Marketing attribution is how you decide which channels, campaigns, and touches actually created revenue, so you can move budget toward what works and stop funding what does not.
Most teams do not have an attribution problem on the surface. They have dashboards from Meta, Google, their CRM, and an email tool. The problem is that each system claims the same conversion in its own way, so leadership ends up with four versions of the truth and no clear answer to a simple question: if we add the next unit of spend, where should it go?
This guide explains the main attribution models in plain language, shows when each one fits, and lays out how to build a measurement approach that survives long sales cycles and offline conversations. It is the hub for the related pieces on attribution, multi-touch tracking, and revenue reporting.
What attribution is really trying to answer
Attribution is credit assignment. A buyer rarely sees one ad and converts. They might read a LinkedIn post, click a Google search ad two weeks later, get a nurture email, then book a call from a retargeting ad. Attribution decides how much of that closed deal each step earned.
The reason this matters is budget. If the search ad gets all the credit, you over-invest in search and starve the LinkedIn content that started the journey. Good attribution is less about perfect math and more about making a defensible budget decision with the data you can trust.
The main attribution models, compared
There is no single correct model. Each one tells a different story about the same buyer journey. Here is how the common models treat credit, and where each one is useful.
| Model | How credit is assigned | Best for | Main weakness |
|---|---|---|---|
| First-touch | 100% to the first interaction | Measuring what creates awareness and new demand | Ignores everything that closes the deal |
| Last-touch | 100% to the final interaction before conversion | Short cycles and direct-response offers | Over-credits bottom-funnel and brand search |
| Linear | Even split across every touch | A fair baseline when you have clean touch data | Treats a minor visit the same as a sales call |
| Time-decay | More credit to touches closer to the sale | Longer B2B cycles where recency signals intent | Under-values the top-of-funnel that started it |
| Position-based (U-shaped) | 40% first, 40% last, 20% to the middle | Teams that value both demand creation and closing | The 40/40/20 split is a convention, not a measurement |
| Data-driven | Modeled weights from your own conversion patterns | High-volume accounts with enough data to model | Needs volume and clean data, and is harder to audit |
A practical rule: start with a position-based model if you run real multi-channel programs, and keep a last-touch view alongside it as a sanity check. The gap between the two views is often where the interesting budget questions live.
Why most attribution breaks in practice
The model is rarely the hard part. The data plumbing is. Three failures show up again and again.
- Broken joins. A lead fills a form as one email, books a call with a personal Gmail, and signs as a company domain. If those records never link, the journey shatters into three half-stories.
- Offline conversions. Calls, demos, and sales notes happen outside the ad platforms. If the closed-won event never flows back to Meta or Google, the platforms optimize toward cheap leads instead of real pipeline. The piece on why Meta ads attribution broke in 2026 covers this in detail.
- Platform self-reporting. Each ad network is graded on its own homework and will claim conversions generously. Adding up platform-reported numbers usually overcounts your real results by a wide margin.
Fixing these is mostly about a single source of truth. The CRM, not the ad platform, should hold the record of what became pipeline and what became revenue.
Multi-touch attribution for long B2B cycles
When a deal takes months and involves several people, single-touch models stop being useful. Multi-touch attribution keeps the whole sequence of interactions and assigns fractional credit across them. The benefit is that you can finally see how an early webinar or a piece of content contributes to deals that close much later.
The cost is discipline. Multi-touch only works if every meaningful touch is captured and tied to the same person and account. That means UTM hygiene on every link, server-side conversion tracking where possible, and a CRM that records the source of each contact and each opportunity. Without that foundation, a fancy model just produces confident nonsense.
From attribution to a budget decision
Attribution is only worth the effort if it changes what you do next. The output that matters is not a pie chart, it is a reallocation. A working loop looks like this:
- Define the one conversion that represents real value, usually a sales-ready opportunity or closed revenue, not a raw lead.
- Tie every channel and campaign to that conversion through the CRM, not through platform reports.
- Review the gap between first-touch and last-touch views to see which channels open deals and which close them.
- Move a small, fixed share of budget each cycle toward the channels with the strongest contribution to qualified pipeline.
- Re-check after a full sales cycle, because moving budget changes the mix you are measuring.
This is where attribution connects to the rest of the system. The tools you choose to wire this together matter, which is why it pairs with the guide to multi-channel campaign analytics tools, and the dashboard view in building a marketing ROI dashboard.
Where attribution meets cost and scoring
Attribution answers which channels create revenue. Two neighbors answer the next questions. Cost per acquisition, covered in true CAC across channels, tells you whether the revenue is profitable. Lead scoring, covered in predictive lead scoring, tells you which of the leads attribution surfaces are worth your team's time first. Read together, the three give leadership a clear picture of where the system creates value and where it leaks.
If you want help connecting ad platforms, your CRM, and sales feedback into one measurement layer, our multi-channel lead generation work is built around exactly this problem.
A worked example: one buyer, six models
Picture a single closed deal with four touches in order: a LinkedIn post, a Google search ad, a nurture email, and a retargeting ad that drove the booking. Here is how the models split the credit for that one deal.
| Touch | First-touch | Last-touch | Linear | Position-based |
|---|---|---|---|---|
| LinkedIn post | All | None | Quarter | 40% |
| Search ad | None | None | Quarter | 10% |
| Nurture email | None | None | Quarter | 10% |
| Retargeting ad | None | All | Quarter | 40% |
Look at what changes. First-touch says LinkedIn deserves the whole budget and retargeting deserves none. Last-touch says the opposite. Linear calls them equal, which is rarely true. Position-based rewards the channel that opened the relationship and the one that closed it, while still giving the middle some credit. None of these is the truth, they are four lenses on the same event, and the lens you pick quietly decides where next quarter's money goes.
Pick a model by your sales cycle
The right starting model depends on how people buy from you.
- Short, direct-response cycles. If buyers convert in one or two sessions, last-touch is usually good enough and far simpler to run.
- Considered purchases over weeks. Position-based fits, because both the channel that created demand and the one that closed it deserve weight.
- Long B2B cycles with many people. Time-decay or a data-driven model fits, because recency carries real intent signal and there are too many touches to weight by hand.
Whatever you choose, write the choice down and keep it stable for a full cycle. Switching models mid-quarter makes every report look like a change in performance when nothing real moved.
Attribution is not the same as incrementality
One trap is worth naming. Attribution tells you which touches were present when a deal closed. It does not tell you which touches actually caused the deal. Brand search is the classic case: it gets last-touch credit for buyers who were already going to find you. The way to separate correlation from cause is a holdout test, where you pause a channel for a slice of your audience and watch whether conversions fall. Use attribution to guide everyday budget shifts, and use the occasional holdout test to check whether a channel is earning its credit or just standing in the doorway.
A tracking hygiene checklist
No model survives messy data, so get the basics right before you argue about weights.
- Put consistent UTM tags on every paid and outbound link, with a naming convention everyone follows.
- Use server-side conversion tracking where you can, so ad-blockers and privacy changes do not blank out your data.
- Record the source on both the contact and the opportunity in your CRM, because the lead source and the deal source are not always the same.
- Send closed-won events back to the ad platforms so they optimize toward revenue, not toward cheap clicks.
- Reconcile platform-reported conversions against CRM reality on a regular cadence, and trust the CRM when they disagree.
A simple way to start
You do not need a perfect model to get value. Pick one conversion that reflects real revenue, make the CRM the system of record, and compare a position-based view against last-touch. That single comparison usually exposes the biggest misallocation in the budget, and it is enough to make a better decision this quarter while you build toward multi-touch over time.