Subscription Ecommerce Google Ads
The first order is not the customer.
For subscription and repeat-purchase ecommerce businesses, first-order ROAS can be deeply misleading. A customer who looks unprofitable today might become one of your best customers over the next twelve months. Another who looks excellent on first order may never buy again.
Yet Google Ads often judges both on the transaction it can see now. We think the better question is: what is this customer actually worth? That changes how you think about acquisition, bidding, product strategy, payback, and how aggressively you should scale.
Subscription economics change the advertising problem.
A one-off ecommerce business can often judge acquisition largely on the immediate transaction. Subscription businesses cannot. You need to understand:
And whether the product used to acquire the customer predicts what happens next.
A customer is a stream of economics, not a single conversion.
Why blended ROAS can look healthier than acquisition really is.
Imagine an account reporting 5.0 ROAS. Great. But that number contains existing subscribers, repeat purchasers, brand searches, customers already familiar with the business, and genuinely new customers. Those are not equally difficult or expensive to generate.
If existing customers account for a large share of revenue, blended ROAS can hide weak new customer acquisition. That is why we separate the two questions: how efficiently are we monetising existing demand, and what is it costing us to acquire somebody genuinely new?
CAC is often reported loosely. We care about nCAC: the cost of acquiring a genuinely new customer. Not total ad spend divided by all customers, because repeat buyers can make acquisition efficiency look considerably better than reality. If the objective is growth, we need to isolate the customers creating that growth.
But nCAC alone isn't enough.
Customer A
Costs £20 to acquire. Generates £30 first-order contribution. Never buys again.
Customer B
Costs £35 to acquire. Generates £10 first-order contribution. Then another £120 over the next year.
Which is the better acquisition? If we only optimise for immediate CPA: Customer A. If we understand customer economics: Customer B may be dramatically more valuable.
Cheap acquisition and valuable acquisition are not the same thing.
The product that acquires the customer matters.
This is one of the most useful questions in subscription ecommerce: which products create your best customers? Maybe a starter bundle. A trial. A specific flavour. A particular skincare product. A larger pack. A lower-margin introductory offer.
Some products can be commercially valuable not because of what happens on the first order, but because of the customers they introduce. We call these Gateway products.
Gateway SKUs
A Gateway SKU's job is not necessarily to maximise first-order profit. Its job is to acquire customers whose wider economics justify the investment.
That means we look at first-order contribution, nCAC, subscription conversion, repeat purchase, payback, lifetime contribution, cross-sell, and retention. Then decide what the business can rationally afford to pay.
This changes how you judge Google Ads performance.
Imagine two campaigns.
Campaign A
ROAS: 5.0
nCAC: £18
12-month customer contribution: £40
Campaign B
ROAS: 3.2
nCAC: £30
12-month customer contribution: £140
Which campaign deserves more investment? The first looks better in Google Ads. The second may create far more value.
First-order efficiency can actively point you towards the wrong customer.
Payback matters.
Lifetime value is useful. But "we'll make the money back eventually" is not an unlimited acquisition strategy. Cash matters. So we also look at how long it takes to recover CAC. 30 days? 90 days? Six months? A year? Different businesses can tolerate different payback periods.
LTV without payback is incomplete.
A customer might be profitable eventually while still creating an unacceptable cash requirement today.
Contribution matters too.
Revenue-based LTV can be dangerously flattering. A customer who spends £400 over twelve months is useful, but what matters commercially is what remains after product cost, discounting, fulfilment, payment fees, shipping, and returns where relevant. Customer value should eventually reconcile to contribution. Not simply gross revenue.
Subscription conversion rate
Not every first-time buyer becomes a subscriber. And different acquisition products can produce very different outcomes. We want to understand which products, campaigns, queries, markets, and customer segments generate subscriptions.
Because a low-margin acquisition product can become extremely attractive if it disproportionately creates high-value subscribers.
Churn changes the equation.
A subscription business with 3% monthly churn has very different acquisition economics from one with 12%. If churn rises, customer value falls, breakeven CAC changes, and payback extends. Google Ads targets should eventually respond.
Acquisition strategy cannot be permanently disconnected from retention performance.
