From $2M to $3M a month, the lever is what each order is worth.

What happens to the price of a customer when you double what you are willing to pay for attention?

Somewhere between $2M and $3M a month in revenue, a founder runs into a quiet fact. The account is well run. The creative pipeline works. New customers are counted properly. And the next thousand new customers still cost more than the last thousand did.

That is not a failure of the media team. It is the shape of a response curve. Every bidding system is built to find the cheapest available buyers first, so it does. Each additional dollar reaches someone a little less ready and a little more expensive. At low spend you are buying the easy customers. At high spend you are buying the ones your competitors want too.

So the question changes. It stops being “how do we get customers cheaper” and becomes “how do we make each customer worth more, so we can afford the ones that cost more.” That is the fourth rung of the ecommerce growth ladder, after creative supply, a trusted second channel, and a working new-customer engine.

The short version: close the gap from the order side

Here is the decision rule, stated once and then worked through. Take the contribution one first order produces, after product cost, shipping, fulfillment, payment fees, and returns. Divide it by the marginal cost of a new customer: what the last increment of spend paid for each extra first-time buyer, not the monthly average. Call the result coverage.

• →Coverage at or above 100%: the first order pays for itself at the margin. Margin is not your constraint yet. Spend more, and go back to the engine.
• →Coverage from 85% to 100%: order value alone can close the gap. Start with AOV, because it moves contribution faster than it moves the price of the order.
• →Coverage from 60% to 85%: you need order value and gross margin together, and usually a better offer than a discount.
• →Coverage below 60%: the first order will not pay back at the margin, whatever you do to it. Either the repeat curve pays and you plan for that explicitly, with a stated payback window, or the last spend increase was too large.

Those thresholds are mine, not an industry benchmark. They come from the arithmetic below, which shows how far each lever moves contribution in the illustrative cost structure before it runs out of room. Replace them with your own once you have run the numbers on your catalog.

Why the next customer costs more, and why the average hides it

Two numbers sound alike here and are not. Average new-customer CAC is total spend divided by total new customers. Marginal new-customer CAC is the extra spend of the last increment divided by the extra new customers it produced. A brand can report a comfortable average while paying far more than an order is worth for the last few hundred customers of the month.

Bidding makes the gap harder to see. Google’s Target ROAS strategy predicts the value of each potential conversion and bids toward the return you set across the campaign. An average target lets cheap conversions subsidize expensive ones. The campaign hits its number while the last dollar earns much less than the first.

None of this means optimization stops mattering. It means a well-run account has already captured most of what optimization can give. Wasted search terms, dead placements, and fatigued creative get found once and then maintained. At $2M a month they are hygiene, not a growth plan.

The worked table: allowable CAC by order value and margin

Allowable CAC is the most you can pay for a new customer and still break even on the first order. It equals first-order contribution before any marketing cost. The table is arithmetic for an illustrative brand, not a client result. It assumes $14 of shipping, pick and pack, and payment cost on every order, held flat for simplicity, and order values net of returns.

Read down a column first. At 60% gross margin, moving AOV from $80 to $100 is a 25% increase in order value. Allowable CAC moves from $34 to $46, a 35% increase. The extra $20 in the basket carries $12 of margin, and none of the $14 per-order cost comes with it, because you ship one box either way.

Now read across a row. At $80 AOV, going from 60% to 70% gross margin lifts allowable CAC from $34 to $42, about 24%. Ten points of margin is a large change. It usually takes a price increase, a supplier renegotiation, or a shift in mix toward higher-margin products. It is worth a lot. It is also slower to get than a better bundle.

That asymmetry is the argument of this rung. Because some costs arrive once per order, every dollar added to the basket is worth more to contribution than the average dollar already in it. The logic also runs in reverse. At $60 and 50% margin, nearly half the gross margin on each order goes to getting the box out the door.

Three levers, one gap: which moves contribution most

Give one illustrative brand a gap to close. AOV is $90, gross margin is 60%, and per-order cost is $14, so first-order contribution is $40. The last increment of spend bought new customers at $48 each. Coverage is 83%, which puts this brand in the “order value and margin together” band.

