Loyalty Adjustment Bidding - How Repeat Buyers Inflate Your ROAS and Mask True Acquisition Cost

The Repeat Buyer Problem

Here's a scenario every ecommerce brand has lived through. Your Google Ads dashboard shows a 6x ROAS on Shopping campaigns. You celebrate. You increase budget. Three months later, revenue is up but profit is flat. The finance team is confused. The agency says "performance is great." The P&L says otherwise.

The explanation is almost always the same: a significant portion of your "conversions" are repeat buyers. People who have purchased before, who know your brand, who were going to come back regardless - and who happened to click a Google Ad on their way in.

Google Ads attributes that conversion to the ad. Your ROAS inflates. Smart Bidding sees a "successful" auction and bids more aggressively for similar users - which, naturally, tend to be more existing customers. A self-reinforcing loop that looks like growth but is actually just an increasingly expensive welcome mat for people already walking through the door.

The question isn't whether this is happening. It's happening in every account. The question is: how much of your ROAS is real acquisition, and how much is repeat revenue wearing an acquisition costume?

How Repeat Buyers Inflate ROAS

Repeat customers convert at 3-5x the rate of new visitors. They have higher average order values. They browse less and buy faster. From Google's perspective, they are perfect prospects - high conversion probability, high revenue, low cost per conversion.

Smart Bidding optimises for auction-level conversions. It doesn't know - or care - whether a conversion is a new customer or someone's fourth purchase this year. It sees a click that converted, credits the campaign, and says "let's find more users like this." More users like this means more existing customers. Higher ROAS. Lower incremental value.

The mechanics of the inflation are straightforward:

• • Branded search: Existing customers search your brand name, click a Shopping ad, and buy. Google credits Shopping. Your brand campaign absorbs spend that would have converted organically.
• • Remarketing bleed: Performance Max serves ads to existing customers via remarketing audiences. The conversion was already in progress - PMax just touched it last.
• • Customer match overlap: Even without explicit targeting, Google's signals identify patterns common to your existing customers. The algorithm gravitates toward them because they convert reliably.
• • Email-driven behaviour: A customer receives your Klaviyo email, searches for your product on Google, clicks an ad, and buys. Google credits the ad. Klaviyo credits the email. The customer was coming back regardless.

None of this is fraud. None of it is a bug. It's the logical outcome of an algorithm optimising for what it can see (conversions) rather than what matters (genuinely new revenue).

Worked Example: The Hidden Inflation

A mid-market fashion brand spends £30,000/month on Google Ads. Dashboard ROAS is 5.2x. Looks excellent. Let's decompose it:

The blended 5.2x ROAS masked a new-customer ROAS of 1.4x. After COGS, fulfilment, and payment fees, the contribution margin on those new customers was barely positive. The "profitable" account was actually an expensive retention channel disguised as an acquisition engine.

This brand wasn't growing. It was paying Google to re-sell to its own email list.

Smart Bidding's Blind Spot

Smart Bidding strategies (tROAS, tCPA, Maximise Conversion Value) optimise for total conversion value. They make no distinction between a first purchase and a fifth. From the algorithm's perspective, a £100 conversion is a £100 conversion - whether it's from a net-new customer or someone who would have searched for you directly.

This creates three structural problems:

• • Budget migration: Spend migrates toward audiences with the highest conversion rates - which are existing customers. Acquisition budgets quietly become retention budgets without anyone noticing.
• • Inflated targets: ROAS targets set on blended data include repeat revenue. When you try to hit 5x ROAS from new customers alone, the maths doesn't work. The target was never achievable with genuinely new traffic.
• • False confidence: Month-over-month ROAS stability masks declining acquisition. You're hitting targets because repeat rates are rising - not because acquisition is working. The moment repeat rates plateau, "performance" collapses.

Google has introduced "new customer" bidding adjustments in some campaign types - notably Performance Max and Shopping. But these require clean customer list uploads, proper conversion tracking, and deliberate campaign architecture. Most accounts don't have any of this configured. Smart Bidding makes agencies lazy precisely because it delivers ROAS that looks good enough to avoid these harder questions.

Segmenting New vs Repeat

The fix starts with measurement. You can't adjust bidding for something you can't see. Here's how to create visibility:

• • Customer match lists: Upload your full customer database as a Customer Match audience. Update it weekly. This gives Google a baseline for who is "existing" vs "new."
• • Conversion segmentation: Use conversion action sets to separate "new customer purchase" from "repeat customer purchase." Pass a parameter indicating first-order vs repeat from your checkout.
• • GA4 audiences: Build audiences based on transaction count. "Users with 0 prior purchases" gives you a clean new-customer segment to compare against "Users with 1+ prior purchases."
• • Reporting dimensions: In your weekly report, split every campaign's conversions into new vs repeat. If more than 50% of a Shopping campaign's conversions are repeat buyers, the campaign isn't doing acquisition work - it's doing retention work at acquisition cost.

The moment you can see the split, the conversation changes. Instead of "our ROAS is 5x," it becomes "our new customer ROAS is 1.4x and our repeat buyer ROAS is 12x." That second statement is commercially useful. The first is dangerously misleading.

How to Adjust Your Bidding

Once you've segmented, there are three practical approaches:

Customer List Hygiene

None of this works without clean customer data. The most common failure mode isn't strategy - it's data hygiene:

• • Stale lists: Uploading your customer list once and forgetting about it. New customers acquired last month aren't in the list yet, so they're still counted as "new" in next month's conversions. Update weekly - ideally via automated feed.
• • Email mismatch: Customers use different emails for purchase vs Google account. Customer Match relies on email matching, so mismatches create gaps. Match rates typically range from 30-70% depending on your sector.
• • Phone + address enrichment: Upload phone numbers and postal addresses alongside emails. Google matches on multiple identifiers, improving coverage from 40% to 60%+ in many cases.
• • Frequency thresholds: A customer who bought once 18 months ago is different from someone who buys monthly. Consider separate lists: "active customers" (purchased in last 90 days) vs "lapsed customers" (90-365 days) vs "new prospects" (never purchased).

The brands that get this right typically automate the process: a nightly export from Shopify or Klaviyo to Google Ads via API. Manual CSV uploads become stale within days, and staleness means misclassification.

Measuring True Acquisition Cost

Once you've separated new from repeat, the next question is: "What should I pay to acquire a customer?" This requires three numbers:

• • First-order contribution margin: Revenue minus COGS, fulfilment, payment fees, and allocated ad spend. If this is negative, you need LTV to justify the investment. If you don't have LTV data, you're flying blind.
• • LTV by acquisition channel: Not all customers are equal. Customers acquired via Google Shopping may have different repeat rates and LTV than those acquired via PMax or organic. Blend carefully - or better yet, don't blend at all.
• • Payback period: How many months until a customer's cumulative contribution margin covers the acquisition cost? If the payback is 9 months and your cash cycle is 60 days, you have a cash flow problem regardless of profitability.

The nCAC ceiling should be set by your CFO, not your agency. It's a commercial decision: "How much can we afford to invest in acquiring a customer given our margins, repeat rates, and cash position?" Setting it based on platform ROAS puts the cart before the horse.

Most brands discover that their true nCAC is 2-3x higher than their blended CAC suggested. This isn't bad news - it's clarity. And clarity is what lets you set achievable targets instead of chasing blended metrics that were never real.

Next Steps

If you can't tell us what percentage of your Google Ads conversions are genuinely new customers, your ROAS is a blended fiction. Start by measuring the split. Then restructure your bidding around the truth. The number might be uncomfortable - but it's the only number worth optimising.