Creating an Attribution Infrastructure
I’ve talked a lot in the past about the importance of using a multi-touch attribution model with data weighted and reported by an unbiased third party. In this blog, I wanted to take this conversation to the next level by explaining the infrastructure you’ll need in order to accurately measure and track your customer’s journey towards a purchase so you can identify which of your vendors are performing – or underperforming.
Don’t misunderstand me. This takes a lot of work. If it were easy, everyone would be doing it already. With that being said, here are 12 key things you need to set-up and the data points you need to truly get accurate multi-touch attribution results. If you are missing any of these, your multi-touch attribution model is flawed.
1. Advanced Tracking Code beyond Google Analytics – If you want to be able to attribute to sales, you need to use a third-party provider that has their own proprietary tracking code that can de-anonymize website traffic and link to sales.
2. Call Tracking Logs – It’s imperative to have your call tracking set-up properly so that you can really know what generated that phone call, not just have it labeled “website.” The tracking logs should be able to tell you whether that phone call was generated from paid search, a click from Facebook, or another specific source. This weighs heavily since many consumers call in rather than convert.
3. Dynamic Call Tracking – This level of call tracking enables you to further drill down on the influences that triggered that phone call. Rather than simply knowing that the phone call was generated through paid search, dynamic call tracking tells you that it came from a specific keyword which triggered the ad, allowing you to attribute and monitor specific campaign results.
4. CRM – CRM data is important as one of your data sources for obvious reasons, but mainly for the ability to do what we call leads-based attribution.
5. DMS – DMS data is also key for the sales data it contains, which is necessary to then credit vendors with weighted sales numbers. Then you are able to measure actual ROI in dollar figures against actual sales and profit. If you only rely on leads, you’ll be missing a large piece of your marketing eco-system in terms of what’s working and what’s not.
6. AdWords & Bing – The most important thing is that you have admin access to your account and data. This will enable ingestion of important data that you can’t get any other way.
7. View-Based Tracking – View-based tracking is one of the most important pieces of data to capture as roughly 90% of consumers don’t click, call or chat from a third-party site — they simply come in and buy a vehicle. In addition, you must take into account all of the display ads and retargeting campaigns that consumers are exposed to. The consumer may see the ads, not click through at that moment – or ever – but still buy a vehicle. If you’re evaluating your display ad vendors based on clicks, you’re missing a big gap in their true performance.
8. Third-Party Auto – As I mentioned above, in order to measure the full impact with third party auto sites, it is imperative to work with a provider that can place tracking code on third party websites across all your dealer pages on the site. This will enable you to effectively measure VDPs where no action was taken immediately, or directly, by the consumer who ends up purchasing a vehicle from you after viewing your VDPs on third party auto sites.
9. Chat – Of course chat logs are important for several reasons – not only can you track the data that initiated the chat, but oftentimes consumers will reveal the source within the chat.
10. E-Mail – Don’t simply rely on attributing a sale to an email campaign just because the email was sent. Install tracking pixels within the email campaigns so that you know whether a particular person opened the email, was exposed to it and you can then attribute it back to that view or click.
11. Direct Mail – Direct mail requires a little more scientific measurement, but you need to make sure that you’re not simply matching to whom a direct mail piece was sent against which of those people bought a car. Utilizing more advanced approaches like tracking numbers or trackable urls in your direct mail should be used at a minimum. There are some more advanced approaches but I will save that for another blog post.
12. TV Spot Logs – If you want to measure TV, it’s imperative to get the TV spot logs so you can overlay that data with spikes in website traffic. However, it’s more than just comparing the two pieces of data as there may be cases where you have overlapping TV spots. There needs to be a way to fractionalize that influence between overlapping TV spots.
13. Anonymous Attribution – According to a Polk Automotive Influence Study, approximately 2 out of 3 car buyers do not contact a dealership via form submission or a phone call prior to their first visit. This data gap leaves a huge blind spot in a marketer’s ability to accurately attribute sales to sources like paid search and third party sites. Identifying anonymous shoppers requires the development of proprietary identification and tracking technology.
14. Data Scientist – Once you have connected all of the marketing touchpoint data, you need to figure out a way to assign credit to each of those touch points. You can leverage rules-based attribution models or data driven models (click here to watch a video blog that outlines the different types of attribution models). You’ll also need the expertise to visualize and analyze the data in order to for it to be really useful.
That’s it! Easy…. right?? Developing a comprehensive multi-touch attribution model, having the ability to access the data needed and making sense of it all is complicated. A lot of it relies on the cooperation of your vendor partners.
If you want to truly measure the real influence your marketing sources are having on sales, you can start by partnering with attribution-friendly vendors. Without their assistance, you will find that properly building your attribution infrastructure is frustrating, inaccurate and, ultimately, near impossible.