We ran one of those free ad audits. The kind every tool offers now: connect your Meta account, connect Google, get a "health score."

For one of the D2C brands it came back as 72/100. "Good, with room for improvement."

Recommendations: Consolidate ad sets (too much fragmentation), increase budget on top-performing campaign, pause 3 underperforming campaigns, test more video creatives.

Standard stuff. We followed the advice for 6 weeks.

Revenue went up 8%. Returns went up 34%. Actual profit went down.

Turns out the "top-performing campaign" the audit wanted us to scale had a 41% return rate that nobody (not the audit tool, not Meta, not Google) could see. And the 3 campaigns it told us to pause? Two of them had the lowest return rates in our account.

The audit was technically correct about everything it could see. It was dangerously wrong about everything it couldn't.

What the Audit Actually Audits

Break down what a typical free ad audit examines:

Account structure. Campaigns, ad sets, targeting overlap. Are you organized cleanly or is it a mess?

Spend distribution. Budget allocation across campaigns. Are you putting money in the right places according to platform metrics?

CPM, CPC, CTR benchmarks. Are you above or below industry averages? Is your cost per click reasonable for your category?

Platform-reported ROAS. Which campaigns are "profitable" according to Meta and Google?

Creative performance. Engagement rates, click-through, frequency. Is your creative fatigued? Is it resonating?

Audience overlap. Are you wasting spend targeting the same people across multiple ad sets?

All of this is ad platform data. It's the equivalent of auditing a restaurant by looking at the menu and the number of orders. Without checking if the food was actually served, if customers sent it back, or if the kitchen is losing money on every dish.

What it doesn't see:

Which campaigns drive high return rates. Not at account level. At campaign level. The variance is massive.

Which campaigns attract COD failures. Some creatives and targeting combinations drive impulse buyers who refuse delivery. The ad platform has no idea.

Whether "conversions" are being double-counted across platforms. Meta says it drove the sale. Google says it drove the sale. Reality: one sale, two claims.

What the actual margin is after fulfillment. Shipping costs, COD charges, packaging, delivery attempts. None of this shows up in ad platform data.

Whether the customer ever came back for a second purchase. A 5x ROAS campaign that acquires one-time buyers is worth less than a 2.5x ROAS campaign that brings repeat customers.

Whether the attributed revenue even matches Shopify revenue. Platforms overclaim. Sometimes by 40-60%. Your audit just accepts their numbers.

The audit scores your ads against what the ad platform thinks happened. Not against what your bank account thinks happened.

The Audit That Recommended Scaling Our Worst Campaign

Let me tell you about Campaign X.

The audit flagged it as the star. Meta ROAS: 4.8x. CPM was low, ₹180 when the account average was ₹240. CTR was high, 2.3% when most campaigns were at 1.4-1.6%. Engagement was strong. Comments, shares, saves. All healthy.

Every metric the audit could see said "scale this."

We did. We tripled the budget over 4 weeks. Went from ₹45,000 a day to ₹1.35 lakhs a day.

Orders went up. Dashboard looked great. Meta kept reporting 4.6-4.9x ROAS. We felt smart.

Then the operations team started complaining.

What the audit couldn't see: Campaign X was a flash sale creative with aggressive discounting. "Flat 60% off + free shipping + extra 10% on prepaid." It drove massive order volume. Mostly COD, mostly from Tier 2-3 cities with historically high failure rates in our data. The creative attracted impulse buyers, not high-intent customers.

The post-purchase data (which no audit tool asked for):

41% of orders returned or refused. Either "didn't like the product" or "changed mind" or straight-up delivery refusal.

Average 2.3 delivery attempts before confirmation. Multiple re-delivery trips eating into margins.

COD failure rate: 38% in the pin codes this campaign was reaching. Customer ordered, delivery agent went twice, customer never picked up the phone.

Net margin after returns and shipping: negative ₹120 per order. Every order was costing us money after accounting for product cost, shipping both ways, COD charges, and wasted inventory on returns.

We were scaling a campaign that was actively destroying the business. And the audit (the free, data-driven, "AI-powered" audit) told us to do it.

Meanwhile, Campaign Y.

One of the three campaigns the audit said to pause. "Underperforming. Low ROAS. Recommend reallocation."

Meta ROAS: 2.1x. Modest. CPM was higher at ₹310. Spend efficiency looked worse on every platform metric.

But Campaign Y was a product education creative. Longer video, less aggressive discount. "Here's how this works, here's why it's worth it." Targeting Tier 1 cities, slightly older demographic, people who'd visited the website before.

