Six months ago, I logged into an e-commerce client's Facebook Ads Manager and saw a 4.1× ROAS. I logged into their Google Analytics and saw 3.5×. I opened their Shopify dashboard and traced actual revenue against total ad spend.

The real number? 2.3×.

This happens in roughly 70% of accounts we audit. Not because the ads aren't working. But because last-click attribution is systematically overstating performance by an average of 60–80%. And here's the thing nobody tells you: both Facebook and Google are telling the truth from their perspective. They're just measuring different things.

After managing $125M+ in ad spend across hundreds of accounts, I've learned that attribution isn't a math problem. It's a communication problem. And it's costing you money every single day.

The Attribution Problem That Every Marketer Faces

Here's what happens in a typical customer journey:

A user sees a Facebook ad for your product on Monday. They don't click. On Wednesday, they see a Google search ad and click. On Thursday, they see a retargeting ad and purchase. That same customer also clicked an email link, visited your site directly, and browsed for 15 minutes.

Who gets credit for the conversion?

Last-click attribution says: the retargeting ad (the last touchpoint before purchase).

Facebook says: the Facebook ad (because of view-through conversion data).

Google says: the Google search ad (because they have the click).

Reality says: all three channels contributed, and the contribution wasn't equal.

This isn't fraud. It's a structural problem with how attribution works. But it costs you money because:

1. You overspend on platforms with good last-click visibility. Google and Facebook will always claim disproportionate credit because they own more of the customer journey data.
2. You underspend on awareness channels. YouTube, TikTok, and brand awareness campaigns get zero credit for the customers they introduce, even though they're often the first touchpoint.
3. You make scaling decisions on phantom data. You increase budgets on "high-performing" channels that are actually just good at claiming credit.

What Last-Click Attribution Actually Measures (vs What It Claims)

Let me be precise here: last-click attribution measures one thing and one thing only: which platform was closest to the conversion moment.

It doesn't measure influence. It doesn't measure impact. It doesn't measure which touchpoint actually changed the customer's mind. It measures timing.

Think about this: if a customer sees your Facebook ad, gets interested, then Googles your product name, then clicks the Google ad and buys—Google gets 100% credit. Was Google more effective? Or was Facebook the decision-maker and Google just the closest click to purchase?

You can't tell from last-click data.

This matters because it systematically undervalues everything except the last touchpoint. Brand awareness campaigns, YouTube views, TikTok content, email nurture—all invisible in last-click attribution. Meanwhile, search ads and retargeting (which happen closest to purchase) get over-credited.

That's not a bug. It's a feature of how last-click was designed.

The 5 Attribution Models: How They Work, What They Miss, and Why They All Lie a Little

Here's what you need to know about each model:

1. Last-Click Attribution (40% of accounts use this)

How it works: 100% credit goes to the last touchpoint before conversion.

Pros:

• Easy to understand
• Easy to implement (literally every platform offers it)
• Answers the question: "Which channel am I closest to the sale with?"

Cons:

• Systematically overvalues bottom-of-funnel channels (search, retargeting)
• Ignores all awareness work
• Incentivizes spending on click-to-conversion rather than actual influence
• Makes scaling painful—you chase diminishing returns while ignoring growth channels

When to use it: Only for paid search optimization. Only. If you're using last-click for your entire marketing stack, you're flying blind.

2. First-Click Attribution (10% of accounts)

How it works: 100% credit goes to the first touchpoint in the customer journey.

Pros:

• Good for understanding which channels introduce new customers
• Helps you see which awareness work matters

Cons:

• Opposite problem of last-click: overvalues awareness, ignores conversion work
• Makes bottom-funnel optimization invisible
• Creates an incentive to spend only on top-of-funnel channels, then wonder why you don't convert

When to use it: As a supplementary view to understand top-of-funnel performance, not as your main model.

3. Linear Attribution (30% of accounts)

How it works: Each touchpoint in the customer journey gets equal credit.

Example: If a customer touches 4 channels before converting, each gets 25% credit.

