Marketing Spend Optimization
Where Do E-commerce Brands Waste Ad Spend?
Your branded search ROAS looks great. Retargeting looks even better. So why does growth keep stalling? Here is where the non-incremental spend hides, and how to prove it.
Typically, e-commerce brands waste ad spend in three places: branded search, retargeting and promotions. Branded search captures shoppers who would have found you anyway, retargeting chases people already about to buy, and promotions pulls demand forward instead of creating it. In all three, the dashboard shows a strong return. In all three, much of that return would have happened without the ad.
In ELIYA’s analysis, on average 20% of ad spend is wasted in many e-commerce brands. Brands typically unlock a further 5 to 15% of incremental growth once that spend is reallocated with ELIYA AI.
Your branded search ROAS looks great. Retargeting looks even better. So why does growth keep stalling? Because platforms grade their own homework. They report every conversion they touched, but not the ones they caused.
This guide illustrates where wasted ad spend hides, review the evidences, and discusses how to find wasted ad spend in your budget.
TL;DR
- Wasted ad spend in e-commerce is non-incremental spend: money that gets credit for sales that would have happened anyway.
- Branded search, retargeting, and promotion-period advertising are the three most common hiding places, because all three reach people who are already close to buying.
- In ELIYA’s analysis, on average 20% of ad spend is wasted in many e-commerce brands, and reallocating it typically unlocks a further 5 to 15% of incremental growth.
- You can test each area directly by switching the spend off for a control group and measuring what happens to sales.
- Beliani, a European e-commerce retailer, found close to 24% of its media budget was overspent on saturated channels and reallocating it drove 2.95M in incremental revenue over six months.
What Counts as Wasted Ad Spend in E-commerce?
Wasted ad spend is money spent on ads that did not cause the sale but they get credit for it. It is different from a bad ad. A beautiful ad is still waste if the customer was going to buy anyway.
Why Does In-Platform ROAS Hide Waste?
Ad platforms rely on attribution, which credits the ads a shopper touched before buying. It never asks what would have happened without those ads. Because attribution has no counterfactual, it cannot separate demand an ad created from demand it intercepted. That is why in-platform ROAS reporting consistently overstates channel performance.
What Is Incrementality?
Incrementality: the additional sales, revenue, or conversions that happened because of a marketing activity and would not have happened without it.
iROAS (incremental return on ad spend): the extra revenue caused by advertising, divided by the ad spend, counting only incremental sales rather than every attributed sale.
ROAS (return on ad spend) asks what your ads touched. iROAS asks what your ads caused. The gap between the two is your waste.
Where Does Waste Hide?
Area | What the dashboard shows | What may be happening |
|---|---|---|
Branded search | High ROAS, low cost per order | Capturing shoppers who were already searching for you |
Retargeting | Often the highest ROAS of any channel | Chasing people who were about to buy |
Promotion-period ads | Sales spike during the sale | Demand pulled forward from later weeks |
Is Branded Search Wasted Ad Spend?
Often, yes. Branded search feels like your best channel because it converts at a high rate for a low cost. But the shopper typed your name into Google. They were already on their way.
Why Does Branded Search Look So Good?
It sits right before checkout, so last-touch attribution hands it the credit. Budget follows the credit. The video and social campaigns that build brand awareness, the thing brand lift studies measure, get starved. Branded search harvests demand. But it rarely creates it.
What Does the Evidence Say About Branded Search Incrementality?
In a large field experiment at eBay, researchers switched off paid ads on brand-name searches for part of the audience. About 99.5% of that paid traffic arrived through free organic results instead. The ads added almost no sales. eBay is a household name, so treat this as the extreme end of the range, not a forecast for your brand.
When Is Branded Search Incremental?
When competitors bid on your name. In Haus’s analysis of brand-search tests, 82% of brands facing heavy competitor bidding measured significant lift, against 35% of brands with little competition. If a competitor take the top slot the moment you switch off, your ad is defending revenue, not wasting spend.
How Do You Test Branded Search Incrementality?
Run a holdout test.
Holdout test: an experiment in which a control group or region is deliberately not shown a campaign, so its sales reveal what would have happened without it.
Pause branded ads in a set of matched regions and keep them running in the rest. Then compare sales. New to this method? Start with what incrementality testing is and how it works for e-commerce brands.
Is Retargeting Wasted Ad Spend?
Often, and the cause is selection, not creative. Retargeting shows ads to warm leads, which are people who already visited your site, viewed a product, or added it to their cart. Those are the shoppers most likely to buy anyway. The platform then counts the purchase that follows.
What Does the Research Say About Retargeting Incrementality?
Retargeting works on some people. As AI Digital summarizes, a randomized study in the Journal of Marketing Research found 14.6% more users returned to an advertiser’s site within four weeks. The same article cites a 2026 experiment in the Journal of Interactive Marketing where prospecting delivered large, consistent incremental returns, while retargeting delivered significantly lower impact.
What Does a Retargeting Holdout Look Like?
Let’s look at an example. Say a brand spends 100,000 a month on retargeting and the platform reports 6x ROAS: 600,000 in attributed revenue. The team holds out a random 20% of the audience. Exposed users convert at 3.0%. The holdout converts at 2.7%.
