A quick primer: why a single transaction is a poor anchor for a repeat-purchase business
If Google Ads optimizes only toward the value of a customer's first purchase, the strategy systematically undervalues customers who cost more to acquire up front but come back and buy again — and overvalues customers who are only ever worth a single sale. In categories with high repeat-purchase frequency (subscriptions, consumables, products with a regular reorder cycle), the gap between single-transaction value and LTV can be a multiple, and optimizing blindly on the first receipt literally sends budget to the wrong place relative to the business's real payoff.
Google doesn't have one button labeled "optimize for LTV" — there are several separate, combinable mechanisms, each covering part of the job:
Specific Google Ads mechanisms for working with LTV
- Conversion Value Rules by audience. Value Rules support adjusting conversion value based on audience membership (alongside location and device) — meaning you can apply a value multiplier for users who belong to a Customer Match list of known high-LTV customers (say, your top 20% by historical LTV), or a GA4 predictive audience with a high predicted value. Technically, this is the closest thing to "partial LTV bidding" among Google's built-in tools.
- GA4 Predictive Audiences. GA4 offers predictive metrics — purchase probability, churn probability, and, particularly relevant here, an audience of "likely most profitable buyers within 28 days." You can push this audience into Google Ads (via the GA4–Google Ads link) and use it as an audience signal in Performance Max/Demand Gen, or as targeting/an observation segment in Display and Search — a way to steer targeting toward a user's likely future value without having a precise LTV prediction model of your own.
- Passing a predicted LTV figure directly as conversion value. If the business has its own cohort-based LTV model (built from CRM data), you can pass Google Ads not the first-order amount, but a calculated expected LTV (or a conservative share of it) as the conversion value itself — the same logic as passing margin instead of revenue. This is the most direct way to actually bid toward LTV rather than a proxy for it, but it requires the infrastructure on the business's side to calculate that figure at the transaction level.
- New Customer Acquisition (value-based). A cruder but standard, easy-to-set-up approximation — assign new customers a higher value relative to returning ones within one optimized goal, reflecting part of their future LTV without building a full predictive model.
- Customer Match for remarketing and retention. Lists of existing customers (especially segmented by LTV or purchase frequency) can be used not only for value adjustment (point 1), but as the basis for separate remarketing campaigns aimed at repeat purchases, upsell, and cross-sell — structurally, this is a different job (retention) than acquiring new traffic, and it naturally has a different, higher achievable efficiency that shouldn't be blended with new-customer acquisition in a single campaign.
What to check before you touch anything
- Are customers currently segmented in ways where LTV differs meaningfully (new/returning, different product categories with different repeat-purchase frequency)? Without that split, it's technically impossible to apply different LTV logic to different segments.
- Does your CRM have enough historical data to calculate a reliable LTV by cohort? On newer businesses, or in categories with a long repeat-purchase cycle, there may simply not be enough data yet to produce anything better than a guess.
- Is the GA4–Google Ads connection set up so that predictive audiences are even available to use? GA4's predictive metrics require a certain volume of conversion data to become "eligible," and they're not automatically available for every account.
- Are current, up-to-date lists of high-value existing customers loaded into Customer Match? Without regularly refreshing these lists, both value adjustment and retention remarketing will be working off stale data.
- Do competitors factor LTV into their acquisition strategy (visible indirectly through how aggressively they bid to acquire new customers)? In categories where competitors are willing to take a loss on the first transaction for the sake of LTV, a strategy based only on single-transaction value will systematically lose out on quality new traffic.
Possible approaches
- If the business has a reliable cohort-based LTV model, the most accurate approach is to pass the calculated LTV (or a conservative share of it) as the conversion value itself, restructuring how conversion value is calculated, rather than simply raising or lowering tROAS on top of an unchanged single-transaction value.
- If a full predictive model isn't available yet, but you do have a list of known high-value customers, you can use audience-based Conversion Value Rules (Customer Match / a GA4 predictive audience) as a partial approximation, boosting conversion value for users on those lists.
- An even simpler first step is New Customer Acquisition value-based bidding, which doesn't require building an LTV model at all — it just reflects the basic difference between a new and a returning customer.
- For retention and repeat sales, it makes sense to run separate campaigns built on Customer Match (remarketing, upsell, cross-sell to your existing base) with their own target — usually either looser or more ambitious on ROAS — rather than blending retention logic with new-customer acquisition logic in one campaign and one goal.
- If you don't yet have enough data for a reliable LTV calculation, it's smarter not to build an unproven estimate into bidding yet — instead, run cohort-based LTV analysis separately, alongside your current single-transaction-based strategy, and move to more advanced mechanisms (passing calculated LTV, audience-based value rules) only once the forecast is statistically sound.