Every conversation about attribution ends up being about tools. Which platform to use, which model to apply, whether to buy the expensive attribution software or build something in-house.
The tool is not the problem.
The problem is structural. And it starts before anyone opens a dashboard.
What Attribution Is Actually Supposed to Do
Attribution is supposed to answer one question: which marketing activity influenced revenue?
Not which activity generated traffic. Not which activity generated leads. Which activity produced customers who paid you money.
Most companies are not measuring that. They are measuring a proxy and calling it attribution. Click-through rates, cost per lead, form submissions. These are useful signals. They are not the same as revenue attribution.
The reason this matters: two campaigns can produce leads at identical cost and identical volume. One produces customers who close quickly and stay for two years. The other produces leads that sit in the pipeline for months and churn after 90 days. If you are measuring cost per lead, those two campaigns look identical. If you are measuring revenue, they are completely different.
The Three Structural Breaks That Cause Attribution Failures
Break 1: The UTM problem
UTM parameters are how you tag traffic so you know where it came from. If your UTM strategy is inconsistent, your attribution data is garbage before anyone ever looks at it.
The most common failure: some campaigns have UTMs, some do not. Some team members use utm_source=linkedin, some use utm_source=LinkedIn, some use utm_source=LI. Those three will show up as three separate sources in your reporting. You will undercount LinkedIn performance by two-thirds and not know why.
UTM governance is not glamorous. It is the foundation everything else depends on.
Break 2: The CRM handoff problem
The lead comes in. The UTM data is captured correctly. Then a sales rep manually edits the lead source field to say “sales prospecting” because that is how they record their outbound. The original attribution is gone.
Or the CRM field that captures lead source is not a required field, so half the time it is left blank and nobody notices until someone tries to pull an attribution report three months later.
I have seen companies with excellent tracking on the marketing side and completely clean data in the ad platforms, where 40% of their CRM records have no lead source because the handoff process was never defined. Everything upstream was working. The data died at the handoff.
Break 3: The time horizon problem
Most attribution reporting looks at a 30-day window. If your sales cycle is 90 days, your attribution window is wrong.
A lead that came in from a LinkedIn campaign in January, went through a six-week evaluation process, and signed a contract in March will show up in your 30-day reporting as having no revenue attached to it. The campaign that generated it looks like it produced zero revenue. You cut the budget. The next cohort of leads dries up in April.
Attribution windows need to match buying cycles. This sounds obvious. It almost never gets implemented.
What Good Attribution Actually Looks Like
I spent two years managing a $12M+ annual paid media budget for healthcare programs at a large EdTech company. The buying cycle for a prospective student was three to six months from first awareness to enrollment. We were driving paid traffic from search, social, display, and programmatic simultaneously.
For attribution to mean anything in that environment, we needed to track every touchpoint across that entire window, connect it to enrollment data in Salesforce, and build reporting that followed the cohort forward in time rather than backward from a conversion event.
That work required Salesforce, Marketo, Adobe Analytics, and Google Tag Manager working in a coordinated data architecture. Not because the tools were complicated. Because the structural requirements were clear and each tool had a defined role.
The result was a reporting model where we could see, for any given week’s paid investment, what the enrollment and revenue outcome was likely to be 90 days later. Budget decisions were made on that projection, not on last-week’s CPL.
That is what attribution is supposed to do.
Where to Start if Your Attribution Is Broken
You do not need a new tool. You need a definition and an audit.
Start with the definition: what is a lead in your business? What makes an opportunity qualified? What does a closed-won deal look like in your CRM? Write those definitions down. Make sure marketing and sales agree on them. You would be surprised how often they do not.
Then do an audit of your current data. Pull 90 days of closed-won deals. For each one, can you identify the original lead source? Can you trace it back to a specific campaign? If you cannot do that for more than half your deals, your attribution infrastructure is not giving you usable information.
Fix the definition first. Fix the tracking second. Then, and only then, worry about which attribution model to apply.
The Revenue Clarity System I use with clients starts every engagement with a Signal Audit for exactly this reason. You cannot build an acquisition strategy on broken data. You need to know what the data is actually saying before you make any decisions about where to spend money.