On this table alone, a 10% cut in marginal CAC and a 10% rise in AOV look similar. They are not, for two reasons the table cannot show.

First, the CAC cut is the hardest number in the account to hold while spend is rising. You are asking the media team to buy more of a scarcer thing for less. The AOV rise does not decay as spend grows. It applies to the ten-thousandth order exactly as it applies to the first.

Second, the CAC cut only touches new customers. AOV and margin gains apply to every order in the business: repeat orders from customers you acquired years ago, and orders that arrive with no ad involved at all. For an established brand that can be the larger share of revenue. Order-side work compounds across the whole order book. Acquisition-side work touches one slice of it.

This is why I tell founders at this size to stop asking the media team for a lower CAC and start asking the merchandising team for a bigger basket. The media team can help. The basket is where the room is.

Order value: bundles, thresholds, and pack sizes

AOV is the first lever in the 85% to 100% band, and the tools that move it are familiar. What is less familiar is that each one can lose money while raising AOV. The test is always contribution per order, never AOV by itself.

Bundles: the one-in-three rule

A bundle sells related products together, usually at a discount. On Shopify it runs through a bundles app, and Shopify’s documentation notes that only fixed bundles are supported on the Google & YouTube channel, which matters if Shopping carries a large share of your acquisition.

The arithmetic to check before launch, illustrative. Two items at $50 each, 60% margin, $14 per-order cost. A single-item order contributes $16. A full-price two-item order contributes $46. A two-item bundle at 10% off sells for $90 and contributes $36. So every bundle sold to someone who would have bought one item earns $20 more, and every bundle sold to someone who would have bought both anyway earns $10 less. The bundle pays if at least one in three bundle buyers would otherwise have bought a single item. That is a question for a holdout, not for the AOV report.

Free-shipping thresholds: count who was already above it

A threshold asks the customer to add product to avoid a shipping charge. The trap is that everyone already above the threshold gets free shipping for nothing. Shopify counts product prices only, at their discounted price, toward a free shipping minimum, so the threshold also interacts with every other discount you run.

Illustrative arithmetic. You charge $8 for shipping today and 30% of orders already exceed $100. Set free shipping at $100 and those orders stop paying $8, which costs $2.40 averaged across every order. If 20% of orders move up by adding $40 of product at 60% margin, each adds $24 of margin and gives up $8 of shipping, a net $16, or $3.20 averaged across all orders. That clears the $2.40. If only 15% move up, it merely breaks even. This ignores any lift in conversion rate, which is the other thing a threshold can buy and which you should measure rather than assume.

Pack sizes: bigger boxes, same shipping

Larger packs, multi-packs, and refill sizes raise AOV without cutting the unit price as deeply, and they spread the per-order cost over more product. For consumables they also change the repeat cadence, which is the bridge to the subscribe-and-save arithmetic. The check is the same: contribution per order, and whether the bigger pack simply replaces repeat orders you would have received anyway.

Offer design: in this example, a 20% discount needs 77% more orders

At this size the offer matters as much as the ad. The ad decides who sees the product. The offer decides what the order is worth once they buy.

Arithmetic for an illustrative $100 product at 60% margin with $14 per-order cost. At full price it contributes $46. At 20% off it sells for $80, product cost stays at $40, and contribution falls to $26. You need 77% more orders at the discounted price to earn the same total. A gift with purchase that costs you $6 to supply and retails at $20 leaves contribution at $40. The customer sees a generous offer. You keep most of the order.

This is the logic behind Q4 offers that protect margin, and it holds all year. A percent-off discount shrinks the contribution of every order it touches, including orders from customers who would have paid full price. Gifts, bundles, bonus sizes, and guarantees change what the customer gets without cutting as deeply into what you keep.

Gross margin and returns: the slow levers that set the ceiling

Gross margin moves slowly, through pricing, supplier terms, and product mix. It is also the lever with the most permanent effect, because it raises allowable CAC for every channel, every campaign, and every order at once.