Pros:

• More fair than single-point models
• Incentivizes omnichannel thinking
• Easier to implement than data-driven models

Cons:

• Assumes all touchpoints are equally valuable (they're not)
• If a customer touches 10 channels, each gets 10% credit—which might reward bad channels
• Still invisible to how much time passed between touches

When to use it: As a starting point for omnichannel campaigns. Better than last-click, but still imperfect.

4. Time-Decay Attribution (15% of accounts)

How it works: More credit goes to touchpoints closer to conversion. The math can vary (exponential, algebraic), but the idea is the same: recent touches matter more.

Pros:

• More sophisticated—acknowledges that recency matters
• Balances awareness and conversion work
• Closer to how actual customer psychology works

Cons:

• Hard to understand and explain to stakeholders
• Still arbitrary (how much more credit does recency deserve?)
• Requires good data to work

When to use it: In accounts with mature data infrastructure, when you're ready to move beyond simple models.

5. Data-Driven Attribution (5% of accounts use this)

How it works: Machine learning looks at successful conversion paths vs. non-converting paths and assigns credit based on statistical impact. Google Analytics 4 offers this. Northbeam and Triple Whale build custom models.

Pros:

• Most accurate—based on actual impact, not assumptions
• Accounts for interaction effects (how channels work together)
• Adapts to your specific business

Cons:

• Requires lots of conversion data (50+ per day minimum)
• Can be expensive ($500-5,000/month if you use third-party software)
• Takes time to train and stabilize
• Still probabilistic—not truth, just better guesses

When to use it: If you have 50+ conversions per day and budget flexibility. Otherwise, start with time-decay and work up.

Why Facebook and Google Both Overclaim Conversions (And Why You Let Them)

Here's a fact that might surprise you: Facebook's 4.1× ROAS isn't fake. Google's 3.5× isn't fake. They're just measuring different things because they have different data.

Facebook's advantage: They track view-through conversions. If someone sees a Facebook ad, doesn't click, but converts within 28 days, Facebook can claim credit through pixel matching. This is legitimate—that user did see the ad. But it also means Facebook's ROAS includes conversions that clicked somewhere else to actually buy.

Google's advantage: They have the click. If someone searched for your product and clicked a Google ad, that click is theirs. Google can track this person across devices and credit the conversion if it happens on any device. Technically accurate, but also inflates the conversion count because multiple clicks for the same user get counted as separate conversions.

The blended impact: Conversions get counted by both platforms, and neither one accounts for the fact that the customer was already going to buy—they just happened to click your ad last.

The Blended ROAS Solution: The Only Number That Can't Lie

Here's what I've done in 100+ accounts to solve this: ignore platform reporting and calculate one number.

Blended ROAS = Total Revenue / Total Ad Spend

That's it. That's the only metric that can't be manipulated.

If you spent $100,000 across Facebook, Google, TikTok, and email, and made $230,000 in revenue, your blended ROAS is 2.3×. No platform can argue with this number.

Is it perfect? No. It doesn't tell you which channel was more efficient. But it's impossible to lie about. It's also the only number that matters to your business.

From there, you work backwards to attribute that revenue across channels using the models above. But the inputs come from your actual revenue data, not platform claims.

This approach has saved clients an average of 15–20% in wasted ad spend by exposing overactive channels within the first 60 days.

How to Build an Attribution Stack Without Expensive Software

You don't need Northbeam or Triple Whale to solve attribution. You need good basics.

Step 1: Implement Proper Tracking

• Pixel on every page: Facebook Pixel, Google Conversion Tag, and TikTok Pixel on your site. No exceptions.
• Conversions API: If you're on Shopify, WooCommerce, or any cart system, set up Conversions API to send events directly from your server. This handles ad blockers and privacy-friendly browsers that kill pixel tracking.
• UTM parameters: Every paid ad campaign should have UTM tags. Format: ?utm_source=facebook&utm_medium=cpc&utm_campaign=summer_sale

Step 2: Set Up GA4 as Your Source of Truth

Google Analytics 4 is free and includes data-driven attribution modeling. It's not perfect, but it's better than last-click.

