What Types of Questions Can Marketing Attribution Answer?

You need to know if your marketing is working. And if you want to do that, you need some kind of marketing attribution model, even if it’s as simple as asking customers “how did you find us?”

So a lot of marketers will ask “which attribution model should I use?” or “which marketing attribution tools should I use?”

But what they really should be asking is a question far more basic: “What am I trying to learn?”

So consider this: your email tool says that email drove 60% of last-click conversions. Should you increase the email budget?

This sounds like a lay-up, like an obvious and easy yes, but reality isn’t so simple. Your answer needs to depend on whether these customers had already decided and if email just happened to push them to the buy button or if email meaningfully influenced their decision.

In this post, I’ll go over what marketing attribution can really tell you, what it can’t, and how to use it alongside metrics like customer acquisition cost (CAC) and customer lifetime value (LTV).

Marketing attribution is how you find out which marketing channels and campaigns lead to sales, conversions, and other goals. Simple as that.

Or to be even blunter, it tells you which marketing is actually working.

Attribution tools help you see what customers did and infer influence from behavior. But you have to be careful about what you assume, because marketing attribution can never really prove causation. And that’s OK, because you can learn a lot with correlation alone, even if you can’t see every offline conversation, hear all word-of-mouth, track every social post, or log every brand impression.

But before we go on, let me make an important distinction:

When you can measure lift, you can measure how effective your marketing is at accomplishing its goals. Smart attribution modelling makes it easier for you to measure lift.

Measuring lift is the goal, modeling attribution is the means of achieving the goal.

Attribution is a great way to see the patterns at play in your business. You can see which marketing channels are present in customer journeys that end in conversions. And you can see how those journeys unfold and where you might be over- or under-investing.

I’ll give you some specifics so you can see what I mean here.

Attribution can show you which marketing channels are contributing the most conversions and which campaigns are the most effective. Key word: “contributing.”

Consider the tight pickle jar. A man struggles to open it, to no avail, passing it to his wife, who opens it immediately. In an effort to save face and appear macho, he says, “I loosened it.”

And maybe he did. It’s impossible to know.

That’s the essential problem with marketing attribution. It can tell which channels opened the jar, and it can even tell you which channels had their hands on the jar. But it can’t tell which channels loosened it.

To this end, attribution models can help you figure out if you’re overspending on top-of-funnel channels with little impact on actual conversions, or under-investing in mid-funnel channels that consistently appear in multi-step customer journeys.

Attribution can show you, to a reasonable degree of accuracy, how many touchpoints it takes to convert a customer. If customers typically encounter 7-10 touchpoints, you know single-touch models will give incomplete information and that you need sustained presence across channels.

Really good attribution models can show the time lag between first touch and conversion. This might be 90+ days for B2B or 3-7 days for eCommerce—it varies massively from industry to industry. But knowing how long it takes is crucial, because if the sales cycle is 90 days, then well, you shouldn’t cut campaigns off after 30 days if you get no results.

When you collect enough data, attribution can tell you which content formats are most engaging at different stages of the funnel. Perhaps customers who attend webinars convert at 3x the rate, or maybe you’ll find that those who engage with three or more content pieces convert better.

These are the sorts of patterns that attribution models can help identify. And that can give you information you need to set up marketing sequences that are more likely to end in success.

Marketing has to answer to accounting at some point. And it’s here that attribution can once again be useful, showing what the marketing team’s contribution to generated revenue was. It can also show which assets drove the most business in specific timeframes.

Attribution can show which blog posts, webinars, campaigns, and so on put people in the pipeline for sales. That lets you double down on what works.

On a more broad level, marketing attribution can help the marketing team in general show ROI. This, in turn, can be used to set an appropriate budget needed to win the amount of revenue the business as a whole wants to generate.

There’s a whole lot you can learn from marketing attribution. But there’s also a lot of limitations and they’re not obvious—a fact which gets a lot of inexperienced marketers into trouble when it comes time to make good strategic decisions.

Here's what attribution struggles with or simply cannot answer:

Email might get last-click credit for 60% of conversions, but many of those customers may have already decided to buy based on brand trust built through other channels. Email just happened to be the mechanism they used to complete the purchase.

And, sadly, marketing attribution can’t show you the decision-making process in the customers’ heads. The closest you can get to doing that is through lift testing or incrementality testing, where you withhold a campaign from half your audience or from a geographic region and see whether there’s a difference in sales.

