The CMO Guide to Marketing Attribution

Let me guess. You've got a marketing budget that's not small, you're running campaigns across five or six channels, and when someone asks "which channel is driving the most revenue," you either point at whatever dashboard you looked at last or you give a carefully worded answer that basically means "we're not entirely sure."

Don't worry. You're in very good company.

I've worked with CMOs at companies spending $50,000 a month on marketing and CMOs spending $2 million a month. The attribution question haunts every single one of them. And here's the uncomfortable truth that nobody in our industry likes to admit: perfect attribution is impossible. It doesn't exist. It never will.

But good-enough attribution? That's absolutely achievable. And it can transform how you allocate budget.

So let's get into it.

Why Attribution Is Broken (And Always Has Been)

Here's a typical customer journey for a B2B SaaS purchase:

1. CEO sees a LinkedIn post from your company (Day 1)
2. Googles your company name, reads your blog (Day 4)
3. Downloads a whitepaper, enters email (Day 4)
4. Receives nurture emails for two weeks, opens three (Days 5-18)
5. Colleague mentions your product in a meeting (Day 22)
6. Visits your pricing page directly (Day 23)
7. Clicks a retargeting ad on Facebook (Day 25)
8. Books a demo through your website (Day 26)
9. Has three sales calls over two weeks (Days 27-40)
10. Signs a $36,000 annual contract (Day 45)

Now. Which touchpoint gets credit for that $36,000?

If you're using last-touch attribution (which most platforms default to), the Facebook retargeting ad gets all the glory. But that ad only worked because of the eight touchpoints that came before it. Kill any of those earlier interactions and the retargeting ad probably never converts.

If you're using first-touch attribution, the LinkedIn post gets the credit. But a LinkedIn post without everything that followed wouldn't have closed anything.

Neither answer is right. Neither is wrong. They're just incomplete.

And this is a relatively simple journey. In reality, buying committees involve 6-10 people, journeys span months, and touchpoints number in the dozens. Attribution models are trying to simplify something that is fundamentally messy. Understanding that is step one.

The Attribution Models Explained (Without the Jargon)

Let me break down each major model, explain when it's useful, and when it'll lead you astray.

Last-Touch Attribution

What it does: 100% of credit goes to the last interaction before conversion.

When it's useful: When you need simplicity and your sales cycle is short. E-commerce with impulse purchases. Direct-response campaigns where the conversion happens in a single session.

When it misleads: Basically everywhere else. It dramatically overvalues bottom-of-funnel channels (retargeting, branded search, email) and undervalues everything that creates awareness and consideration. If you make budget decisions based purely on last-touch, you'll eventually starve the top of your funnel and wonder why your pipeline dried up.

My honest take: Every analytics platform gives you last-touch by default, so you're probably already using it. It's fine as one data point. It's dangerous as your only data point.

First-Touch Attribution

What it does: 100% of credit goes to the first recorded interaction.

When it's useful: When you're specifically trying to understand which channels are best at generating initial awareness. Good for evaluating top-of-funnel campaigns and content marketing effectiveness.

When it misleads: It ignores everything that happens after initial awareness. A channel might be great at attracting visitors who never convert, while getting all the credit just because it was first.

My honest take: I actually like first-touch more than most marketers. It provides a counterbalance to the last-touch bias that dominates most reporting. Use both, compare, and the truth is somewhere between them.

Linear Attribution

What it does: Equal credit distributed across every touchpoint. If there were 8 touchpoints, each gets 12.5%.

When it's useful: When you genuinely believe every touchpoint contributes equally. Which... rarely. But it's a decent starting point for understanding the full customer journey.

When it misleads: It's lazy, honestly. It assumes your welcome email is as important as the product demo request. That's rarely true. But it's less wrong than single-touch models for complex journeys.

Time-Decay Attribution

What it does: More credit goes to touchpoints closer to the conversion. Recent interactions get more weight than earlier ones.

When it's useful: For longer sales cycles where recent interactions genuinely matter more. B2B with 30-90 day cycles. Considered purchases where the final research phase is critical.

When it misleads: It can undervalue brand awareness and top-of-funnel content. The LinkedIn post that started the entire journey might get 2% credit while the retargeting ad gets 35%.

My honest take: This is probably the best "set it and forget it" model for most B2B companies. It's imperfect, but its biases are less harmful than single-touch models.

Position-Based (U-Shaped) Attribution

What it does: 40% credit to the first touch, 40% to the last touch, and the remaining 20% distributed among middle interactions.

When it's useful: When you believe both introduction and conversion touchpoints are critical, and middle-funnel nurture is supporting cast. This is my favorite model for most businesses with moderate sales cycles.

When it misleads: The 40/40/20 split is arbitrary. Why not 30/30/40? Or 50/20/30? There's no science behind the specific numbers.

My honest take: Despite the arbitrary numbers, this model reflects a truth about marketing: the first impression matters, the closing interaction matters, and the middle stuff helps but isn't as decisive. For companies doing $1-10M in annual marketing spend, this is usually the sweet spot.

Data-Driven (Algorithmic) Attribution

What it does: Uses machine learning to analyze your actual conversion data and assign credit based on patterns. Different touchpoints get different weights based on their statistical impact on conversion probability.

When it's useful: When you have high volumes of conversions (500+ per month) and enough touchpoint data for the algorithm to identify meaningful patterns.

When it misleads: With low conversion volumes, the algorithm doesn't have enough data and its assignments are essentially random. It's also a black box — you can't easily explain to your board why the algorithm decided podcast ads deserve 27% credit.