• Link GA4 to Google Ads, Facebook Ads, and any other paid channels
• Create a custom event for actual revenue (not just button clicks)
• Set up audience exports back to your ad platforms

Step 3: Build a Simple Attribution Spreadsheet

I'm not kidding. One of our best clients uses a Google Sheet with four columns:

1. Date
2. Channel
3. Spend
4. Revenue (from Shopify export)

Updated daily. ROAS calculated automatically. Trends visible in 30 days. Cost: $0. Time investment: 5 minutes per day.

This simple spreadsheet catches platform over-claiming faster than most attribution software because it forces you to reconcile claimed conversions against actual revenue.

When to Actually Pay for Attribution Software

I'm not against tools like Northbeam, Triple Whale, or Littledata. I use them in bigger accounts. But they only pay for themselves under specific conditions:

• Monthly ad spend over $50,000: Below that, the insight gain doesn't justify the cost.
• 5+ active channels: If you're mostly Facebook ads, you don't need fancy attribution. Last-click is close enough.
• Mature data infrastructure: Your tracking needs to be solid. If you're still fixing pixel issues, a $2,000/month tool won't help.
• Willing to change based on insights: Most accounts don't. They buy the software, ignore the insights, keep running the same campaigns. If that's you, skip it.

If you meet all four criteria, software pays for itself by reallocating 10–15% of budget to more efficient channels.

The Offer-Attribution Conversation With Clients and Teams

This is where I see most breakdowns happen. You discover that attribution was inflated, and suddenly people think the Facebook manager is incompetent or the campaigns are broken.

Here's the script I use:

"Your actual blended ROAS is 2.3×, but Facebook reports 4.1× and Google reports 3.5×. This isn't anyone's fault. Here's why it happens: [explain the model above]. The good news is we now know our real baseline. From here, we optimize toward it. We're not broken—we're just seeing clearly for the first time."

Most stakeholders understand this. Some push back. Those people usually have a political interest in high-reported numbers. Address that directly.

Real Example: Where Last-Click Was Lying (And How We Fixed It)

E-commerce client, beauty products, $30K/month ad spend. Last-click attribution showed:

• Facebook: 4.1× ROAS
• Google Search: 3.7× ROAS
• TikTok: 1.2× ROAS

Budget allocation: 60% Facebook, 30% Google, 10% TikTok.

Actual revenue data told a different story. We built a time-decay model and tracked view-through conversions properly. Real performance was:

• Facebook (awareness): 2.1× ROAS
• Google Search (conversion): 3.2× ROAS
• TikTok (awareness + conversion): 2.8× ROAS

TikTok looked like it was failing, but it was introducing customers who converted later through Google. We were punishing it for not closing the sale in 28 days.

We rebalanced to 40% Facebook, 30% Google, 30% TikTok.

Three months later, blended ROAS went from 2.3× to 2.7×. Same budget, better allocation, all because we stopped trusting last-click.

The Weekly Attribution Review: 3 Numbers to Watch Every Monday

I spend 15 minutes every Monday morning on this:

Number 1: Blended ROAS (Total Revenue / Total Spend)

Week-over-week change. Simple trend. If it's going down, something's wrong. If it's stable, you're optimizing well.

Number 2: Platform Delta (Platform Reported Conversions - Actual Conversions)

Facebook reports 500 conversions, Google reports 400, actual revenue shows 650 conversions. That's an 250-conversion gap. Track this weekly. If it grows, your platforms are over-claiming.

Number 3: Channel Efficiency (Blended ROAS per Channel)

Using your spreadsheet or GA4, calculate actual ROAS per channel without letting the platform claim credit. This is your real performance baseline.

If you do this once a week, you'll catch attribution problems in days instead of months.

The Bottom Line

Last-click attribution wasn't invented to deceive you. It was invented because it was easy to measure. But it's 2024, and easy doesn't mean accurate.

Your competition is probably still making budget decisions on last-click data. That's your advantage. Implement proper attribution, calculate blended ROAS, and start allocating budget to channels based on real influence, not just who got lucky enough to be last.

This one change—moving from last-click to blended ROAS—has improved account performance by an average of 18% across our portfolio. It costs nothing. It just requires honesty about what you're actually measuring.