Why is Multi-touch Attribution So Hard?



The holy grail of marketing data always seems to aim towards a multi-touch attribution model. A magical place where you know the exact impact of your investments and exactly where to put your next dollar. Seems simple enough to understand, track engagement, track your spend against that engagement, and then report on what’s working best and what’s not working. If it’s so simple, how come so few companies actually do it? And, when they do, why is it not giving them the visibility they expected?

First, what is multi-touch attribution? And why do we want it?

The high level concept is pretty simple. For most B2B companies, a sale rarely takes place with just one interaction with the company’s marketing efforts, it usually takes quite a few interactions before a deal is ultimately signed. So, rather than just giving credit to the first interaction or the last interaction, it would make sense to spread the credit across multiple lead sources. That way, you know which lead sources are performing well and where you should double down on your investments. It may look something like this.

Stealth Startup closed a deal worth $100,000 and we have tracked that the primary contact downloaded a white paper from our website, joined one of our webinars, got a demo booked with one of our SDRs, met with the sales team at our big industry trade show, and then finally closed.

If you’re following a first touch model, credit would go to the white paper, which doesn’t seem fair given the high engagement with the webinar and the tradeshow. If you’re following a last touch model, the trade show would get credit, but what about the SDR that booked the first demo? They helped too right? So let’s do multi-touch, all we have to do is split the $100,000 over those lead sources…and this is exactly where it gets tricky.

How do you decide how to distribute the credit?

Tracking the activities is relatively straightforward, this is the most difficult part. Here are a few ways you can do it.

Evenly distribute the credit:

You can simply divide the contract value across the four lead sources. It might look something like this:

White paper: $25,000

Webinar: $25,000

SDR: $25,000

Tradeshow: $25,000

This is easy and you can implement something like this right away, but what is this really telling you? Did the white paper really have the same impact as an SDR getting them on the phone and booking a demo? And would you actually say that the ROI is justified for each of these investments? Is the Tradeshow getting the credit it deserves by facilitating an in person meeting with a sales person to close? Would you feel comfortable dividing your spend by this amount and saying that’s the lead performance?

Develop weights for each lead source: You can weight the lead source credit by ascribing effectiveness value to the lead source. One might say, trade shows and SDR interaction are worth more of the credit than white papers and webinars, and you may see something like this.

White paper: $5,000

Webinar: $10,000

SDR: $30,000

Tradeshow: $55,000

The only issue is, these weights are arbitrary and it’s tough to truly say if the tradeshow in this case actually helped move the needle along more than any other deal. It would be like coming up with sales stage conversion rates without it being based on actual conversion data.

Hire a data scientist to develop an attribution model: The best way to get the weights you’re looking for is to do a proper data study on the causal factors that move a deal to the finish line. Now, this assumes you have enough data to be statistically significant and you are collecting the right kind of data that can be used to do the study. This will also need to be refreshed constantly.

For big companies this might be doable, they have the budget to dedicate the resource and the cost can justify the improvements in their customer acquisition cost.

Use a multi-touch platform: There are tools that will attempt to automate this for you. However, just like any tool that you bring on, it’s only going to be as good as the data you collect. If you don’t have enough data or it’s not collected in a way that’s usable, you’ll have a tough time getting one of these to work.

There is a cost to doing multi-touch

This is not something you achieve without a cost. The resources dedicated to doing this properly as well as the operational rigor required to feed data to these resources both come with a cost. For some companies, it’s completely justified and will absolutely create a competitive advantage. For others, it may be an endeavor that costs more than it’s worth. Here are a few costs:

  • Time for marketers, sales people, and customers to provide data and organize in a way that is useful
  • Dedicated person with enough experience to manage it properly and articulate what the data is saying to other members of the organization
  • Tools and technology to scale the efforts

If multi-touch is too much for us, what should we do instead?

For the majority of early stage startups: seed, series A, series B, you can get meaningful intelligence that will give you enough data to make investment decisions and it will cost significantly less than trying to roll out a multi-touch attribution model. Here is a recommendation that should work for most of you out there.

  • Track first and last touch these are usually the most important interactions a prospect has with your business. How are they hearing about you and what is the last interaction they had before turning into a sales qualified opportunity?
  • Report SQLs and Bookings by first touch and last touch: Give full credit to each one and report on it separately.
  • Set up influence reporting: By each lead source, simply track how many SQLs and bookings interacted with that lead source. You may find that there’s an unsung hero in the middle of the sequence that tends to always have a helping hand in creating SQLs and closed won deals and can justify continued investment even if they don’t get the first touch/last touch spotlight.


Multi-touch attribution is ideal, it’s just hard to do, expensive, and for most it’s overengineered for what you really need. Tracking marketing activities is not the hard part, you should absolutely track all marketing interactions, it’s distributing the credit that gets tricky. If you can’t do multi-touch, give credit to first touch and last touch lead sources and set up influence reporting for all of your lead sources. At the end of the day, you need enough data to justify your decisions, it’s never perfect, but as long as you can feel comfortable putting a dollar behind a lead source that’s what matters.

If you want to discuss this topic deeper I’m always happy to connect