What Is Marketing Attribution? Models, Limitations, and How to Choose the Right One
Marketing attribution assigns credit for conversions to the touchpoints that contributed to them - but every model makes different assumptions. Here is the full breakdown of six models, when to use each, and why the sum of platform ROAS always exceeds actual revenue.
What Is Marketing Attribution? Models, Limitations, and How to Choose the Right One
Marketing attribution is the practice of assigning credit for conversions to the marketing touchpoints that contributed to them. It answers "which channels and campaigns caused this sale?" - but the answer always depends on which attribution model you use, and every model makes a different assumption about what "caused" means.
Marketing attribution: the process of assigning credit for conversions across the touchpoints in a buyer's journey, using a defined model that determines how credit is distributed between channels and interactions.
Attribution is not a single number. It is a lens. The same conversion can be attributed to paid search (last-touch model), social media (first-touch model), or distributed across five interactions (linear model) - and all three answers come from the same data. Understanding which model fits your question is the starting point for using attribution correctly.
Key takeaways
Attribution Is a Lens Not a Single Number
The same conversion can be attributed to paid search, social media, or distributed across five interactions depending on the model used - and all three answers come from the same data. Which model fits the decision being made determines which answer is correct.
Without Attribution Every Platform Claims One Hundred Percent Credit for the Same Conversion
Summing per-platform reported conversions produces a total higher than actual conversions. Summing per-platform attributed revenue produces a total higher than actual revenue. This double-counting is not a reporting error - it is the structural consequence of each platform applying its own attribution model.
The Six Primary Attribution Models Differ Only in How Credit Is Distributed
Last-click gives 100% to the final touch, first-click gives 100% to the first touch, linear distributes equally, time-decay weights toward recent touches, position-based assigns 40% to first and 40% to last, and data-driven uses ML weighting. The choice determines which channels appear to be performing well.
Every Attribution Model Is an Approximation of True Causal Contribution
True causal contribution of each touchpoint is not directly measurable from click data alone - which is why incrementality testing with holdout groups is the only methodology that directly measures whether a channel caused conversions rather than merely correlated with them.
The Correct Attribution Model Depends on the Decision Being Made
Last-click is most useful for identifying which channels are closest to conversion. First-click works for discovering awareness channels. Time-decay fits sales-cycle-aware budget allocation. Using one model for all decisions is the most common attribution error because it produces correct answers for one question while being misleading for the others.
Why attribution is necessary - and why it's always an approximation
Every marketing team faces the same structural problem: money goes into multiple channels simultaneously, and the same customer sees a Meta ad on Monday, clicks an email on Wednesday, searches on Google on Thursday, and converts through a direct visit on Friday. Which channel gets credit?
Without attribution, teams default to whatever platform dashboard they check - which always claims full credit for the conversion. Meta reports one ROAS. Google reports another. The sum routinely exceeds actual revenue by 200-300%.
Attribution creates a single system of record that accounts for the full conversion path rather than letting each platform self-report. But no attribution model can objectively determine which touchpoint "caused" the sale - causation in marketing is unobservable. Every model is an approximation, and different approximations produce different answers.
The practical implication: choose the attribution model that fits your decision context, not the one that maximises ROAS for your preferred channel.
Of the six models, last-click is the most common default and the most damaging when used for cross-channel budget decisions. Switching from last-click to multi-touch attribution typically reveals 30-60% of B2B SaaS marketing spend was misallocated. For the specific failure modes and the migration path, see why last-click attribution is broken.
The six attribution models and when to use each
First-touch attribution gives 100% credit to the first touchpoint in the conversion path. Useful for measuring which channels generate awareness and drive new prospects into the funnel. Problem: ignores everything that happened between awareness and conversion, making it useless for evaluating mid-funnel or conversion channels.
Last-touch attribution gives 100% credit to the final touchpoint before conversion. The default for most analytics platforms - 35% of B2B SaaS organisations still use it as their primary model. Useful for measuring direct response and conversion channel efficiency. Problem: systematically undercredits awareness and mid-funnel channels that built the intent last-touch then captures.
