Why Does ROAS Look Good but Revenue Is Down?

Why ROAS Can Look Good While Revenue Falls

ROAS can look good while revenue is down because ad platforms are not measuring the same thing as your store. Meta, Google, and many blended dashboards can report strong attributed revenue while your ecommerce platform shows weaker net sales, lower margin, worse customer quality, or fewer truly incremental orders.

The most common explanation is that the ad report is giving credit for revenue that does not fully translate into business value. That can happen when paid ads are credited for orders from returning customers, when retargeting captures buyers who were already likely to purchase, or when the reported revenue is based on gross order value before refunds, discounts, product costs, shipping costs, payment fees, or repeat-purchase quality are considered.

That means the first move is not automatically to cut budget or trust the ad dashboard. The first move is to reconcile three layers:

1. What the ad platform says it influenced. Purchases, conversion value, and ROAS by campaign or ad set.
2. What the store actually recorded. Orders, customers, gross sales, net sales, discounts, refunds, products, and repeat purchase behavior.
3. What the business kept. Contribution margin, refund-adjusted revenue, customer quality, and payback.

ROAS is useful, but it is not the final source of truth. It is an ad-reporting metric. Store order data tells you whether the revenue was real, retained, profitable, and attached to customers worth acquiring again.

Key terms to separate before you diagnose

The Source-of-Truth Reconciliation Map

When ad ROAS and store revenue disagree, do not start with opinions about which dashboard is “right.” Start by mapping what each system is allowed to know.

The ad platform can tell you which campaigns it believes influenced purchases. Your store can tell you which orders actually happened. Your order economics can tell you whether those orders were valuable after discounts, refunds, costs, and repeat behavior.

Minimum order export fields for the investigation

Pull a clean order export for the period where ROAS looked good but revenue weakened. If possible, also pull the prior comparable period so you can see what changed.

• Order ID
• Order date
• Customer ID or hashed email
• First-order flag or customer order count at purchase
• Channel, source, medium, campaign, or landing page if available
• Discount code and discount amount
• Gross sales
• Net sales
• Refund amount and refund date
• Tax and shipping amounts
• Product title, SKU, variant, and quantity
• COGS, gross margin, or margin proxy by SKU
• Payment or fulfillment cost if available
• Repeat purchase status
• 30-day, 60-day, and 90-day customer revenue where available

Do not worry if every field is not perfect. A partial order-level view is still more useful than trying to solve the issue from aggregate ROAS alone.

Diagnostic Decision Tree: Which Leak Explains the Gap?

Use this decision tree when Meta, Google, or a blended dashboard says performance is healthy but the store says revenue quality is getting worse.

1. Are ad-attributed purchases higher than store orders?

If yes, the first suspect is attribution inflation or duplicated credit. The ad platform may be modeling conversions, claiming purchases that happened through another path, or counting the same customer journey differently than your store.

• Compare platform-reported purchases to actual store order count for the same dates.
• Check whether multiple platforms are claiming the same order value.
• Look for retargeting campaigns with unusually high ROAS and high returning-customer share.
• Review attribution windows and whether view-through credit is included.

Decision: Do not scale based only on platform ROAS. Reframe reporting around store orders, blended MER, new-customer CAC, and incrementality testing.

2. Is gross revenue up but net revenue down?

If yes, the issue is likely happening after the order is placed. Discounts, refunds, returns, cancellations, taxes, shipping adjustments, or product-level economics may be erasing the apparent gains.

• Compare gross sales, net sales, discount amount, and refund amount by week.
• Separate refunded revenue by product, campaign, discount code, and customer cohort.
• Check whether refunds are posting after the ad platform has already counted the sale.

Decision: Treat ROAS as incomplete until it is adjusted for the value you actually keep.

3. Is AOV falling while ROAS holds steady?

If AOV is falling, the campaign may be converting customers into smaller baskets. That can happen when traffic shifts toward entry-level products, single-unit purchases, aggressive discounts, or landing pages that do not encourage bundling.

• Compare first-order AOV by campaign or source.
• Review units per transaction and product combinations.
• Check whether free-shipping thresholds, bundles, or upsells are being missed.

