What Is Marketing Attribution? A SaaS Guide

Marketing attribution is the process of assigning credit for a conversion to the marketing touchpoints that contributed to it, so you can find out what really drives revenue rather than guessing. In short, marketing attribution answers the following question: “What made a customer buy?”

This guide explains what marketing attribution is, how it works, the attribution models, their pros and cons, why the default attribution is misleading, and how SaaS companies can attribute revenue. It is designed for entrepreneurs and marketers who want to learn attribution in order to make better budgeting decisions, not for memorizing the models.

The problem attribution solves

Consider a customer’s path to signing up and becoming a client. He/she reads a blog post several months ago. Then he clicks an ad. He subscribes to a newsletter. He goes to a comparison article. He tries your service and upgrades.

And now you need to report what marketing touchpoints helped you convert that customer.

This is not just a hypothetical question: your budget depends on the answer. If you attribute the conversion to the wrong touchpoint, you overspend on one channel and neglect another that actually convinces customers. Do this for many campaigns and for each customer, and attribution will silently distribute your marketing budget.

However, giving an honest answer — “All of them contributed in various ways” — doesn’t help with decision making. Therefore, you start using attribution rules that allow assigning credit to marketing actions in some way. The rules are called attribution models, and your choice determines what your data says.

Thus, attribution is an attempt to turn a complicated and multi-stage process into a convincing statement about its parts. Successful attribution tells you what marketing efforts drive revenue. Failure makes you invest in the wrong channels.

Attribution is not the same as reporting

Before going deeper, it is helpful to distinguish two related but quite different concepts.

Reporting is a set of facts about the past, for example, you received 500 sign-ups, 40 of them converted, and your revenue was X. Such a report is necessary, but it is not informative.

Attribution attempts to explain what caused what happened in a reporting. Of 40 conversions, which touchpoints contributed? This is a causal analysis, and it is quite difficult because you usually cannot conduct an experiment in which you expose a customer to one touchpoint only.

Why it matters is that some teams consider dashboards of marketing data as attribution. They are not attribution; reporting shows the results. Attribution requires you to choose a model and distribute the credit among marketing efforts.

How attribution works, at least technically

There are three components of the process.

Tracking the touchpoints. You need to know all the marketing touchpoints that preceded the conversion: what pages the customer visited, which campaigns were running, which emails he/she opened, and through which channels the customer appeared in your funnel. Without tracking the touchpoints, attribution is impossible. In SaaS, it means you have to connect your analytics and marketing tools to the product and billing system so that you could trace the customer’s journey until the conversion.

Choosing a model. This is the part that makes attribution interesting. The model is a rule according to which the credit is distributed among the touchpoints. Give all the credit to the first? To the last? Distribute it equally? Give weight to some touchpoints and not to others? This choice is the core of attribution, and it is the source of many discussions.

Connecting to revenue. You can’t just attribute the customer to the channels. The customer is important because he contributes to your revenue. You need to attribute the revenue to the marketing efforts. Attribution of conversions is worse than attribution of revenue in SaaS: some customers can become profitable within a month while others bring value after several years. Proper attribution connects the touchpoints not only to conversions but to the recurring revenue and its longevity.

In other words, track the customer journey, apply a rule to distribute credit, and tie this credit to the revenue.

The main attribution models

There is no correct attribution model; every model makes some assumptions and is appropriate in certain cases and misleading in others.

Last-click attribution (also known as last-touch attribution) gives all the credit to the final touchpoint before the conversion. It is the default attribution in most analytics tools, which is why it is widely used, and widely misleading. It is simple and clear — you give credit to what the customer did before paying. However, it distorts reality because it systematically favors whatever happens to be the last touchpoint (usually branded searches or retargeting ads) and completely ignores everything that had been done before.

First-click attribution (also known as first-touch attribution) gives all the credit to the very first touchpoint. This is the opposite of the previous model and makes the opposite mistake: it systematically favors discovery touchpoints and ignores whatever nurtures and closes the conversion.

Linear distributes credit equally among all touchpoints. Its merit is that it acknowledges the multi-touch nature of the journey. Its flaw is that it considers each touch equally important, which is not often true: one click on an uninteresting link should not be equivalent to an interesting demo call.