Linear attribution distributes equal credit across every touchpoint in the path. Useful for getting an approximate full-funnel view when you lack data to justify more complex models. Problem: treats a 30-second retargeting impression the same as a 10-minute demo watched on YouTube.
Time-decay attribution weights recent touchpoints more heavily, with exponentially decreasing credit for earlier interactions. Useful for B2B with long sales cycles where the most recent interactions are more influential. Problem: undersells top-of-funnel investment in organic content and social that built awareness months before conversion.
Position-based (U-shaped) attribution gives 40% to first touch, 40% to last touch, and distributes 20% across middle touchpoints. A common compromise for teams that want to value both acquisition and conversion channels without ignoring mid-funnel activity.
Data-driven attribution uses statistical modelling to assign credit based on each touchpoint's measured impact on conversion probability - comparing paths that converted versus those that didn't. Requires significant data volume to produce reliable outputs. Google Ads and Meta both offer proprietary versions, but their weighting methodologies are not publicly documented.
Put what you just read into one place
Prooflytics unifies every source into one brief — and remembers what worked.
14 days free · no credit card
What the data shows about attribution model adoption
Gartner's 2025 UK Digital Marketing Survey found that only 24% of B2B organisations currently use multi-touch attribution. In B2B SaaS specifically, 67% of marketing teams still rely on last-touch as their primary model - crediting the final touchpoint while ignoring every prior interaction.
The consequence is systematic budget misallocation: teams over-invest in bottom-of-funnel channels (branded search, retargeting) and under-invest in top-of-funnel channels (content, social, awareness) that last-touch makes invisible. Over time, top-of-funnel pipeline dries up while the team reports improving ROAS on channels that convert existing demand rather than generate new demand.
Prooflytics addresses this with a transparent rules-based attribution layer that reconciles platform-reported data against actual outcomes - surfacing the discrepancy between platform-claimed ROAS and confirmed revenue, rather than collapsing it into one model's output.
The ROAS overlap problem: why platform numbers always exceed actual revenue
Meta, Google, and TikTok each use their own attribution windows. Meta defaults to 7-day click / 1-day view. Google Ads defaults to a 30-day click window. When a customer sees a Meta ad on Tuesday and a Google ad on Thursday before converting on Friday, both platforms claim the conversion in full.
For most marketing teams running two or more paid channels simultaneously, the sum of platform-reported conversions exceeds actual orders by 30-70%. For teams running Meta, Google, TikTok, and email simultaneously, the overlap can reach 100% - meaning total platform-claimed conversions equal or exceed actual revenue.
The only defensible performance number is actual revenue from your commerce system or CRM, divided by actual spend. Platform ROAS is a useful comparative signal within one platform - for creative testing, bid strategy evaluation, and audience comparison - but it cannot be summed across platforms and presented as total ROAS without producing a number that is structurally inflated.
Attribution vs. measurement vs. marketing intelligence
Attribution answers: which channels got credit for this conversion?
Marketing measurement answers: what was the actual revenue impact of this marketing activity?
Marketing intelligence answers: why did performance change, and what should we do next?
Attribution is one input into measurement, which is one input into marketing intelligence. Teams that stop at attribution - focusing on who gets the credit - miss the more operationally valuable questions: why did overall CPL shift, what competitor activity correlates with the movement, and which actions produce the best expected outcome.
For the framework on how attribution connects to a full marketing intelligence stack, see the marketing intelligence guide.
How to choose the right attribution model
Three questions determine the right model for your context:
1. What decision are you trying to inform? Budget allocation across channels to use multi-touch. Conversion channel efficiency to last-touch is informative as one signal. Awareness channel ROI to first-touch or data-driven.
2. How long is your average sales cycle? Under 2 weeks: last-touch is defensible. 2-8 weeks: position-based or linear. Over 8 weeks: time-decay or data-driven.
