Media buying · Inventory

Which product should get more ad budget when stock is limited?

By Mirow · October 23, 2026 · 6 min read

Give the next ad dollars to products that can add contribution and still fulfill the resulting orders. A product with better economics may deserve only a small increase if most of its stock is already needed for expected demand. A warehouse full of a weaker product does not make its advertising profitable.

Make the decision over a stated window: for example, the next 14 days. Estimate how many more units you can sell in that window, then compare the contribution each additional order could leave after media. Keep a separate decision for the budget already running.

Count the units an increase can actually use

Start with sellable stock at locations that can fulfill the advertised variant and destination. In Shopify, available inventory excludes committed, unavailable and incoming units. On-hand inventory includes committed and unavailable units, so it can overstate what a new customer can buy. Safety stock may already be classified as unavailable. Shopify inventory states.

For planning, use:

Other activity includes the campaigns you plan to keep running, email, repeat purchases and unpaid discovery. Estimate that demand for the same window, with upcoming promotions included. The buffer covers uncertainty in that estimate and fulfillment. Subtract only a buffer that has not already been excluded from available stock. Do not subtract existing committed orders a second time.

These are planning assumptions, not a platform inventory field. If the result is zero or negative, hold the increase and review the existing demand plan. An incoming shipment needs a separate receipt and fulfillment check; a supplier's expected arrival date does not create sellable units today.

Work the stock and cost table together

Here is a hypothetical 14-day planning example in USD, unrelated to Mirow client results. A brand has $2,000 of additional media budget to consider. Each order contains one unit of one product. Product A's next receipt is expected on day 21 and B's on day 28; C has no confirmed receipt. No replenishment is counted within the 14-day window.

Scroll horizontally to see all columns.

The next table assumes each extra order is caused by the additional advertising and neither replaces another product purchase nor brings forward an order from a later period. That is a simplifying assumption for the calculation, not something an attributed purchase report establishes. Net sales are after discounts and exclude tax; customers pay no shipping. Variable costs include product cost, payment fees, packaging, pick-and-pack and outbound shipping. Assume no returns or cancellations here; a real plan must account for their effect on both sales and costs.

Scroll horizontally to see all columns.

Under these assumptions, allocate $250 to A and $1,750 to B. A can support 10 extra units: 10 × $25 of media = $250. B's allocation models 70 extra units, within its 200-unit allowance. C receives no increase: its assumed media cost exceeds the $15 available before media.

The modeled 80 additional orders produce $6,600 in net sales and $3,300 before media. Subtracting the full $2,000 increase leaves $1,300 in contribution after media, a 19.7% margin on those additional sales. Fixed overhead, agency fees and new creative production are outside this calculation. Deduct any extra costs needed to execute the plan before approving it. This is not the store's total contribution or net profit.

When media costs differ, compare expected contribution after media per additional media dollar, subject to stock and your required margin. Here A leaves $1 per extra dollar spent ($25 ÷ $25), B leaves $0.60 ($15 ÷ $25), and C loses $0.25 (−$5 ÷ $20). These rates are assumed constant only within this example; larger increases can change them.

A's larger contribution per order does not justify putting all $2,000 behind it: that would model 80 extra units against an allowance of 10. B is the next candidate because its assumed economics and available units support the remaining spend. If B cannot achieve those economics, leaving some budget unspent is a valid outcome.

Replace the two weakest assumptions first

Demand without the increase. If A's expected remaining demand rises from 90 to 105 units, its allowance becomes 120 − 105 − 20 = −5. Stop the proposed increase and review existing promotions. Do not keep spending just because the spreadsheet still shows a positive per-order contribution.

Orders caused by the increase. Historical platform-reported cost per purchase describes past spend and attributed purchases. It does not prove how many extra orders the next dollars will create. Google describes Conversion Lift as a controlled comparison of exposed and control groups to measure conversions caused by ads; the tool is not available to every account. Google Conversion Lift.

Use relevant experimental evidence when available, with its uncertainty. Otherwise label the assumption, limit the commitment and test a plausible range. For B, $35 per additional order would leave $5 before overhead; $45 would lose $5. A change in product mix or returns can move that threshold too. The break-even contribution worksheet helps establish the cost boundary.

Make stock the stop rule

The $250 assigned to A is a scenario allocation, not an inventory-safe spending limit. If an extra order costs $15 instead of $25, $250 models about 17 units, already above the 10-unit allowance. Faster organic sales can consume the same stock. Monitor actual available units and total order quantities across channels; do not wait for ad attribution to catch up.

1. Set an owner and a cutoff. Recalculate available units minus expected remaining demand and buffer as orders arrive. Stop the increase when that allowance is exhausted. Use inventory alerts and a review frequency suited to order velocity and reporting delay; once a day may be too slow for a scarce item.
2. Check the campaign controls. Confirm how the media buyer can limit or exclude each product. If one shared campaign cannot enforce the proposed allocation, treat the table as a merchandising constraint and revise the execution plan before spending.
3. Keep availability truthful. For Merchant Center, the product data, landing page, checkout and any availability markup must agree, including the advertised variant. If an item remains for sale but you want to stop promoting it, use the relevant pause or destination-exclusion control instead of falsely labeling it out of stock. Google's availability requirements.
4. Release more budget after a real change. Recount when replenishment is received and sellable, when a promotion ends, or when the economics change. Rebuild the remaining-window calculation rather than carrying yesterday's allowance forward.

For the next budget review, bring one row per product with available units, expected remaining demand, the additional buffer, contribution assumptions and a named stop-rule owner. That makes the decision specific: how much can this product absorb now, and what evidence would justify the next increase?