Measuring Marketing ROI When the Buyer Journey Is Messy
By Rick Elmore ·
Every founder I talk to wants a single number that proves their marketing works. The buyer journey rarely cooperates.
Marketing ROI measures the revenue generated by marketing relative to what you spent to generate it. When the buyer journey is messy — multiple touchpoints, long sales cycles, dark social — you stop chasing perfect attribution and instead build a layered measurement system that ties spend to pipeline and revenue with reasonable confidence.
Why marketing ROI is hard to measure now
The clean funnel died a while ago. A buyer might hear you on a podcast, ignore you for three months, get retargeted, ask a peer in a Slack community, read two of your posts, then type your brand name into Google and "convert" as organic search. Your analytics tool credits the last click. Every touch before it disappears.
This is the core problem. The touchpoints that actually move buyers are often the ones your tools can't see. Dark social, word of mouth, and offline conversations don't leave clean UTM trails. Meanwhile the trackable channels get over-credited simply because they happen to be last in line.
Sales cycles make it worse. If it takes 90 days to close a B2B deal, the spend that created the opportunity happened last quarter. Any ROI calculation that compares this month's spend to this month's revenue is measuring two unrelated things.
So the goal isn't a perfect attribution model. It's a measurement system honest about its own uncertainty, that still points you toward better decisions. That's a RevOps problem, not a marketing-dashboard problem.
What marketing ROI actually means
The textbook formula is simple: (revenue attributed to marketing − marketing cost) ÷ marketing cost. The trouble is every word in that sentence is doing heavy lifting.
"Revenue attributed to marketing" depends on your attribution model. "Marketing cost" should include salaries, tools, and agency fees, not just ad spend — otherwise you're flattering yourself. And "attributed" implies a confidence level you probably don't have.
I tell teams to hold two versions of ROI at once. There's the defensible number, built only on touches you can actually track, which understates reality. And there's the directional number, which layers in self-reported and modeled data to get closer to the truth but carries more assumptions. Report both. Decisions made on the defensible number are conservative; decisions made on the directional number are informed. Pretending you have one precise figure is where teams lose credibility with their board.
How to measure marketing ROI when the journey is messy
Here's the framework we deploy inside client revenue engines. It moves from crude and reliable to nuanced and directional, so every layer checks the one below it.
1. Start with the blunt instrument: blended ROI
Before you touch attribution models, calculate total revenue divided by total marketing spend over a trailing period long enough to cover your sales cycle. If your cycle is 90 days, look at rolling quarters. This blended number ignores which channel did what, and that's the point. It's hard to game and it captures the effect of untrackable touches, because they still show up in closed revenue somewhere.
Blended ROI is your ground truth. If your fancy attribution says one thing but blended ROI is flat, trust the blended number.
2. Add self-reported attribution
Put one open question on every demo or intake form: "How did you first hear about us?" It's low-tech and it catches exactly the touches your tracking misses. Nobody's pixel records a podcast mention, but buyers will tell you. Over a few hundred responses, patterns emerge that no analytics platform will surface.
Treat this as a signal, not gospel. Buyers misremember. But when self-reported data and blended trends agree, you're onto something real.
3. Track pipeline, not just closed revenue
Because deals close slowly, waiting for revenue means you're always driving by the rearview mirror. Measure marketing-sourced and marketing-influenced pipeline as a leading indicator. Sourced means marketing created the opportunity. Influenced means marketing touched a deal sales already had. Both matter, and confusing them is how marketing and sales end up arguing over the same closed-won.
4. Use cohorts to respect the sales cycle
Group leads by the month they entered, then follow that cohort forward as it converts. This is the only honest way to connect a specific month's spend to the revenue it eventually produced. It takes patience, but it kills the apples-to-oranges comparison that wrecks most ROI reporting.
5. Run holdout tests where you can
When attribution is genuinely murky, incrementality testing cuts through it. Turn off a channel in one region or audience and watch what happens to pipeline. If nothing changes, that channel wasn't driving what you thought. Geo holdouts and pausing tests are more truthful than any model, because they measure cause instead of correlation.
