maybe the city-shoe visitors will love the trail boot. Maybe nobody will. The point of stating a replacement basket is to expose the bet you are about to fund.

Try the retailer’s bet

The same gap asks very different things.

For each ad, suppose 60 future orders would have included the missing boot. Choose the complete purchase you expect instead, and how many orders contain it. These are modeled purchases, never an identifiable later customer segment.

The shop chooses $1.50 GP per ad dollar—an operating requirement, not a Google bidding target or universal break-even point. Each ad pays $4,000 for 10,000 clicks and needs $6,000 GP. Other baskets supply $4,800. The replacements have to earn the remaining $1,200.

Try both stories. We will follow your standard-boot choice into the arriving ledger.

Standard-boot ad

What do they buy instead? How many of those 60 orders? 30 replacement orders · 50%

$1,050 of room

$4,800 from other baskets + 30 replacement orders at $75 = $7,050 GP.

Needs 16 replacement orders: 26.7% of the modeled 60.

GP per ad dollar: 1.76

City-shoe ad

What do they buy instead? How many of those 60 orders? 30 replacement orders · 50%

$750 short

$4,800 from other baskets + 30 replacement orders at $15 = $5,250 GP.

Needs 80 replacement orders: 133.3% of 60. Even all 60 cannot meet the requirement.

GP per ad dollar: 1.31

The controls need JavaScript. The displayed starting case is complete: at 50% recovery the standard-boot ad returns 1.7625 and the city-shoe ad 1.3125. At 100%, they return 2.325 and 1.425 respectively.

Bring your bet to the orders

The basket story meets a ledger.

Stay with the standard-boot ad. The left column carries your live choice above. On the right, suppose a later cohort bought 10,000 clicks for $4,000 and produced these 126 retained orders. This is still an invented shop; assume this ledger’s attribution and return adjustments are complete.

Your standard-boot model: trail boot + socks, with half of 60 modeled orders becoming replacement orders.

Your model can change. These arriving orders stay fixed.
Complete basket	Your current model	Arriving ledger
Other orders	$4,800 GP	100 standard boots $4,800 GP
Trail boot + socks $75 per order	30 orders $2,250 GP	14 orders $1,050 GP
Clearance wide shoe + socks $15 per order	0 orders $0 GP	12 orders $180 GP
Socks only $10 per order	0 orders $0 GP	0 orders $0 GP
Total GP	$7,050	$6,030

Modeled order counts are expectations and can be fractional. Dollar amounts are rounded; displayed rows can differ from the total. Revise your standard-boot model ↑. Open all 126 arriving orders (CSV).

The ledger includes 12 clearance baskets your all-trail premise did not allow.

Arriving GP is $1,020 below your modeled total.

We can count 26 alternative-product orders and their $1,230 of GP. We cannot divide 26 by the model’s 60 and call it recovery. Some shoppers might have chosen the trail boot with the wide boot still in stock. The ledger shows what arrived, not which sales were rescued.

The ledger’s historical return stays $6,030 ÷ $4,000 = 1.5075. Attributed profit does not establish incremental profit caused by advertising.

What would you pay for another 10,000 clicks?

Compare your current model with a simpler expectation: these arriving baskets repeat.

Assumed price of the next 10,000 clicks

This price also updates the model above. It compares the same 10,000 clicks; it does not predict what a budget cut will buy.

Change the model’s other baskets and the shop’s requirement

Other-basket GP in your model, relative to $4,800 100%

Required GP per ad dollar 1.50

These controls also update the model above. The arriving ledger and a repeat of its baskets keep their own $4,800 of other-order GP. Changing your model does not change those orders.

Required replacement orders = max(0, required GP − other-basket GP) ÷ replacement-basket GP. Dividing by 60 gives required recovery. Fractional counts describe an average over repeated cohorts.

Your current model$1,050 above requirement

If arriving baskets repeat$30 above requirement

The arriving mix narrowly earns the next cohort at this price. It does not earn the richer story.

My judgment

The fittings give the trail boot a reason to be offered. They do not make $30 of room comfortable enough for me to risk another $4,000 unchanged. I could defend a smaller exposure only with a reason to take it and an amount the shop could afford to lose. Refusing the spend would not make the boot a bad replacement.

The basket and your spending choice

The trail basket promises $75 GP. The arriving mix earned $6,030: only $30 of room for another $4,000 cohort. Take both into the next stockout conversation.

With the starting premise and $4,000 price, the forecast exceeds the requirement by $1,050. Repeating the arriving ledger exceeds it by only $30. At $4,200, that repeat falls $270 short. Recording your own decision needs JavaScript.

A confirmed delivery could justify more spending: a deliberately tolerated shortfall until the wide boot returns, with a spend limit and ending date. “Soon” cannot do that work. A slipped delivery would remove the reason for that exception.

The boot may be worth offering. The next spend still needs your yes.

For your own shop. Start with complete, deduplicated orders, sold-item IDs and quantities, net item revenue and reliable COGS, tied to the same attribution window and advertised-product basis. Google documents cart-data metrics and cross-sell reports; missing COGS can make reported profit incomplete. Keep observed order evidence and your substitution assumptions distinguishable. Availability should reflect whether the specific variant can actually be ordered.

Where the order data and requirement come from

Google’s conversions with cart data can distinguish what was advertised from what was bought. This investigation additionally needs complete baskets on one consistent attribution basis; its shop-to-basket picture is our construction, not a native Google report. Gross-profit reporting needs cost-of-goods data.

The shop starts with a chosen requirement of $1.50 GP per ad dollar. It is an operating choice, not a Google bidding target or universal break-even point. It establishes the arithmetic threshold; it does not establish a willingness to risk the spend.

Back to the shop ↑