Google Ads and retention are different jobs.
We specialise in Google Ads. We do not manage your email or subscription retention programme. But retention data still matters to our advertising decisions, because the quality of the customers Google Ads acquires determines what those customers are worth. We don't need to own retention to understand its economic consequence. That's the distinction.
Product-level LTV can be more useful than account-level LTV.
One account-wide LTV number can hide significant differences. Customers acquired through Product A may behave very differently from customers acquired through Product B. Maybe Product A attracts bargain hunters, Product B attracts subscribers, Product C attracts high-AOV repeat buyers, and Product D creates customers who churn immediately.
Acquisition product can predict customer quality. That's useful information for Google Ads.
Search intent can predict customer quality too.
Not every query should be valued equally. Someone searching "cheap dog food" may behave differently from someone searching "cold pressed dog food subscription". The second search may be more expensive. It may also produce a materially better customer.
We look beyond cost per conversion and ask: what happened to the customer afterwards?
Brand vs generic matters.
Brand traffic often converts exceptionally well. Existing customers may also use branded searches to reorder. That can make Google Ads performance look excellent. But if the commercial objective is acquisition, we need to understand how much of that result represents genuinely new demand.
Retention revenue should not be allowed to disguise acquisition economics.
Performance Max can blur this further.
PMax can combine brand, remarketing, existing customers, new customers, Shopping, and broader discovery into one campaign result. Useful for execution. Less useful for understanding customer economics. So we look at customer composition, brand contribution, product mix, new customer signals, Gateway products, and first-order contribution.
PMax should have a defined customer job.
Explore Performance MaxNew customer bidding
Google provides tools intended to help advertisers prioritise new customers. Useful. But the business still needs to answer: what is a new customer worth? Are all new customers worth the same? Which acquisition products create better cohorts? How reliable is the new-customer identification? How much premium can we rationally pay?
The setting is the easy part. The economics determine what number belongs in it.
Subscription products can have different jobs.
Not every product needs to be a Gateway. BOI® gives each product the objective that matches its actual commercial role.
Scale
Strong first-order economics and strong repeat behaviour.
Profit
Mature products generating excellent contribution.
Gateway
Lower immediate margin but exceptional customer value.
Protect
Strategically important demand.
Recovery
Inventory still exists even in subscription businesses.
Trials and starter products
Trial offers can be extremely powerful. They can also be an excellent way to spend money acquiring customers who never become profitable. The important questions: what does the trial cost, what does acquisition cost, what percentage convert to full customers, how quickly, what is their retention, what is their eventual contribution?
Trial volume is not the objective. Customer value is.
Discounts
A heavily discounted first order may improve conversion, lower CPA, increase volume, and absolutely destroy first-order economics. That can still be rational if conversion to subscription is strong, retention is strong, and payback is acceptable.
Discounts should be judged against cohort economics, not simply conversion uplift.
Cohort analysis matters.
The customers acquired in January may behave differently from those acquired in June. Promotions, products, channels, pricing, seasonality, creative, and market changes all play a part. We want to understand cohorts by acquisition period, product, campaign, customer type, and market, where data volume allows.
LTV should be observed, not worshipped as one permanent number.
Historical LTV can mislead. If prices changed, churn changed, product mix changed, acquisition expanded into new markets, or retention deteriorated, last year's lifetime value may not describe tomorrow's customer. Customer economics need updating like every other commercial input.
Scaling subscription acquisition
Suppose an account currently spends £50,000 per month, with nCAC of £25 and 12-month contribution per acquired customer of £100. Should you spend £100,000? Maybe. But the next £50,000 may acquire customers at £30, £40, or £60, and their quality may differ too.
Scaling depends on marginal customer economics. Not the average CAC of customers already acquired.
Marginal CAC
As spend increases, the next customer often costs more than the previous one. That is normal. The important question is not "did CAC increase?" It is: did CAC increase beyond what the customer is worth?
A rising CAC can still be commercially attractive. A flat CAC can still be poor if customer value falls.
Acquisition ceilings should come from the business.
Instead of "target CPA = £30 because that's what the campaign historically achieved", we want first-order contribution, expected downstream contribution, payback requirement, desired profit, and risk tolerance. Then derive: what can we rationally afford to pay?