Time-decay gives more credit to touchpoints close to the conversion and less credit to touchpoints that happened earlier. The idea behind it is that recent contacts played a larger role in customer’s decision-making process. It may be reasonable in many cases, but it can underestimate the importance of the early content that creates the customer’s awareness and intent.

Position-based (also called U-shaped) attribution gives credit mostly to the first and last touchpoints and equally divides the remaining credit among the intermediate touchpoints. It is based on a natural assumption: finding the customer and closing the deal are very important, while nurturing in between plays a secondary role.

Multi-touch attribution is the general term for attribution models that give credit to more than one touchpoint. All attribution models above are multi-touch except the first and last models.

The straightforward conclusion is that the single-touch models (the first and the last) are misleading, and the multi-touch models are honest but require more effort. However, what is more important is the explanation in the next section.

Why last-click attribution is misleading for most teams

The fact that last-click attribution is the default makes it widely applied without much thinking. And you should understand how it misleads you.

Last-click attribution gives credit to the last step in the journey. Usually, this step consists of a low-effort action when a customer already has his intention to buy: a branded search (he types your name because he already knows about you), a retargeting ad (it appears to customers who already visited your website), and a direct visit.

However, the touchpoints that build awareness and intention receive zero credit because they happened earlier in the journey. As a result, the attribution shows “branded searches and retargeting ads are the main drivers of our revenue.” The team reallocates budget to these touchpoints and neglects the top-of-funnel efforts that create the demand. The numbers remain high until growth stagnates due to lack of demand.

This is the single most dangerous attribution mistake, and the danger is that it is not obvious because last-click is the default model. The solution is not changing the model to a data-driven one; quite often it is sufficient just to move to a multi-touch model.

Attribution is an estimate, not a truth

It is important to treat attribution with the proper skepticism because otherwise it will mislead you.

No attribution model gives you the truth. The perfect experiment — showing a customer one touchpoint in isolation — is not available. Every attribution model is a simplified representation of incomplete data. Some representations are better than others but there is no perfect truth.

It doesn’t mean that attribution isn’t useful. It means that attribution is not a magic tool that reveals the truth; it is a decision aid, like lead scoring and most other models. The goal of attribution is not to give you an accurate number for each customer but to show a sufficiently accurate picture for making better budget decisions.

A rough and multi-touch attribution picture that stops you from defunding your top-of-funnel marketing is better than a precise last-click number that leads you to a cliff.

When teams overestimate the accuracy of an attribution model and base their decisions on tiny differences in channels’ performance, attribution fails.

How SaaS teams can implement attribution

Whatever model you choose, a proper attribution in SaaS will have several features.

Revenue is connected, not vanity metrics. The goal is to know what campaigns and channels drove the revenue, not just how many sign-ups they generated. A channel that attracts lots of customers who don’t pay is not a good channel, even if the statistics is impressive. Only revenue attribution reveals the truth.

Attribution is multi-touch. Any multi-touch attribution beats the default last-click attribution, because the latter lets one channel take credit for the entire journey. The particular attribution model is much less important.

Tools are connected. Attribution needs you to stitch the customer journey across analytics, marketing platforms, and product/billing data. It may be the biggest part of attribution because it is separate from the selection of a model; the latter is easy once you have the data.

Attribution is constantly updated. The customer funnels evolve, and channels change, and attribution that described the reality last year can start lying to you. The successful teams regularly review whether their attribution reflects the reality.

Most of the above — continuous tracking, revenue connection, multi-touch crediting — is done as an integrated system, which is why it usually takes place inside automation rather than inside Excel sheets that someone updates manually. And that is why done-for-you marketing attribution provides this continuous process of stitching customer journeys and distributing credit. The general idea of marketing automation for SaaS — revenue reporting — reflects it: the attribution model is executed continuously against the real revenue.

The mindset that makes attribution useful

The most useful approach to marketing attribution is to consider it as a decision aid, not an ultimate truth.

You will never know for sure which touchpoint produced a specific customer. However, you can achieve a view that is accurate enough and not misleading to show you what efforts drive revenue. It means avoiding the default last-click attribution, tying the credit to actual recurring revenue, and not being misled by numbers.

Do this and attribution will become the useful tool that guides your budget decisions. Don’t do this or stick to the misleading last-click attribution, and attribution will silently allocate your budget to whatever happens to be the last touchpoint while the engine of your growth remains unfunded.

Frequently asked questions