Bidding for new customers instead of cheap conversions

A bid strategy optimizing to all conversions will find the cheapest ones, and for an established store the cheapest conversions are people who were already going to buy. That is a reasonable outcome if the store’s goal is efficiency this month. It is the wrong outcome if the goal is a larger customer base next year, and the two goals produce very different accounts.

New customer acquisition bidding tells the system to value a first-time buyer above a repeat buyer. The mechanics are simple. The judgment, which is how much more a first order is worth, is the part that decides whether it works.

Start with what a new customer is worth, not with the setting

Before changing anything, put a defensible number on a first order. Four inputs:

1. Contribution margin on the first order. Revenue minus product cost, payment fees, fulfillment, packaging and expected returns. Not gross revenue.
2. Repeat behavior. What share of first-time buyers order again, and how many times, within a window you can actually observe.
3. Margin on those later orders, which is often better because there is no acquisition cost attached.
4. The payback window the business can fund. A store financing inventory from cash flow cannot wait a year to recover acquisition cost, however good the lifetime value looks.

The output is two numbers: the total contribution a new customer is expected to produce within the payback window, and the contribution of a single order from an existing customer. The difference between them is the honest value of newness. Work it through with the CAC payback calculator, and read customer lifetime value and CAC payback period for the definitions that keep the arithmetic clean.

Use observed cohort behavior rather than a projected lifetime. A twelve-month observed figure that understates the truth is safer than a five-year projection that cannot be checked.

Two modes, two different jobs

Platforms that support this generally offer two shapes, and they suit different situations.

Mode	What it does	Fits when
Value bonus	Bids on all conversions but treats a new customer as worth more by a stated amount	The store wants growth without giving up repeat revenue, which is most cases
New customers only	Bids only toward first-time buyers	The store has a strong retention channel already capturing repeat demand, and enough budget to fund acquisition alone

The bonus mode is the default recommendation for most catalogs, because repeat purchases still have value and refusing to bid on them is an expensive way to make a point. The exclusive mode makes sense when a mature email and SMS program is reliably capturing repeat orders without paid support, so paid spend can concentrate on the part retention cannot do. The retention side of that arrangement is built in post-purchase flows.

The customer list is the load-bearing part

The system can only distinguish new from returning if it can recognize returning. That recognition comes from your own data, so the quality of the customer list decides the quality of the bidding.

• Upload the full purchaser list, not just recent buyers, using every identifier the store holds and is permitted to use.
• Keep it fresh with an automated sync rather than a manual upload that ages between campaigns.
• Check match rates after each sync, since a low match rate means most returning customers look new to the system.
• Respect consent and the platform’s terms for what may be uploaded and how it was collected.
• Pass a reliable first-party signal from the store where the platform supports it, so the order itself confirms whether the buyer was new.

Where the site’s own tagging is the source of truth, verify it before trusting it. A store that cannot answer how do you track new versus returning customer revenue from its own analytics is not ready to bid on the distinction.

Setting the bonus without inflating the account

The bonus is a value the bid strategy adds to a conversion when the buyer is new. Set it from the arithmetic above, not from optimism, and be conscious that it changes what your reported return means.

An illustrative calculation. Say average order value is 80 dollars, contribution margin is 40 percent, so a first order contributes 32 dollars. Say observed cohort data shows a first-time buyer produces, on average, one further order within twelve months at a similar margin. The additional expected contribution attributable to newness is roughly 32 dollars, and a bonus somewhere below that figure is defensible while leaving room for the estimate to be wrong. Publishing a bonus larger than the incremental contribution simply buys revenue at a loss with extra steps.

Two constraints:

• The bonus should never take the effective target below the break-even floor for the products involved. The break-even ROAS calculator gives that floor.
• Products with different margins support different bonuses. Where the catalog varies widely, segment first using the approach in margin-based Shopping segmentation, rather than applying one bonus to everything.

Reporting changes the day you switch

Two things happen at once, and both need explaining to whoever reads the report.

First, reported conversion value rises without a matching rise in bank revenue, because the bonus is a modeling value rather than money received. Keep a second view of unadjusted revenue so the store’s actual return stays visible.

Second, the efficiency numbers get worse before the business does better. Acquisition costs more per order than order capture, so a campaign that shifts toward new customers will show a lower return in the short term. That is the intended trade, and it needs to be agreed in advance rather than defended in month two.

Add three columns to the monthly report: new customer count, cost per new customer, and new customer share of orders. Track them next to the brand and non-brand split described in brand versus non-brand, because acquisition and brand capture are the two things this setting is meant to separate.

Guardrails

• Give it a full learning period. Changing the bonus weekly prevents the strategy from ever settling.
• Keep brand campaigns out of it. Brand traffic contains plenty of first-time buyers who found you elsewhere, and paying a bonus for them buys nothing.
• Watch the payback, not the month. The test of whether this worked is whether the cohorts acquired under the new setting reach payback within the window you set.
• Recheck the value input twice a year. Repeat rates move, margins move, and a bonus set from last year’s numbers can drift a long way from the truth.

When this is the wrong tool

Skip it, or postpone it, when any of these are true:

• The store is new and has too few conversions for the strategy to learn from, in which case the priority is volume and clean measurement.
• The category is genuinely one-purchase, where repeat behavior is rare and the first order is the whole relationship.
• Customer data cannot be matched reliably, so the new and returning labels are guesses.
• The retention program is not built. Paying more to acquire customers the store cannot bring back is the most expensive way to run this setting.

The platform mechanics are covered in how to set new customer acquisition goals in Google Ads, the target-setting context is in what is a good ROAS for Google Shopping, and the surrounding account work sits in Google Ads.

---

Sources

• Google Ads Help: About conversion tracking
• Google Ads Help: About Customer Match