Decision: Fix offer architecture, merchandising, bundles, or landing-page paths before assuming the channel is healthy.

4. Are discounts rising faster than revenue?

If discounts are rising, ROAS may be rewarding conversion volume while profit gets worse. A campaign can produce more attributed revenue and still hurt the business if customers require deeper incentives to buy.

• Segment orders by discount code, discount percentage, and customer type.
• Compare discounted AOV against full-price AOV.
• Check whether the discount is mostly going to new customers, returning customers, or customers who were already likely to purchase.

Decision: Test tighter thresholds, bundles, gift-with-purchase offers, or lower discount depth instead of scaling the same promotion.

5. Are refunds clustered by product, campaign, or cohort?

If refunds cluster around a specific SKU, variant, promise, or acquisition cohort, the ad report is showing demand but the post-purchase experience is failing. Common causes include sizing confusion, product quality, misleading creative, shipping damage, fulfillment delays, or mismatched customer expectations.

• Rank SKUs by refund amount, refund rate, and refund reason.
• Compare refund behavior between first-time and returning customers.
• Review whether one campaign is pushing a product with above-average refund pressure.

Decision: Pause or reduce promotion of the problem SKU until the product page, creative, sizing, quality, or fulfillment issue is fixed.

6. Is paid revenue mostly returning customers?

If paid campaigns are getting most of their revenue from returning customers, acquisition ROAS may be overstated. Retargeting, branded search, and high-intent remarketing can capture revenue that email, organic, direct, or natural repeat behavior might have captured anyway.

• Split paid-attributed orders into first-time and returning customers.
• Compare new-customer CAC against contribution margin and expected payback.
• Check whether retargeting budgets are growing while new-customer volume is flat.

Decision: Separate acquisition reporting from retention capture. Do not let returning-customer ROAS hide weak new-customer economics.

7. Are new customers failing to repeat?

If first-order ROAS looks acceptable but 30/60/90-day repeat purchase is weak, the campaign may be buying low-quality customers. They purchase once, often with a discount, then do not come back soon enough to recover acquisition cost.

• Group customers by first purchase month, source, campaign, product, and discount code.
• Measure repeat purchase rate and cumulative revenue at 30, 60, and 90 days.
• Compare cohorts acquired through paid social, paid search, organic, email, affiliates, and direct.

Decision: Shift budget toward sources, products, and offers that create better customer cohorts, not just cheaper first orders.

8. Is the campaign selling low-margin or one-time-purchase products?

A campaign can produce strong ROAS by selling products that are easy to buy but poor for long-term economics. Examples include low-margin hero SKUs, clearance products, accessories with low basket expansion, or one-time-purchase items with little repeat potential.

• Join campaign performance to SKU margin and repeat behavior.
• Compare product-level contribution margin, not just product-level revenue.
• Check whether landing pages are steering paid traffic away from higher-quality products.

Decision: Redirect spend, creative, landing pages, and merchandising toward products that produce better retained value.

Order Data Metrics to Check Before Changing Budget

Before you cut or scale spend, build a short diagnostic view from your order data. The goal is to see whether the ROAS gap is caused by attribution, revenue quality, customer mix, or product economics.

Start with the revenue bridge

Create a simple bridge from ad-attributed revenue to the revenue the business actually keeps.

Example: strong campaign ROAS, weaker business outcome

Imagine a campaign reports strong ROAS because it generated a large amount of attributed gross revenue. Store-level analysis shows a different picture:

• The campaign relied on a deep first-order discount.
• Most orders contained a low-margin entry product.
• A high share of customers were returning buyers who already had recent email engagement.
• Refunds were concentrated in one promoted SKU.
• New customers from the campaign had weak 60-day repeat purchase.

In the ad platform, that campaign can look like a winner. In the order data, it may be a margin leak, a retargeting over-credit problem, or a low-quality acquisition source.

Questions to answer before touching spend

• Did actual store orders increase, or did only attributed orders increase?
• Did net sales increase, or only gross sales?
• Did contribution margin increase after discounts, refunds, COGS, and shipping?
• Did new-customer count increase?
• Did new-customer CAC stay within an acceptable payback window?
• Did first-order AOV change?
• Did the product mix shift toward lower-margin or higher-refund products?
• Did customers from the campaign repeat at the expected rate?