The most common manifestation of this problem is simply giving the last stage in the sales process all the credit. But there’s all kinds of ways that conversion attribution can go off the rails—a topic I’ve written at length about.

Attribution only sees trackable interactions. You can’t hear every word spoken about your brand at kitchen tables and conferences. You can’t tell if people are talking about you on their social media DMs. You can’t tell how many drivers really looked at your billboard on the highway, and you can’t tell if someone saw your logo on an old custom pen made 20 years ago and that they were prompted to Google you afterward.

Someone might see your LinkedIn post, text a colleague about your company, and that colleague then searches for you directly and converts. The point is: marketing is irreducibly messy because there are so many factors to consider. We’ll be lucky if we can, in our lifetimes, get to the level of precision that weather forecasters enjoy in modeling day-to-day odds of rain.

In the meantime, so many of the examples I just cited are going to show up in your dashboards as "direct traffic" or "organic search," missing the entire influence chain that drove the conversion.

Display ads might build awareness that converts six months later through a completely different channel. Attribution captures journey sequences within a defined window, but it doesn't measure brand equity accumulation over time.

Similarly, consistent content marketing might establish your authority in a way that makes all your other channels more effective, but attribution reports won't necessarily show this clearly.

This is the kind of problem that brands have to account for when they sponsor sporting events and conferences. They come around so rarely that lift testing is incredibly slow, so they just have to make a reasonable guess about what true brand impact is going to be.

Attribution inherently looks back in time. So it can’t tell you if you should enter a new market. It can’t tell you how to future-proof the business.

Likewise, attribution alone can’t tell you what’s going on in people’s heads. It can’t tell you how your messaging makes your ideal customers feel. It can’t tell you if your pricing is high or low or just right. These thorny problems require their own special kinds of analysis.

I am not impressed by most marketing metrics. In fact, I think that unless you can tie a metric, in some reasonable fashion, to revenue and profitability, you shouldn’t spend too much time thinking about it.

Even saying that, if you’re serious about marketing profitably, I strongly recommend you start with customer acquisition cost (CAC) and lifetime value (LTV). If your marketing can sustain a CAC that's profitable relative to your LTV, then your business has the potential to keep growing.

CAC is how much it costs to win a new customer. This includes advertising spend, marketing and sales salaries, content creation costs, software tools, and any other expenses involved in converting a lead into a paying customer.

LTV is how much net profit a customer is expected to generate over the span of their relationship with a business. It shows you what the potential value of a customer is to your business and helps you make strategic decisions.

It’s said that the ideal LTV:CAC ratio is 3:1, meaning you should make three times what you spend to acquire a customer. This gives you room for costs, overhead, and profit margin.

Reality is more complicated.

In reality, it comes down to unit economics. If you take a customer's LTV and subtract both the CAC and the cost to serve that customer, and the result is positive, your business is fundamentally viable.

If margins are really high, then an LTV:CAC of 1.5:1 might be good.

If margins are really low, then you might need an LTV:CAC of 10:1.

It depends really heavily on the business.

Even so, once you understand CAC and LTV, you understand the basic levers of marketing. And attribution, you’ll be pleased to know, goes a long way toward helping you figure out which channels contribute to your CAC.

There are three main categories of attribution models, each with different strengths:

There’s no single correct attribution model. In fact, if anything, I’d recommend you use more than one at a time and check them both regularly. If different models tell you the same thing, then you can have more confidence in your decisions. Where there’s disagreement among models, you’ll be prompted to ask better questions and run better experiments.

Marketing attribution is a tool. It will help you understand patterns in your customer journeys and find opportunities for improvement. But as precise as the numbers may seem on your computer screen, that doesn’t mean attribution models provide perfect answers. They just provide you information you need to ask better questions.

Attribution works best when you understand its limits and use it as one input among several. Combined with lift or incrementality testing, an understanding of what makes the business profitable, and good old-fashioned clear strategic thinking, it can help you make better marketing decisions.

The goal isn't perfect measurement. In fact, that’s impossible. Customer journeys, and for that matter, life, are too complex and many influences are untrackable.

The goal is making better decisions with imperfect but useful information. Attribution, used thoughtfully, helps you do exactly that.