How to Report on Multi-Touch Attribution
A Multi-Touch attribution model assumes that all touch points in the customer journey have an influence on conversion. There are three main types of Multi-Touch attribution models: Even-Weighted (each touch gets equal value), U-Shaped (weights first and last touch higher), W-Shaped (weights first-touch, touch-prior-to-conversion, and touch prior to deal higher), and Time Decay (touches near the beginning or end of the customer journey are weighted more).
Why You'd Use This Model
Customers interact with many different events during the buying journey, and multi-touch attribution models are helpful in gaining insight into the different touch points along said journey and how each impacts conversion. Multi-touch models paint a more holistic picture than Single Touch models for marketers looking to better understand the right combination of channels and messages used to achieve higher ROI for their marketing programs. They are also helpful for marketers looking to shorten their sales cycle by engaging customers with fewer but more impactful marketing messages.
WHO THIS is valuable for
Depending on the size of your organization, you may have different stakeholders. If you're a small business or midsize enterprise, this data will be valuable for:
- Demand Generation teams
- Marketing Analysts
- VPs of Marketing
- Chief Marketing Officers
- AVPs, VPs of Sales, Chief Revenue Officers
If you're an Enterprise organization, you can expect this data to be valuable for all of the above roles, aside from AVPs, VPs of Sales, and Chief Revenue Officers.
DATA You Need
- Any data that represents a “touch”. Lead Source, UTM, Salesforce campaign, event attendee list, etc.
- Opportunity data (Opened, Closed Won, Closed Lost) and/or the conversion event data you wish to measure.
- Agreed upon “time-bound.” For longer sales cycles, you may wish to look back 180 days or even 365 days prior-- relative to the conversion date. This sets the stage to how far back should data be counted.
data sources required
The more [clean] “event” data set you can add in, the more robust the resulting model will be. For example, a log of website visits + email clicks + form submits + sales activity is better than just having one of those sources. This includes any marketing or sales activity involving an interaction with the customer. This dataset will become the source data for attribution. These events (or touches) are all eligible to get the credit for the conversion
- Marketing Automation Platform (Marketo, Hubspot, Eloqua, Pardot, etc)
- CRM (Salesforce, SAP, etc)
- AnalyticsJS / product behavior
- Web Analytics (Google, Adobe)
- Ad Platforms (Google Ads, LinkedIn, Facebook)
Key Considerations When Selecting a Multi-Touch Attribution Model
There are many ways to build multi-touch attribution -- some more robust than others! Below are some key things to consider when determining if multi-touch attribution is right for you and how to apply it to your own data.
Be Aware of the Caveats
Marketing teams everywhere are adopting multi-touch models, and they are certainly an improvement from single-touch and a step in the right direction towards accurately reporting on attribution. This being said, each model has its caveats, and they must be considered when selecting which attribution model is right for your business.
Check out our webinar on attribution models from our 2019 Revenue Marketing Innovation Series here, where we discuss each of the models mentioned in a bit more detail.
Gain Alignment on Time Constraints
It’s crucial that before you begin building a multi-touch model, you gain alignment on the timeframe in which touch points must have occurred relative to conversion in order for them to be included (ex. “Only touches up to 60 days before a closed opportunity will count”).
Without determining how far in advance (or how recent) a touch point either becomes invalid or too significantly weighted, you may be including touch points that aren’t actually related to the revenue you’re reporting on.
Start with a Pivot Table
To better understand the concepts in multi-touch, sometimes it’s easier to take a first pass at a data model.
To build a basic multi-touch model, start by looking at a data set that pulls in campaign types, how many days from touch point to close, how much revenue is associated with each touch point, and how you’ll weight them.
Here's an example of how to download and format a data set:
| Class ID | Touch ID | Touch Date | Days to Close | Campaign Type | Attribution Pct | Attribution $ | Revenue |
|---|---|---|---|---|---|---|---|
| A | 1 | 11/31/2018 | 31 | e-Book | .5 | $50,000 | |
| A | 2 | 12/15/2018 | 16 | Webinar | .5 | $50,000 | |
| A | A | 12/31/2018 | 0 (too new) | Opp closed (exclude) | -- | $100,000 | |
| B | 3 | 12/01/2018 | 39 | e-Book | .33 | $40,000 | |
| B | 4 | 12/15/2018 | 24 | Webinar | .33 | $40,000 | |
| B | 5 | 12/31/2018 | 7 | Free Trial | .33 | $40,000 | |
| B | B | 1/7/2018 | 0 (too new) | Opp closed (exclude) | -- | $120,000 | |
| C | 6 | 1/30/2018 | 180 (too old) | Blog subscription (exclude) | -- | -- | |
| C | 7 | 5/15/2018 | 55 | Free Trial | .25 | $20,000 | |
| C | 8 | 6/08/2018 | 22 | e-Book | .25 | $20,000 | |
| C | 9 | 6/28/2018 | 2 | Filled out form | .25 | $20,000 | |
| C | 10 | 6/29/2018 | 1 | Webinar | .25 | $20,000 | |
| C | C | 6/30/2018 | 0 (too new) | Opp closed (exclude) | -- | $80,000 |
This will enable you to create charts like this:
More Playbooks
The Weekly Revenue Marketing Report
Sign up to receive a weekly personalized email from Chris Nixon, the host of the Revenue Marketing Report.