A customer views a product twice, adds it to their basket, then sees a dynamic ad on Instagram and purchases that evening. Meta reports a conversion. Google may report one too if they also clicked a branded search ad. Your email platform may claim the sale as well. The question is not whether the conversion happened. It is: does retargeting cannibalise sales that would have happened without the ad?

For established eCommerce and lead generation businesses, this is one of the most commercially important questions in paid media. Retargeting often produces the strongest reported return on ad spend because it targets people already close to converting. But a strong reported ROAS is not the same as incremental growth. If retargeting is simply claiming credit for existing demand, increasing spend can make dashboards look better while profit, new customer growth and total revenue stand still.

That does not mean retargeting is a problem. It means it needs to be managed as a demand-conversion tool, not treated as proof that your acquisition strategy is working.

Does retargeting cannibalise sales or reveal demand?

Retargeting cannibalises sales when it receives attribution for conversions that would have occurred anyway through direct traffic, email, organic search, branded search or another paid channel. In practical terms, you pay to reach a customer who had already decided to buy.

It can also cannibalise the wider media mix. A campaign that relentlessly follows recent site visitors may pull budget away from prospecting, creative testing or Google Shopping activity that brings genuinely new customers into the funnel. This is particularly common when account optimisation is driven solely by platform-reported ROAS. Ad platforms naturally favour audiences that are easiest to convert, and recent visitors are usually the easiest audience available.

However, retargeting can be genuinely incremental. A shopper may need a reminder after getting distracted, reassurance through reviews or social proof, a time-sensitive offer, or a clearer reason to choose your brand over a competitor. For lead generation businesses, a thoughtful follow-up ad can bring a high-intent visitor back to complete a form or book a consultation after they have compared several providers.

The difference comes down to causality. Did the ad change behaviour, or did it merely appear shortly before a conversion? Attribution reports are designed to answer the second question. Profitable scale requires answering the first.

Why reported ROAS makes retargeting look stronger than it is

Most platform attribution models award credit after an ad click or view within a set conversion window. That is useful directional data, but it does not account for what would have happened in the ad’s absence.

Consider a repeat customer who directly visits your site to replenish a product every month. If they happen to see a retargeting ad three days before ordering, the platform can claim the purchase. The reported revenue may be accurate, but the claimed influence is overstated. The same issue arises when a customer searches for your brand after seeing a prospecting ad, then converts after a retargeting impression.

This effect gets stronger when audiences are too broad. Including all website visitors for 180 days, existing customers, email subscribers and recent purchasers in one retargeting pool creates a large audience full of people with very different intent. It also creates plenty of low-cost attributed conversions that can conceal inefficient spend.

View-through attribution deserves particular scrutiny. It can be valuable for understanding exposure, especially on visual platforms, but it should not be used as the sole basis for allocating budget. A person seeing an ad is not evidence that the ad caused their purchase.

The answer is not to switch retargeting off by default. The answer is to use a measurement framework that separates attributed efficiency from incremental value.

Start with the business metrics, not the ad account

The first signal of cannibalisation is often visible outside the advertising platform. If retargeting spend rises sharply while total revenue, new-customer revenue, contribution margin and conversion rate remain flat, it is worth investigating. A growing retargeting ROAS alongside declining prospecting reach is another warning sign.

For eCommerce brands, evaluate blended performance alongside channel-level results. Look at total marketing spend as a proportion of revenue, new customer acquisition cost, repeat purchase behaviour and contribution margin after advertising. For lead generation, focus on qualified leads, sales-accepted leads, booked meetings, close rate and revenue by cohort. A cheap platform lead that never reaches the sales team is not growth.

Tracking infrastructure matters here. Clean purchase events, reliable first-party data, consistent UTMs and CRM feedback make it far easier to understand where prospects first entered the journey and what happened afterwards. Without that foundation, teams tend to optimise towards the neatest-looking platform report rather than the commercial outcome.

How to measure whether retargeting is incremental

The most reliable approach is a controlled holdout test. Create a group of eligible users who do not receive retargeting ads, then compare their conversion rate and revenue with a similar group that does. The difference between the two groups is the incremental lift created by retargeting.

Platforms offer experiments that can support this, but the test design matters more than the button you click. The audiences must be large enough, the test must run long enough to account for purchase cycles, and the exclusion group must genuinely be withheld from the relevant ads. For lower-volume brands, a geo-based test can be more practical: reduce or pause retargeting in matched regions and compare the movement in total outcomes.

Do not judge a test by retargeting ROAS alone. Measure the change in total revenue, new customer orders, conversion rate and profit against the cost saved. If withholding ads produces little or no decline in total sales, the campaign was likely harvesting demand. If sales fall meaningfully and profit improves when ads return, you have evidence that retargeting is adding value.

A simpler diagnostic can help before a formal experiment. Reduce retargeting spend gradually, rather than turning it off overnight, and monitor blended performance over several weeks. This is not as conclusive as a randomised holdout, because seasonality and other campaigns can influence the result, but it can reveal whether a supposedly essential campaign is carrying more credit than impact.

Segment audiences by intent and recency

A single retargeting campaign for everyone who has visited your site is rarely the best use of budget. Intent and recency should shape both the message and the bid.

Someone who abandoned a checkout yesterday is very different from a person who read a blog post 90 days ago. The first group may respond to delivery information, returns reassurance or a reminder of the product they considered. The second may need an entirely different creative angle, or may be better reached through prospecting later rather than expensive retargeting now.

Exclude recent purchasers unless there is a clear cross-sell or replenishment strategy. This is an easy way to avoid paying to re-convert customers who have already completed the action you wanted. Also separate existing customers from prospects. Retention can be highly profitable, but it should have its own targets and reporting rather than inflating acquisition results.

For most accounts, shorter recency windows deserve the greatest attention because intent is highest. That does not mean bidding aggressively without limit. As audiences become smaller, frequency can climb quickly. Excessive repetition can waste budget, damage brand perception and encourage customers to wait for discounts.

Keep retargeting in proportion to prospecting

Retargeting cannot create demand at scale if prospecting is not continually bringing qualified people into the funnel. An account with an oversized retargeting allocation often looks efficient precisely because it is feeding on demand generated elsewhere.

There is no universal budget split. Brands with short purchase cycles, high traffic volumes and a clear abandoned-basket problem may warrant more retargeting investment. Brands with longer consideration periods or limited site traffic may need to prioritise prospecting and nurture through email, content and sales follow-up. The right balance depends on audience size, buying cycle, margins, repeat purchase rate and the strength of your owned channels.

Treat retargeting as part of one connected acquisition system. Prospecting creates future demand. Landing pages and offers convert interest. Email and SMS capture recoverable intent. Retargeting supports the journey where it can prove an incremental contribution. When each channel has a distinct role, the team can make decisions using data rather than allowing the last touch to take all the credit.

Make better decisions with incrementality in mind

A performance-focused team should be comfortable reducing spend on a high-ROAS campaign when the marginal return is weak. That can feel counterintuitive, especially when platform dashboards are reporting exceptional numbers. But protecting budget from cannibalisation creates room to test new audiences, improve creative, strengthen conversion rate and acquire customers who would not otherwise have found you.

At Lightspeed Digital Media, we treat attribution as a decision-making tool, not a scoreboard. The aim is not to make retargeting look good. It is to understand where each additional pound of spend creates profitable, sustainable growth.

The most useful closing question for your next performance review is simple: if these retargeting ads disappeared for a month, what would change in the business – not just in the ad account? Build your testing plan around that answer, and your media budget will start working harder for real growth.