The real data:

87% prepaid orders. High-intent buyers who weren't just clicking because of a discount.

9% return rate. They knew what they were buying. They kept it.

Net margin: ₹680 per order after all costs. These were profitable customers.

Repeat purchase rate: 3.2x higher than the account average. They came back. Campaign Y wasn't just acquiring customers, it was acquiring the right customers.

The audit recommended killing the most profitable campaign and scaling the most destructive one.

And it was right by every metric it had access to.

That's the problem.

What a Real Audit Looks Like

The audit nobody offers for free. Because it requires data nobody connects.

Layer 1: Platform reconciliation

Sum all platform-attributed revenue across Meta and Google. Then compare to Shopify.

If Meta claims ₹52 lakhs in attributed revenue and Google claims ₹33 lakhs and Shopify shows ₹58 lakhs in actual revenue, your platforms are collectively overclaiming by 46%.

Now you know how much to discount every platform metric. When Meta says a campaign drove ₹12 lakhs, the real number is probably closer to ₹7-8 lakhs. You need a correction factor before you make any scaling decision.

Most audits skip this. They just accept platform data as truth.

Layer 2: Post-purchase overlay

Map each campaign (not your account, each campaign) to delivery outcomes. Return rate. COD success rate. Number of delivery attempts. Net margin after fulfillment costs.

This is where you find out that Campaign A has a 12% return rate and Campaign B has a 38% return rate even though they have identical platform ROAS. The variance between campaigns is massive and it's completely invisible from the ad platform.

You can't optimize what you can't see. And if you're only looking at Meta and Google, you're flying blind on the metrics that actually determine profitability.

Layer 3: Customer quality scoring

Which campaigns drive one-time buyers? Which drive repeat customers?

A campaign that acquires customers with a 2.8x LTV-to-CAC ratio over 6 months is worth 5x more than a campaign that drives impulse purchases with identical upfront ROAS. But platform data treats them the same.

Track cohorts by campaign. See who comes back. See who refers others. See who upgrades to higher-value products. That's customer quality, and it's the difference between growth and churn.

Layer 4: Cross-platform attribution

Is Campaign X on Meta actually driving conversions, or is it just touching customers who were already going to buy through Google Search?

De-duplicate. See the real incremental contribution. Otherwise you're counting the same customer twice and making decisions on inflated numbers.

This requires matching user journeys across platforms, not just accepting each platform's self-reported attribution. Most audits never do this. They stack platform reports side by side and call it analysis.

This is a full audit. It requires ad data, Shopify data, delivery and logistics data, and return data in one place. Connected. Mapped. At campaign level, not account level.

That's why nobody offers it for free. It's not a dashboard scan, it's a data integration problem. And most tools can't (or won't) integrate outside the ad platforms.

The 3 Questions Your Free Audit Can't Answer

"Which of my campaigns is actually profitable after returns and delivery costs?"

Not "which campaigns have good ROAS on Meta." Which campaigns, when you account for returns, COD failures, shipping both ways, re-delivery attempts, and actual margin per product, actually make money?

Your free audit doesn't know. It can't know. It doesn't have the data.

"Am I scaling campaigns that look good on the platform but lose money in real life?"

This is the nightmare scenario. You follow the audit's advice. You scale the "winner." Revenue goes up. You feel like you're winning.

Then the returns come in. The COD failures stack up. The profit disappears. You were optimizing for a metric that had no connection to business reality.

Your free audit won't catch this. It's scoring you on platform performance, not P&L.

"Which campaigns drive customers that actually come back?"

Customer lifetime value. Repeat purchase rate. Referral behavior. These are the metrics that determine whether you're building a business or renting revenue.

A campaign that brings one-time discount shoppers will always look worse in an audit than it should. A campaign that brings loyal customers will always look better. But the audit can't tell the difference because it stops at the click.

If your audit can't answer these three questions, it's auditing the ad. Not the business.

And an ad audit that can't see the business is just confirming your existing dashboard in a nicer PDF.

What to do?

Every ad tool and agency offers a "free ad audit." Import your Meta and Google accounts, get a health score, see where you're "wasting budget."

It looks thorough. It feels data-driven.

But here's what every free audit misses: it can only see what the ad platform sees. And the ad platform has no idea what happened after the click.

An ad audit that stops at the platform is like a financial audit that only looks at revenue. The real story (returns, COD failures, attribution overlap, actual margin) is invisible to any tool that only reads ad data.

This is the gap we built Predflow to close. We're connecting ad platform data to Shopify, delivery, returns, and customer data so that when you audit your ads, you're auditing what actually happened to the money. Not just what Meta thinks happened. 👇