3. How much conversion data do you have? Under 1,000 conversions per month: avoid data-driven attribution - the model lacks sufficient data and produces noise. Use position-based or linear instead.
No model is universally correct. The right model is the one that most accurately represents how your specific buyer journey works - and is consistent enough over time to support trend analysis.
Bottom line
- Marketing attribution assigns credit for conversions to contributing touchpoints - but every model makes different assumptions about what "caused" the conversion
- 67% of B2B teams still use last-touch, which systematically undercredits awareness channels and concentrates budget at the bottom of the funnel
- The sum of platform-reported ROAS across channels always exceeds actual revenue due to attribution window overlap - the only defensible performance number is actual revenue divided by actual spend
- The right model depends on your decision context, sales cycle length, and conversion data volume - no single model is universally correct
- Attribution is one input into marketing measurement, which is one input into marketing intelligence - the explanation and action layers above attribution are where operational value is created
- Read independent Prooflytics reviews on G2 and compare attribution approaches across marketing intelligence platforms
Frequently asked questions
What is the difference between marketing attribution and marketing analytics?+
Marketing analytics measures performance across channels - what happened, at what cost, with what volume. Marketing attribution is one component of analytics: it determines which channels get credit for conversions. Analytics without attribution relies on platform self-reporting, which inflates each channel's contribution. Attribution creates a shared denominator.
Why don't Google Ads and Meta Ads attribution numbers match?+
Google and Meta use different attribution windows and models. Meta defaults to 7-day click / 1-day view. Google Ads defaults to a 30-day click window. When a customer interacts with both platforms before converting, both claim the conversion. The overlap is why summing platform-reported ROAS across channels produces a number that exceeds actual revenue.
What is data-driven attribution and when should I use it?+
Data-driven attribution uses statistical modelling to assign credit based on each touchpoint's measured impact on conversion probability. It is theoretically more accurate than rules-based models, but requires sufficient data volume - typically 3,000+ conversions per month - to produce reliable outputs. For teams below this threshold, a transparent rules-based model (linear or position-based) is more reliable.
Can I use different attribution models for different decisions?+
Yes. Many teams use last-touch for conversion reporting, first-touch for awareness channel evaluation, and a multi-touch model for budget allocation decisions. The key is to be explicit about which model produces which number - and never present different-model outputs in the same report as if they are comparable.
What is the difference between attribution and media mix modelling?+
Attribution assigns credit to specific digital touchpoints in a conversion path - the Meta ad, the email, the Google search. Media mix modelling (MMM) is a statistical approach that estimates the revenue contribution of all marketing activities - including offline channels (TV, radio, direct mail) that produce no trackable clicks - using regression analysis on aggregate spend and revenue data. Attribution and MMM answer different questions and require different data inputs.
Put what you just read into one place
Prooflytics unifies every source into one brief — and remembers what worked.
14 days free · no credit card
Continue reading
Multi-Touch Attribution: Models, Why Last-Click Fails, and How to Implement It
Last-click attribution misallocates budget by ignoring every touchpoint except the final one. This guide covers the four multi-touch attribution models - linear, time-decay, position-based, and data-driven - and how to implement them for B2B SaaS teams.
What Is Incrementality Testing in Marketing? A Practical Guide for In-House Teams
Attribution tells you who got credit. Incrementality testing tells you what actually caused the conversion. Here's how in-house marketing teams measure true ad impact without a data science department.
Marketing Analytics Guide: From Scattered Data to a Unified View
Marketing analytics means joining data from all your channels - paid, email, CRM, and revenue - into a single view. This guide covers what it includes, where most stacks break down, and how to build one that gives reliable answers.
End-to-End Marketing Analytics: Connecting Ad Spend to Revenue
End-to-end marketing analytics closes the loop between your paid channels and actual revenue events in your CRM or payment processor. This guide explains the full-funnel stack, why most B2B SaaS teams are still stuck at the reporting layer, and how to move to the intelligence layer where decisions happen.