Mix is the one founders underrate. If two products sell at the same price and one carries 70% margin while the other carries 50%, moving acquisition spend toward the first raises blended margin without changing a single price. The product-level view in the losers campaign ROAS hides is where that shows up.

Returns are the lever nobody on the media side owns. A returned order costs outbound shipping, return shipping, handling, and the margin on anything that cannot be resold, on top of the lost sale. Illustrative arithmetic: if 12% of orders come back and each return costs $20 all in, that takes $2.40 off the average order. On the $40 base above, that is 6% of contribution, gone before the ad team spends a dollar. Size guides, clearer product pages, and honest imagery are margin work, even if nobody files them that way.

Bid to profit once the order side is fixed

Once contribution per order is known and moving, tell the bidding systems about it. Many Shopping and Performance Max accounts bid on revenue, because revenue is the conversion value the store sends. Revenue treats a $100 order at 70% margin and a $100 order at 40% margin as the same order. Your P&L does not.

Google Ads can report gross profit when you send conversions with cart data and supply a cost_of_goods_sold value for each product in your Merchant Center feed. That gives you profit by campaign and by product. Changing what the bidder optimizes toward is a separate step, covered in profit on ad spend. The order matters: fix the order side, then teach the bidder what an order is worth, then hold it to your breakeven ROAS.

What “lower your CAC” gets wrong at $2M a month

The most common advice to a brand at this size is to lower CAC by optimizing harder: tighter structure, better bidding, more testing. I think that advice is wrong at this rung. Not because optimization is useless, but because it aims the effort at the lever with the least room left.

Three reasons. Scarcity: a well-run account has already harvested its cheap improvements, and each new one costs more effort for less return. Decay: CAC improvements erode as spend rises, because the curve keeps bending. Scope: CAC work touches only new customers, while order-side work touches every order.

There is a quieter cost too. When the brief to the media team is “lower CAC,” the easiest way to comply is to buy more returning customers and brand searches, which are cheap and flattering. CAC falls. The number of new customers falls with it. That is how a brand optimizes itself into a plateau while every dashboard improves, and why new-customer CAC, not blended CAC, has to be the number on the brief.

When this advice is wrong

The order-side argument has limits. It is the wrong advice in at least four situations.

• →The account is not actually well run. If search terms, placements, and tired creative are still leaking money, fix those first. They are cheaper than a new bundle.
• →Coverage is already above 100%. If the first order pays for itself at the margin, your constraint is volume. Spend more and revisit this when the curve bends.
• →The product caps the basket. A brand selling one durable item per household cannot bundle its way to a bigger order. Margin, returns, and referral are the levers left.
• →The business runs on repeat, not the first order. For consumables and subscriptions the right comparison is payback over a defined window, not first-order coverage. The first order can lose money by design, as long as the window is explicit and the cash to fund it exists.

In every case the principle survives: compare what an order is worth with what the marginal customer costs. Only the lever changes.

What to do this month

1. Compute first-order contribution for your actual product mix: AOV net of returns, times gross margin, minus shipping, fulfillment, payment fees, and the cost of returns. The MER target calculator turns contribution margin and fixed costs into the blended target you need.
2. Estimate marginal new-customer CAC from your last two spend increases: the extra spend divided by the extra first-time customers, matched against your own customer list rather than a pixel.
3. Divide the first number by the second. That is coverage, and it tells you which band you are in.
4. Pick one order-side test for your band: a bundle with a holdout, a threshold change, a new pack size, or a gift with purchase in place of a discount. Judge it on contribution per order, not AOV. November is the wrong month to read a clean test, so use the peak to try a gift with purchase in place of a discount, as in Q4 offers that protect margin, and schedule bundle and threshold holdouts for after Cyber Monday.
5. Rewrite the brief to the media team around new-customer CAC and contribution, so the account is never rewarded for buying returning customers.

The full cost structure behind these numbers is in ecommerce unit economics. If you want a second set of eyes on your own, the free audit is a place to start on the ad account, and a call is where we work through what an order is worth and what the next customer costs.