Only one conversion in ten was caused by the ads. Incremental revenue is about 60,000, so iROAS is 0.6. The dashboard says 6X. The experiment says 0.6X. (These numbers are hypothetical, for illustration). A step-by-step lift analysis shows how to run this test on your own campaigns.
Do Promotions Pull Demand Forward Instead of Creating It?
Sometimes. Promotions create a spike in sales. But part of that spike is customers who would have bought at full price next week.
Promo pull-forward: when a promotion moves purchases that would have happened later into the sale window, so sales spike during the promotion and dip afterwards, with little or no net gain.
Why Does Pull-Forward Inflate Ad Performance?
Promo weeks are when you spend the most and when conversion rates peak. Every platform claims the surge. Nobody counts the dip that follows. This is specifically common after Black Friday. Weekly ROAS looks excellent while the quarter’s revenue barely moves. Part of the credit belongs to the discount, not the ad.
How Do You Separate Promotion Effects from Ad Effects?
Pull-forward is invisible inside a single week. You should look across weeks and control for the promotion itself. That is exactly what a Marketing Mix Model is designed do.
Marketing Mix Modeling (MMM): a statistical method that uses aggregated spend and sales data over time to estimate how much each marketing channel, promotion, and outside factor contributes to sales.
For Beliani, ELIYA added the retailer’s own promotion calendar, event effects such as Black Friday, weather, and Google Query Volume as controls.
Step by Step; How Do You Find and Fix Waste in Your Ad Budget?
- Sort spend into intercept versus create. Label each campaign as reaching people already close to buying (branded search, retargeting, promo-period ads) or reaching new demand (prospecting, video, social).
- Hold out branded search. Pause brand ads in matched regions and compare sales against regions where they keep running.
- Hold out a retargeting audience. Withhold ads from a random slice of your retargeting pool and compare conversion rates.
- Put your promo calendar in the model. Add promotion dates, if possible add discount depth, and event effects as controls so pull-forward shows up as a dip.
- Calibrate your MMM with the test results. Feed experiment results back into the model so its channel estimates match measured reality.
- Reallocate, then re-test. Move budget away from channels at diminishing marginal returns and toward proven incremental return. Then confirm with a new test.
How Does ELIYA Find Non-Incremental Spend?
ELIYA starts with an incrementality audit. The audit tests each major campaign against one question: did this spend cause sales, or just collect credit for them? Next, ELIYA builds an always-on Marketing Mix Model, refreshed monthly, that includes your promotion calendar and event effects.
ELIYA AI then turns the calibrated model into a specific budget plan across every channel, reviewed by ELIYA’s experts before you act. The AI does the heavy analysis. Human judgement makes the call. This is the core of ELIYA’s e-commerce MMM solution. To see what an audit would find in your own budget, book a free measurement audit.
Who Should Audit Their Ad Spend for Waste?
Best Fit | Not the Right Fit Yet |
|---|---|
Growth-stage e-commerce brands with $2M or more in annual ad spend, running branded search, retargeting, and regular promotions | Brands running a single paid channel at a spend level where in-platform ROAS is directionally good enough |
Teams whose ROAS looks strong while overall growth stays flat | Very early-stage brands still validating product-market fit rather than optimizing an existing budget |
CMOs and Heads of Growth who must defend budget reallocation in front of finance | Teams that only need day-to-day, campaign-level optimization |
What Should E-commerce Brands Do Next?
Wasted ad spend hides where customers are already close to buying. Branded search intercepts demand, retargeting follows buyers, and promotions borrow sales from next week. Dashboards credit all three. Only experiments and models that separate cause from correlation show what is really working.
Start by holding out branded search and a retargeting audience, and incorporate your promo calendar into your MMM model.
If you’re a growth-stage e-commerce brand with $2M or more in ad spend and you want to know how much of your branded search, retargeting, and promotion budget is non-incremental, ELIYA AI is built for this.
Frequently Asked Questions
How Much Ad Spend Do E-commerce Brands Typically Waste?
In ELIYA’s analysis, on average 20% of ad spend is wasted in many e-commerce brands. The exact share depends on your channel mix, how promotion-heavy your calendar is, and how much of your spend targets people already close to buying.
Should You Turn Off Branded Search Completely?
Not without testing first. Brand bidding can be close to pure waste for a strong brand that no competitor bids against, and valuable protection where rivals do bid on your name. Pause it in a few matched regions, compare sales and decide accordingly.
Is Retargeting Ever Worth It?
Yes, when it is incremental. Retargeting can lift return visits, but its value depends on the audience, the ad frequency, and how close the shopper already is to buying. A holdout test shows which audiences are worth paying for.
Can You Find Wasted Ad Spend Without Running Experiments?
Partly. A Marketing Mix Model estimates each channel’s contribution from historical spend and sales data, so it needs no user-level tracking. It works best with at least two years of weekly data and will get better when it’s calibrated with experiments.
How Long Should an Incrementality Test Run?
It depends on the channel, your traffic, and your customer journey length. According to Haus’s analysis, branded search effects show up almost immediately. Channels that build demand over time, such as video, need longer windows to show their effect.