If the answer to several of these is no, strong ROAS should not be treated as permission to scale.

Channel and Customer Quality Analysis

A channel can produce revenue that looks efficient and still create poor customer economics. That is why the diagnosis has to separate channel performance from customer quality.

The most useful cut is not just campaign versus campaign. It is campaign, customer type, first product purchased, discount exposure, refund behavior, and repeat purchase timing together.

Segment paid performance by customer type

• First-time customers: Use this group to evaluate acquisition quality, new-customer CAC, first-order AOV, first-product mix, refund rate, and repeat purchase.
• Returning customers: Use this group to evaluate retention capture, retargeting dependency, promotional pressure, and whether paid media is taking credit for existing demand.

If returning customers are driving most of the ROAS, the campaign may still be useful, but it should not be judged as a pure acquisition engine.

Run these cohort cuts

How to tell if ROAS is coming from new or returning customers

Export orders for the campaign period and mark each customer as first-time or returning based on their order history before the purchase date. Then compare the split of revenue, order count, discounts, refunds, and margin between those two groups.

If the ad platform reports strong ROAS but most store orders are returning customers, the campaign may be benefiting from existing brand demand. If first-time customers are present but show low AOV, high discounts, high refunds, or weak repeat purchase, the issue is customer quality rather than attribution alone.

ROAS vs Revenue Leak Matrix

Use this matrix to move from symptom to action. The goal is to identify the specific leak before changing budget.

What to Do Next Based on What You Find

The right action depends on the leak. Avoid the generic response of “cut spend” or “scale what has the best ROAS.” Use the finding to decide what should change.

If attribution is inflated

• Use store orders as the baseline for revenue truth.
• Track blended MER alongside platform ROAS.
• Separate prospecting, retargeting, branded search, and retention capture.
• Run holdout or incrementality tests where possible.
• Manage acquisition decisions around new-customer CAC, not total attributed ROAS.

If discounts are the issue

• Compare profit and repeat purchase by discount code.
• Test minimum order thresholds instead of deeper blanket discounts.
• Use bundles, gifts, or value-add offers where margin allows.
• Limit discounts for returning customers who would likely purchase without them.
• Watch whether reduced discounting changes conversion, AOV, and contribution margin together.

If refunds are clustered

• Identify the products, variants, campaigns, and cohorts with the highest refund impact.
• Audit product pages for sizing, materials, compatibility, shipping expectations, and use-case clarity.
• Review creative claims that may be creating the wrong expectation.
• Check fulfillment timing, packaging, damage, and support tickets.
• Slow spend to the affected SKU until the refund driver is fixed.

If AOV is compressed

• Review whether campaigns are pushing low-priced entry products without basket expansion.
• Test bundles, quantity breaks, cross-sells, and post-add-to-cart upsells.
• Adjust free-shipping thresholds based on actual order distribution.
• Route paid traffic to landing pages that support multi-product buying.
• Measure AOV by first-time and returning customers separately.

If product mix is hurting margin

• Rank products by contribution margin, refund rate, repeat purchase influence, and acquisition volume.
• Shift budget away from products that generate revenue but weak retained value.
• Update landing pages and creative to feature higher-quality products.
• Build campaigns around products that lead to stronger second orders.
• Use margin-adjusted reporting in budget reviews.

If repeat purchase is weak

• Find the expected second-order window for each acquisition cohort.
• Trigger lifecycle campaigns before customers fall behind expected repeat timing.
• Segment post-purchase journeys by first product purchased.
• Use replenishment, education, bundles, and complementary-product flows where relevant.
• Adjust CAC targets if payback requires a second or third order.

If paid media still looks good but net sales are down

Reduce budget when the order data shows that paid spend is creating non-incremental, low-margin, heavily discounted, high-refund, or low-retention revenue. Keep spending, or cautiously scale, when store orders, net sales, contribution margin, new-customer quality, and repeat purchase behavior support the ROAS story.