Building An AOV-Weighted Paid-CAC Ceiling Across SKU Tiers
When your catalogue spans a €35 entry SKU and a €165 bundle, one blended CAC ceiling is wrong for both. Here's how to build a tier-weighted ceiling and feed it into separate campaign budgets.
Quick answer
If your AOV spread is wider than about 2×, replace the single blended CAC ceiling with one ceiling per SKU tier, weighted by that tier's share of paid orders. Then split Meta and Google campaigns by tier so each ceiling actually binds a real budget — not an average that fits neither end of the catalogue.
AOV-weighted paid-CAC ceiling across SKU tiers
A paid-CAC ceiling calculated per SKU tier and weighted by that tier's share of paid orders, instead of one blended ceiling across the whole catalogue.
An AOV-weighted paid-CAC ceiling replaces the single catalogue-wide CAC target with a set of tier-specific ceilings. Each tier — entry SKU, mid, bundle, premium — gets its own contribution margin, its own target POAS, and therefore its own maximum allowable paid acquisition cost. The tier ceilings are then combined into a blended reporting number using order-share or revenue-share weights, but the ceilings that bind campaigns are the per-tier ones. This matters whenever a catalogue's AOV range is wide enough that the average AOV is not representative of any actual buyer.
The failure mode is easy to spot in the P&L. Blended CAC lands under target, ROAS looks fine, but contribution margin is flat or shrinking. Almost always, paid is over-acquiring the entry SKU at a loss and under-acquiring the premium bundle where the real margin sits.
A single ceiling assumes every paid order carries the average AOV. For a catalogue with a €35 serum and a €165 regimen, no order actually does. The ceiling is simultaneously too generous for the serum and too tight for the regimen.
Why one blended AOV ceiling breaks a mixed catalogue
Start with the arithmetic. If a skincare brand sells 70% entry-tier serums at €35 and 30% regimens at €165, the blended AOV is €74. At a 30% contribution margin and a 3× target POAS, that produces a paid-CAC ceiling near €22.
That €22 ceiling is €7 too high for the serum (whose real ceiling is closer to €12) and €30 too low for the regimen (which can defensibly absorb €55). Meta's algorithm cheerfully finds serum buyers all day at €18 — under the blended cap, over the real one. This is the pattern documented in diagnosing when a single blended AOV ceiling is hiding a loss-making entry SKU.
The tell in your dashboard
Blended CAC is at or under target, blended ROAS looks healthy, but contribution-margin-per-order is drifting down month over month. That's a mix shift toward the cheap SKU — the ceiling isn't catching it because the ceiling was built on last year's mix.
Grouping SKUs into tiers that behave alike
Two to four tiers is almost always enough. More than four and campaign structure collapses; fewer than two and you're back to blended. The cut points should follow buyer behaviour, not neat price bands — the details are in grouping SKUs into AOV tiers.
For most beauty and apparel catalogues, the natural split is entry (single-item trial purchase), core (repeat-buyer basket), and bundle/premium (considered purchase, higher intent). Each tier has a distinct contribution margin, return rate, and repeat-purchase profile, which is what makes a shared ceiling misleading.
Assign every SKU to exactly one tier. Bundles that include an entry SKU get their own attribution rule so they don't double-count — see handling the bundle SKU that cross-sells from entry tier.
What tiered ceilings look like in practice
Two-tier ceiling for a skincare catalogue (€35 serum, €165 regimen), 30% CM, 3× target POAS
| Tier | AOV | Order share | Contribution margin | Paid-CAC ceiling | Blended ceiling would be |
|---|---|---|---|---|---|
| Entry — serum | €35 | 70% | €10.50 | €10-12 | €22 |
| Bundle — regimen | €165 | 30% | €49.50 | €50-55 | €22 |
| Weighted blended (order-share) | €74 | 100% | €22.20 | €22 | — |
Read the last column. The blended ceiling reports €22 either way, but the tier ceilings are €10 and €55. Enforcing €22 across both campaigns is the exact behaviour that quietly kills margin — Meta spends up to €22 on serum orders that only support €10.