My honest take: It's the future, but most companies aren't ready for it yet. You need scale, clean data, and a unified tracking infrastructure. Google Analytics 4 offers this, but their model still has significant limitations.

The Practical Framework: What I Actually Recommend

Look, I could write a dissertation on attribution theory. But you need something you can implement. So here's my practical framework.

Step 1: Accept Imperfection

No model will give you "the truth." Every model is a lens that reveals some truths while distorting others. Your goal isn't perfect attribution — it's better decision-making.

Step 2: Use Multiple Models Simultaneously

Pull reports using at least three models: first-touch, last-touch, and either time-decay or position-based. Look at all three when making budget decisions. If all three models agree that a channel is performing well, you can be confident. If they disagree sharply, dig deeper.

Step 3: Layer in Incrementality Testing

Attribution models tell you what correlated with conversion. Incrementality testing tells you what caused it. Run holdout tests: take 10% of your audience in a geography and stop showing them Facebook ads for a month. Did sales drop proportionally? That's your actual Facebook impact.

I've seen this test reveal that branded search was getting credit for sales that would've happened anyway. We cut branded search spend by 40% for one client and revenue didn't change at all. That was $12,000/month going straight to waste.

Step 4: Add Post-Purchase Surveys

After someone converts, ask them: "How did you first hear about us?" It's low-tech and there's recall bias, but it captures touchpoints that digital tracking misses entirely — word of mouth, podcasts, events, billboards.

One of our clients discovered through surveys that 30% of their customers first heard about them from a podcast appearance that showed up nowhere in their analytics. They doubled down on podcast guesting and it became their most efficient acquisition channel.

Step 5: Quarterly Marketing Mix Reviews

Every quarter, sit down with all your data — platform reports, incrementality tests, survey data, revenue numbers — and make allocation decisions. Don't do this weekly. The data is too noisy at short intervals. Quarterly gives you enough signal to make smart moves.

Common Attribution Mistakes That Burn Budget

Mistake 1: Optimizing for Clicks

Clicks tell you almost nothing about value. I've seen campaigns with a 0.3% click-through rate generate 5x ROAS while campaigns with a 2.5% CTR generated nothing. Clicks measure curiosity. Conversions measure intent. Optimize for the latter.

Mistake 2: Ignoring View-Through Conversions (Or Over-Counting Them)

A view-through conversion is when someone sees your ad, doesn't click, but later converts. Should this count? Partly. Display and video ads have genuine brand impact even without clicks. But counting a 30-day view-through window inflates display's apparent performance massively. I recommend 1-day view-through windows for most analysis.

Mistake 3: Letting Platforms Self-Report

Facebook will tell you Facebook is amazing. Google will tell you Google is amazing. Both are correct and both are lying. Platform-reported conversions always exceed reality because each platform claims credit for overlapping conversions.

Use a third-party source of truth — your CRM, your Shopify dashboard, your actual revenue numbers — and reconcile platform claims against it. In my experience, the combined "conversions" reported by all platforms together exceed actual conversions by 30-60%.

Mistake 4: Treating Attribution as a Finance Function

Attribution isn't accounting. You're not trying to balance a ledger where every dollar of revenue maps to a specific ad. You're trying to understand directionally which investments drive growth. Treat it as strategic intelligence, not bookkeeping.

Mistake 5: Set-and-Forget Attribution

Your attribution model should evolve as your business changes. Launching a new channel? Your model needs to account for it. Changed your product offering? Customer journeys shift. Had a big PR moment? First-touch data is going to look completely different for the next few months.

The Privacy Elephant in the Room

We can't talk about attribution without acknowledging that the tracking landscape is fundamentally changing. iOS privacy changes nuked a huge portion of user-level tracking. Third-party cookies are disappearing. Consent regulations are getting stricter globally.

This doesn't mean attribution is dead. It means it's evolving.

What's replacing granular user tracking:

• Marketing Mix Modeling (MMM): Statistical analysis of aggregate data — total spend per channel correlated with total revenue over time. No user-level tracking needed. It's old-school but it's making a comeback for good reason.
• Incrementality testing: Geographic or audience-based holdout experiments that measure true causal impact.
• First-party data strategies: Building direct relationships with customers through email, accounts, and loyalty programs so you have consented, owned data.
• Aggregated measurement APIs: Platform-provided tools like Meta's Aggregated Event Measurement that give directional insights without individual tracking.

The companies that will thrive are the ones building measurement frameworks that don't depend on tracking every click. Start diversifying your measurement approach now, because the user-level tracking era is ending whether we like it or not.

Building Your Attribution Stack

For most companies spending $20,000+ monthly on marketing, here's what I'd recommend:

1. Google Analytics 4 as your central analytics platform with data-driven attribution enabled
2. A CRM (HubSpot, Salesforce) tracking the full customer journey from first touch to closed deal
3. UTM discipline — every link, every campaign, every channel tagged consistently
4. Post-purchase surveys capturing self-reported attribution
5. Quarterly incrementality tests on your biggest spend channels
6. A dashboard that shows multiple attribution models side by side

Total additional cost for most of this: zero to a few hundred dollars a month. The technology isn't the barrier. The discipline is.

The Bottom Line

Attribution isn't about finding "the answer." It's about having enough insight to make better decisions than your competitors. If you can figure out that your podcast sponsorship drives 3x more qualified leads than display ads, while your competitor is still guessing, you win.

Don't let the complexity paralyze you. Start with what you have, layer in better measurement over time, and always — always — question what the dashboards are telling you.

Because the dashboard is a model. And as the statistician George Box said, all models are wrong. But some are useful.

Make yours useful.