Attribution models compared
No single attribution model is correct. Each one answers a different question and carries a different bias. The mistake is picking one and treating it as reality. Here's how the common approaches stack up for a messy B2B journey.
| Model | How it assigns credit | Best for | Main weakness |
|---|---|---|---|
| First-touch | 100% to the first interaction | Understanding what drives awareness and top-of-funnel | Ignores everything that closes the deal |
| Last-touch | 100% to the final interaction before conversion | Simple reporting, short cycles | Over-credits branded search and retargeting |
| Linear multi-touch | Equal credit across every touchpoint | Long cycles with many known touches | Treats a trivial touch like a decisive one |
| Self-reported | Credit to what the buyer says influenced them | Capturing dark social and offline touches | Memory bias, low response quality |
| Blended (spend vs. revenue) | No per-channel credit; total in vs. total out | Ground-truth sanity check | Can't tell you which channel to cut |
| Incrementality testing | Credit based on measured lift vs. holdout | Proving true causal impact | Requires scale, time, and discipline |
In practice we run blended ROI and self-reported as the backbone, use a multi-touch view for channel-level tuning, and reach for incrementality tests when a channel's budget gets big enough to justify the effort. The models triangulate. Where they agree, you act with confidence. Where they disagree, you investigate.
The metrics that actually matter
A pile of dashboards is not a measurement system. Most marketing dashboards track activity — impressions, clicks, sessions — none of which pay salaries. Strip it back to the metrics that connect spend to money.
- Blended CAC: total sales and marketing spend divided by new customers. The most honest efficiency number you have.
- CAC payback period: how many months of gross margin it takes to recover the cost of acquiring a customer. This is where cash flow lives or dies.
- Pipeline coverage: qualified pipeline relative to your revenue target. Tells you whether marketing is feeding the machine enough.
- Marketing-sourced revenue %: the share of closed revenue that started with marketing. Tracks the function's real contribution over time.
- LTV:CAC ratio: the long-term test of whether acquisition economics work at all.
Notice what's missing: cost per lead in isolation, MQL counts, engagement rates. Those are diagnostic metrics — useful when a headline number moves and you need to know why, useless as goals in themselves. Optimize for cheap leads and you'll get plenty of cheap leads that never buy.
The connective tissue here is clean data. If your CRM doesn't reliably tie spend to opportunities to closed revenue, none of these numbers can be trusted. This is exactly where RevOps earns its keep, and it's the foundation we build first in every engagement. You can see how that fits into a full system in our packages.
Putting it together as a system
Measuring marketing ROI in a messy journey isn't about finding the one true model. It's about layering imperfect signals until the picture is clear enough to act on, and being honest about the confidence at each layer.
The operator's version looks like this. Blended ROI is your north star and reality check. Self-reported attribution catches the dark funnel. Cohort analysis respects your sales cycle. Pipeline metrics give you leading indicators so you're not always looking backward. Incrementality tests settle the arguments that models can't. And a handful of economic metrics — CAC, payback, LTV:CAC — tell you whether the whole thing is worth doing.
When those signals point the same direction, move budget with confidence. When they conflict, that conflict is information — usually pointing at a data problem or a channel that looks better than it performs. Either way you learn something. That's a far better position than a single dashboard number nobody quite believes.
Frequently asked questions
What is a good marketing ROI for B2B?
It depends entirely on your margins and sales cycle, so be wary of universal benchmarks. A more useful lens is CAC payback: many B2B teams aim to recover acquisition cost within 12 months and target an LTV:CAC ratio around 3:1 or better. Measure your own trend over time rather than chasing someone else's number.
How do you attribute revenue when buyers don't tell you where they came from?
You combine methods. Add a "how did you hear about us" question to catch what buyers will volunteer, lean on blended ROI to capture the effect of invisible touches in aggregate, and use holdout tests to measure a channel's true lift. No single tool sees everything, so you triangulate across several imperfect ones.
Should I use first-touch or last-touch attribution?
Neither on its own. First-touch tells you what creates awareness; last-touch tells you what closes. For a long B2B journey, use both as diagnostic views and anchor your actual decisions to blended ROI and pipeline contribution, which don't depend on a single touchpoint being right.
How long should I wait before judging a marketing channel's ROI?
At least one full sales cycle, and ideally two, before drawing conclusions. If deals take 90 days to close, judging a channel after 30 days measures noise. Use pipeline created as your early read, then confirm with closed revenue once the cohort has had time to convert.
If your ROI numbers don't add up or your tools disagree with each other, the fix is usually in the data plumbing, not the marketing. Book a Revenue Systems Audit and we'll show you exactly where your attribution is leaking.