Order-share vs revenue-share weighting
The blended number you report to the board is a weighted average of the tier ceilings. Order-share weighting matches how the algorithm counts conversions; revenue-share weighting matches how finance counts euros. They give different blended numbers and drive different budget splits — the trade-off is worked through in weighting SKU-tier AOVs by order-share vs revenue-share.
Rule of thumb: use order-share weighting for setting campaign cost-per-result goals (that's what the platforms optimise against) and revenue-share weighting for the P&L reconciliation. Don't mix them in the same table.
Feeding tier ceilings into Meta and Google
A tiered ceiling only works if the campaigns are actually split by tier. One catch-all Advantage+ campaign with a €22 cost-per-purchase goal will drift toward the cheap SKU every time. Separate the product sets, set separate cost-per-result goals, and cap budgets independently — the mechanics are in splitting Meta and Google campaigns by SKU tier and feeding SKU-tier ceilings into Meta Advantage+.
Refresh the tier weights quarterly, not monthly. Weekly refreshes whipsaw the learning phase and cost more in re-learning than they save in precision. Refreshing tier weights quarterly without whipsawing campaign budgets covers the cadence.
Edge cases worth naming
If the entry SKU is a deliberate loss-leader — priced to acquire, monetised on the second order — the ceiling can sit below contribution margin, but only when it's paired with a retention guardrail on 90-day repeat rate. See when the entry SKU is a deliberate loss-leader for the guardrail structure.
If a bundle contains the entry SKU, decide once whether the bundle's CAC ceiling attributes to the bundle tier or splits pro-rata across tiers. Both are defensible; inconsistency between the two is not.
Frequently asked questions
Roughly 2× between the lowest and highest tier AOV. Below that, a blended ceiling is close enough. Above 2×, the blended number stops being representative of any real order and the mix-shift risk becomes material.
Two to four. Two tiers (entry vs bundle) covers most beauty and apparel catalogues. Four is the ceiling before campaign structure becomes unmanageable in Meta or Google. If you think you need five, you probably have two tiers with sub-segments — model those inside the tier, not as new tiers.
Order share for setting platform cost-per-result goals, revenue share for reconciling with the P&L. They answer different questions. The full trade-off is covered in the dedicated weighting page.
Pick one attribution rule and stick to it: either the bundle's CAC lands entirely in the bundle tier, or it splits pro-rata across the tiers of its component SKUs. Inconsistency between campaigns is what breaks the ceiling, not the choice itself.
Quarterly. Monthly refreshes cost more in Meta re-learning than they save in weight precision. If mix is drifting fast enough that quarterly feels too slow, that's usually a signal to add a tier, not to refresh more often.
Yes, but only if you split the product sets by tier and give each campaign its own cost-per-result goal. A single Advantage+ campaign with one blended goal will drift to the cheap SKU because that's where conversions are cheapest. Two campaigns with two goals hold the line.
Only if you attach a retention guardrail — typically 90-day repeat rate above a defined threshold. Without the guardrail, a sub-CM ceiling is just a slow way to lose money. The structured version is in the loss-leader page.
Target POAS is the input; the tiered ceiling is the output. You pick a POAS target per tier (they can differ — premium tiers often carry a higher POAS target because CM is richer), then derive the ceiling from AOV × CM% ÷ POAS.
Then the blended ceiling is dominated by that tier and looks fine — but the other 20% is where the mistake hides. Tier anyway. The point of tiering is to prevent budget leakage into the minority tier, not to rebalance the majority.
If margins vary more than AOVs across your catalogue, yes — margin-tiering is the underlying goal. AOV tiering is the practical proxy because AOV is easier to segment campaigns by. For catalogues where a low-AOV SKU has an unusually high margin, use CM% directly.
Track CAC, channels, and funnel conversion in one place
Metricuno connects ad spend, funnel events, and revenue so you can see CAC by channel, cohort, and campaign — without stitching together five tools.