That's a commercial CAC target.
Google Ads is not your LTV model.
We do not believe every customer-lifetime calculation should be stuffed directly into Google's conversion value. There are challenges: latency, data quality, changing cohorts, returns, churn, attribution, and uncertainty. We use customer value to inform advertising decisions. How directly it should influence bidding depends on the reliability and maturity of the data.
More data does not automatically mean more data should be fed into Smart Bidding.
POAS still matters.
Subscription economics do not make first-order profitability irrelevant. If a customer has strong lifetime value but first-order acquisition consumes enormous cash, the business may still have a scaling constraint. POAS helps us understand immediate economic return. LTV adds the longer-term view. We need both time horizons.
Explore POASA simple example
Imagine £100,000 monthly ad spend generating £400,000 of Google Ads revenue. ROAS: 4.0. Looks good.
£180,000
of revenue came from existing customers
£20
true nCAC across 5,000 new customers
£12
average first-order contribution
-£8
first-order result after acquisition
Terrible? Maybe. Then cohort data shows average 12-month downstream contribution of £85. Now the economics look completely different. But if payback takes 14 months and the business only has enough cash to support six, the strategy still may not scale.
No single metric answers the whole problem. That's the point.
How we manage subscription Google Ads
1.
Separate acquisition from repeat demand
Understand what is actually generating growth.
2.
Establish nCAC
The cost of acquiring a genuinely new customer. Not blended CPA.
3.
Understand first-order contribution
What does acquisition cost immediately?
4.
Connect customer value
What happens after the transaction?
5.
Identify Gateway products
Which products create disproportionately valuable customers?
6.
Establish payback
Can the business fund the acquisition curve?
7.
Build Google Ads around those economics
Search. Shopping. PMax. Product jobs. Bidding. Budgets.
8.
Monitor cohorts
Does the customer still behave like we expected?
9.
Scale to marginal economics
Not historical averages.
What we don't do
Subscription acquisition needs both customer economics and advertising discipline.
Results
Wilsons Pet Food
A subscription-led ecommerce business where first-order transaction value did not tell the whole customer story. By managing acquisition around customer intent and longer-term economics:
+212%
Revenue
+112%
Contribution
+83%
Customer LTV
Who this approach is for
It becomes particularly valuable when:
Frequently asked questions
What is nCAC and why does it matter?
nCAC is the cost of acquiring a genuinely new customer. Total ad spend divided by all customers is misleading because repeat buyers make acquisition efficiency look considerably better than reality. If the objective is growth, we need to isolate the customers creating that growth.
Should we feed lifetime value directly into Smart Bidding?
Not automatically. There are real challenges: latency, data quality, changing cohorts, returns, churn, attribution, and uncertainty. We use customer value to inform advertising decisions. How directly it should influence bidding depends on the reliability and maturity of the data. More data does not automatically mean more data should be fed into Smart Bidding.
Do you manage retention and email as well?
No. We specialise in Google Ads. We do not manage your email or subscription retention programme. But retention data still matters to our advertising decisions, because the quality of the customers Google Ads acquires determines what those customers are worth. We don't need to own retention to understand its economic consequence.
Is first-order ROAS irrelevant for subscription businesses?
No. Subscription economics do not make first-order profitability irrelevant. If first-order acquisition consumes enormous cash, the business may still have a scaling constraint even with strong lifetime value. POAS helps us understand immediate economic return. LTV adds the longer-term view. We need both time horizons.
How do you handle introductory discounts and trials?
By judging them against cohort economics, not conversion uplift. A heavily discounted first order can improve conversion, lower CPA, increase volume, and absolutely destroy first-order economics. That can still be rational if subscription conversion, retention, and payback are strong. Trial volume is not the objective. Customer value is.
The subscription question.
Not "what ROAS should we target?" And not even "what is our LTV?" The better question is: what can we rationally afford to pay for the next customer? Based on what they contribute today, what they're likely to contribute tomorrow, how quickly the cash comes back, and how confident we are in that expectation.
The first order is not the customer. Cheap acquisition is not necessarily good acquisition. And LTV is useless without payback.