As a marketer in the digital landscape, the struggle to measure the performance of an ad and to give credit to a specific attribution model as the possible reason for conversion is a challenge.
One wrong move and you end up measuring the wrong results and making business decisions based on the wrong data.
Bram De Jonge, from his experience as Senior Account Relationship Manager at AdRoll, thinks that measuring marketing impact is not just about measuring marketing results, but focusing on business results for better long-term business operations.
Q: At the risk of sounding simple… why is measuring ad performance so hard? It seems like the more advanced tech we get, the harder it is to get “simple” answers. How do you make sense of this phenomenon at AdRoll?
Consumer behavior has changed a lot in the last decade. We have been using more and more devices and basically became more dependent on the internet. This has changed the customer journey, the number of potential touchpoints with ads, and the ways brands interact with potential customers.
For example, if your TV is no longer working, you probably start your journey looking for a replacement by searching online. You might remember a couple of brands or stores and check their websites. In order to figure out what price point is right, price comparison sites are used. You might even check videos and websites for reviews. This process is mostly done in between other online activities (reading some news websites, checking social media, etc.). The more touchpoints of marketing within a customer journey, the more difficult it gets in order to figure out if each of these touchpoints — and to what extent — has a true added value in the purchase decision.
Many marketers struggle to figure this out and so choose to stick to a more simple default model, mainly last-click. At AdRoll we have seen this trend by our advertisers and started building a measurement platform that allows tracking the impact (both clicks and impressions) of all your digital marketing channels within one view. Being a display advertisement platform originally, AdRoll has always had the challenge to prove the impact of display on the overall marketing mix. Display retargeting mainly triggers indirect actions with no click involved, and this impact is simply not measured in a limited model like last-click.
Q: In your presentation, you talk about the different eras of marketing attribution: the Click era, the KPI era, and the Impact era. Can you explain a bit about what each means and what their respective limitations and opportunities are?
This part of the presentation is meant to highlight that technology and shopping behavior has evolved a lot, and that the way marketers track and attribute belongs to an era already past. Attribution models are often far behind the current marketing tech.
The click era relates to the first years of digital marketing. Online shopping was not yet baked into our commercial ecosystem, the internet was not yet the main source of information, and therefore online customer journeys were not yet complex. There was not much more to do than clicking on an ad within a search engine and the number of touchpoints online was limited. Measuring clicks made sense at that time and it was easy to set up. The main complication here: as soon as you have multiple touchpoints, this model no longer holds. Why should only the last interaction get all the credit?
When more and more online advertisement options were created (display, email, affiliate, etc.) and adoption of the internet further developed, it became difficult to track the performance of all channels in just one platform. The market started to adopt a more siloed approach, where KPIs like a CPA or ROAS were measured within the different marketing platforms used or per channel. Typically, a single-touch, click-based attribution model is still at the basis, but it is at least a first attempt to compare performance between channels.
Currently, in the impact era, you see that marketers recognize the flaws in click-only and/or single-touch attribution, as the wrong channels are being rewarded or conversions are attributed although they are not incremental. It’s for that reason that they start to wonder what the true impact is of each of the channels on their overall business results. This is exactly the question each marketer should ask, but it also brings up the challenge of “how can we actually do this?”. Measuring all touchpoints and multi-touch attribution is required to start this journey.
Q: Just to follow up on that, what kinds of different attribution models work best for different types of companies/products/services? Can you make any specific recommendations for B2B vs B2C brands?
The quick answer: Ideally, each company should have a data-driven and multi-touch algorithmic attribution model (clicks + impressions). There is no difference between B2B or B2C in this case, as this holds for any company that has a customer journey that consists of multiple touchpoints through content and marketing campaigns.
In general, as soon as you have more than one marketing channel in your mix, any single-touch attribution model (first-click / last-click) doesn’t make sense anymore. You let channels compete to win the same conversion, where you actually want them to work together in order to bring incremental revenue. Optimize the entire mix instead of individual channels. This can only be achieved when an attribution model is in place that actually can give credit to this cooperation. Additionally, if you have any marketing channel in your mix that has a visual impact on consumer behavior — such as display, video, email, or branded search — only tracking clicks is not sufficient, and tracking impressions as touchpoints is crucial.
Display, as an example — the majority of its effect is not through a click, but through the unconscious effect it has on the consumer (building brand awareness and recognition to develop demand and increase purchase intention). Not only is the proportion of click interactions on display low (± 16% of impact), the clicking proportion has almost no correlation to generating conversions. That’s already proof that only measuring clicks for this channel — and attributing based on click only — doesn’t make sense and shouldn’t be an optimization focus.
Last click attribution can actually harm business models as it is not rewarding the true incremental channels, causing marketing budget to be distributed to the wrong platforms. Adidas recently announced that they have seen this negative effect, causing them to over-prioritize short term digital marketing and not credit their branding activities properly.
Q: When it comes to measuring marketing impact, you stress that it’s important to focus on business results, not just marketing results. In my experience, this can lead to a bit of a chicken-or-the-egg scenario, in which it’s impossible to prove business results without good data, but that requires new (expensive) technology and internal resources. So, which comes first?
I agree with you, and this is probably also the reason why there is not enough movement on this front. At Spark, I touched on the concept of exponential thinking that is required in order to take steps forward. With clients, I often start with data analysis and providing studies to highlight that their current model has a lot of flaws and that there is a significant risk that the company is measuring the wrong results and therefore making business decisions based on the wrong data.
There are low-cost tests that can be implemented (an A/B cookie split in Google Analytics in order to measure the effect of a marketing channel on the marketing mix, for example) as a start in order to give the first indication. But for true measurement, an investment in technology is required indeed.
So to answer your question, I would say that first of all a company needs to have the will and accept the need to change, which should lead to the business decision to invest in better technology that can fill the gaps currently not measured. This must be seen as a business operations decision, and not as a marketing investment, as it will ultimately lead to improved long-term business operations just like optimizations in a supply chain can do.