Answers to Questions Marketers are Asking about Customer Journey Attribution
Q&A from our recent webinar on how to move to a multi-click attribution model
Our first webinar of the year, Getting Started with Customer Journey Attribution, took place earlier this month, hosted by Lothar Krause, our VP of Global Advisory Services, who shared insights and best practice from more than 17 years in digital, working for companies such as Zanox, eBay and Zalando. With more than 600 registrations and a multitude of questions at the end of each session, we could see that this is a topic that keeps a lot of marketers up at night.
Want to see the webinar recording? Email us to request the link.
In fact, there were so many great questions, that we decided to compile the greatest hits, and share the questions and their answers with you here.
Marketers, which attribution model are you currently using? Vendors/Agencies, what is the model most seen by your advertising clients?
In fact, the first question came from Sociomantic to the audience, in the form of a live poll. Answered by more than 80% of the nearly 200 participants, the results showed that the last-click model is still the most common, but that more advertisers than we expected are already working with more advanced, multi-click models that represent a fuller picture of the customer journey to purchase:
The remaining questions were posed by users throughout the webinar, and answered by Lothar in the Q&A session at the end or via email follow-up after. Here are the highlights.
Before even deciding how to weigh and model marketing touchpoints, we need to set up a way to track the user. How much volume is needed, and what’s involved in that process?
Attribution should be a part of your marketing planning right from the beginning. If you’re a new company, with a smaller budget, there are free analytics tools you can use, like Google Analytics. The free version may not be as sophisticated as paid options, but for the beginning it’s more than enough to get started.
We recommend that even startups should start with a more sophisticated, multi-click attribution model from the beginning, because that means you will allocate your limited budgets more wisely right away, and you won’t have to deal with the resource-intensive model switch on down the line. Also, young companies using cost-per-action/order should consider measuring additional KPIs such as customer acquisition cost (CAC), which will support new customer growth.
How many marketing touchpoints do you normally see in a customer journey?
The number of marketing touchpoints prior to purchase depends on how aggressively you run your online marketing channels. This is especially true for retargeting, which can be very aggressive depending on the vendor or vendors you work with. This is one of the reasons why we made our frequency capping algorithms transparent to our clients, so everyone can see how a particular user group is targeted throughout the journey.
It’s important to note that only a few digital channels, such as programmatic advertising, make it possible to focus tactics at the user level; with other channels it’s not possible to define the desired number of touchpoints per user with such precision.
How can you track the customer journey in the app environment?
In the last few years, the app environment has been of rising importance in the overall online marketing mix. As such, it’s important to track the performance of your app and app marketing in your attribution model.
The app environment is different than traditional browser-based environments, primarily because tracking is based on device ID (android ID or IDFA) instead of browser cookies. The great thing is, that within the app itself, you can track every event on a user level every time he is using the app again, because you always have the same identifier. But when you switch environments (from app to browser and vice versa), you have to work with specific companies that specialize in this type of tracking. You need to connect the two environments and evaluate the cookie-based impressions with the click/performance-based device ID (i.e., what’s happening in the app).
What it the typical late conversion window you are seeing in the market?
This question is not easy to answer, because the answer depends on the business model and whether your website visitors tend to buy products/service quickly, or if they need some time to consider each purchase. In recent years, marketers came to the decision that the standard attribution window click-to-sale should include the last 30 days.
The situation looks a bit different if we’re talking about view-based attribution. For performance campaigns, this window could be up to 48 hours, but for branding campaigns (or channels with a branding effect), you could easily increase this time frame. Also consider the exceptions on when to include a view, based on time proximity to another trackable action (as discussed in the webinar), such as when a click from another channel comes immediately following a view.
When should media KPI’s such as bounce rate overrule the traditional performance KPIs?
This depends on the campaign and on the channel — one must measure performance campaigns with performance KPIs, but media KPI’s can be useful for getting a better understanding of branding/reach campaigns, especially if you don’t generate enough website visitors via a branding campaign, or you need to adjust your spending before the whole attribution window has finished.
In this case, you could also first run evaluations with media KPIs before your late conversions are kicking in; for example, you normally don’t buy a trip for your whole family on short notice, but you might be able to identify campaigns with a nice branding effect based on how users acted on the site, even if they didn’t convert.
For the travel industry specifically, and for hospitality in particular: considering the competition between online travel agencies, do you believe brand search should be excluded from the model?
I agree that when Google allowed for other people to bid on your brand, competition grew. However, if you want to include brand search with a lower percentage that might be fine, but it shouldn’t get the same credit as the click on another SEM keyword search. When you see a click before the brand name search, you can assume the user already knew the brand. It would make sense to run two parallel reports, and from that see what makes more sense for your brand.
Do you have suggestions of KPIs specifically for display?
For a reach/branding campaign, you need to work with the view. With these more traditional display campaigns, we recommend you use the media KPIs to measure the quality of the audience. Look at things like whether the users bounced or the time on site of the users who arrive from a particular campaign, especially if it’s at beginning of the customer journey, to see if the campaign is bringing relevant traffic or just bringing bouncers or even some false clicks.
How often should I adjust my marketing budget according to my attribution model?
The frequency of budget adjust should really does depend on the amount of marketing budget you want to spend, because the more you spend, the more quickly you can, or better say, should adjust.
In general the departments (and of course the CMO) should receive reporting daily. I think that monthly you should look at the overall budget and analyze it. Weekly, for example every Monday, do some adjustments, even if just a change of 10% plus or minus. And be careful: if you have something more specific, like an email campaign going to millions of users, look specifically at how attribution went on that day, or even ignore the specific day in your reporting.
When switching from one to the next (more sophisticated) attribution model, how much time should they run in parallel?
It’s difficult to track everything correctly with a new attribution model, especially when you switch from a single-click model to a multi-click one; the fact that every channel must have adjusted spends according to the new model will make it even harder. Be aware that some departments will have doubts whether the new spend allocation is correct or fair. To arrange a smooth transition period, we recommend running both attribution models at least four weeks in parallel. During this time, the spends should still be allocated according to the old model.
Be aware that your internal departments may have to create new channel-specific reports to support the new model. Besides this, many departments might need to explain to marketing vendors and partners why their operative KPI is changing, especially if they will be valued with a lower KPI under the new model.
For best results, we recommend that you explain your new model and updated KPIs to your partners in a transparent manner, so that you can learn together how best to adjust to the new model.