CMO & Growth Head
Setting up attribution models correctly will enable marketing budgets to be allocated to the right channels and campaigns
A couple of years ago, I was helping a friend run a digital campaign to sell pet supplies. We scaled up the campaigns to drive 10X traffic with a reasonable conversion rate. During optimising the campaigns, we saw that CPS for top of the funnel keywords like ‘Dog Food’ way higher than other campaigns. We, therefore, decided to ramp down the budget for this campaign and allocate budgets to better performing campaigns. Standard operating procedure till here.
Things took a bad turn when we saw that conversion numbers drop. The ‘optimisation’ has resulted in lesser number of people entering the funnel. Buyers would come back after a few days and search for the brand name which wasn’t happening any more since the brand name was no longer in the consideration set. This reduced the overall traffic and sale.
We were committed to last-click attribution at that point in time. It was only after taking a step back and when we mapped the user journey to purchase, did we realise that we were committing a rookie mistake. We were not attributing conversions to the top of the funnel and mid-funnel campaigns appropriately.
With default settings in various tools set to last-click attribution it’s quite often that I notice this trend with many marketers. Only when we analyse the user journey and setup attribution models correctly will the marketing budgets be allocated to the right campaigns and optimisation efforts be fruitful.
A decade ago, attribution was linear. Marketers would start out with a TV Ad, coupled with some print campaigns and radio. The campaign would build up over a week and sales would hopefully start picking up. With digital mediums proliferating, today, the consumer’s journey to purchase is complex. It may start by seeing an ad on Facebook and landing on the website to understand more about the product. This may be followed by checking out review sites, doing competitor research, searching for discount coupons before finally purchasing the product. The attribution in today’s world cannot just rely on single touch.
Adding to this each channel has its own attribution methodology as the channel wants to attribute as many conversions to itself. In earlier days, Facebook attribution was a bit too good to be believable. It has become increasingly challenging for brands to get attribution right. Brands need to now invest in multichannel attribution tools for conversion-based marketing strategies. Besides this, brands need to understand the conversion path of a typical user so that they can assign weights to different brand touchpoints in the user’s journey to purchase.
An attribution model is a mechanism to determine the value of different interactions (like a social media ad, an email campaign or a search ad) towards conversion.
Proper attribution permits marketers to understand the entire customer journey from lead to purchase. It may happen that the Instagram ads or RLSA are not working, then maybe it’s a good time for a campaign strategy reset. Or perhaps the ROI from email marketing is awesome, in which case since it works, the marketer would want to step up on email outreach.
In a utopian world of a marketer, the consumer would buy the product in the first interaction. That’s very rare. Similarly, it is also not possible that only the last click drove the user to conversion.
What’s the correct answer then?
Each brand lends itself to a different model. While the website/app is where the conversion takes place, assigning weights to the various interactions over a defined period of time (lookback window) correctly will help marketer allocate budgets more efficiently and optimise campaigns.
Let’s take a look at the different attribution models:
These models offer marketers the simple solution of assigning full credit to a single touchpoint in the customer journey. These are the easiest to implement and understand.
This model assigns 100% credit to the very first interaction that engages the user and drives him to your website. The conversion may happen much later, often via a different ad campaign. This is very often used by brand marketers when running video campaigns to measure top of the funnel efforts. This model, however, does not do justice to a marketer’s retargeting effort. In today’s environment with multichannel interactions, this attribution model is dated.
A user opened an emailer about new phone model recently launched. He then lands on the brand website. After a few days, he sees the phone video ad on YouTube. He also sees ads on social media due to a retargeting campaign. He then goes on to buy the phone from the website. All credit is given to the emailer in this case.
This model assigns 100% of the credit to the last touchpoint before conversion.
Last-click attribution modelling was the default option in Google Analytics for several years. Many app install campaigns are still run on last-click attribution today. Again, this model does not give any weightage to the customer journey across different channels. It ignores the incremental impact of each touchpoint preceding the last click towards driving conversion.
A user sees an ad for a payment app on a social media site. He then watches a TrueView ad on YouTube. After a few days, he sees an interstitial ad during gameplay and clicks on it to install the payment app. 100% credit is given to the interstitial ad in this case.
This is similar to the last click model only in this case 100% credit is given to the last marketing activity and not the last touchpoint before the user converts. Marketers often relate direct traffic to users who already were impacted by a previous marketing effort. This model filters out direct traffic.
A user sees an Instagram ad for a study table (yeah WFH). She lands on the website and browses for options. After a few days, she goes directly to the website and makes the purchase. 100% credit is given to the Instagram ad.
Here every touchpoint is given credit towards the conversion. The marketing mix involves various channels with different objectives; from creating awareness to engaging and retargeting. Multi-touch attribution gives a complete view of the customer journey by assigning a value to each customer touchpoint.
This model gives equal credit to each ad a user clicks along the way to a conversion, regardless of where in the conversion path the interaction occurs. It’s a simplistic approach to assign the same credit to each touchpoint. Following this method makes planning and budgeting very straightforward - not the most ideal approach though since it is inflexible in assigning greater weights to more influential interactions.
A user interacts with an Instagram ad, followed by a video ad and an RLSA before completing the purchase. In this case, equal credit is given to each interaction.
Here more credit is given to ads that the user has interacted with closer to the conversion time. This model gives less credit to the first interaction and far more credit to the last interaction. Typically, the credit assigned diminishes with a half-life of 7 days. Which means that an interaction done 7 days ago is give half the credit of the interaction done just before the conversion. This model is therefore not very suitable for products with a short duration sales cycle.
If the user searches for a pair of shoes and clicks on the ad, followed by clicking on the social media creative 4 days hence and a banner ad by 8 days hence; more credit will be given to the banner ad and least credit to the search ad.
In this model, the first and the last interactions get 40% credit. The remaining 20% credit is shared between the other interactions. This model weighs in heavy on creating brand awareness (first interaction) and conversion (last interaction). All other interactions are given equal weightage for assisting the conversion. While it gives a higher weightage to major touchpoints, this model ignores the key assists by some of the interactions in between.
While searching for shoes the user first sees the brand ad for shoes. In between the user sees brand videos and discount coupons for the product, but goes on to convert after landing on the website by clicking on a display ad. Here the display ad and the search ad are given 40% credit each while the remaining 20% credit is split between the video ad and the coupon ad.
Similar to the U shape model, this model assigns maximum credit to the interactions creating brand awareness, generating the lead and the last interaction before conversion. Which means that if there are 5 interactions before the conversion, 90% of the credit is split between the first, third and fifth interaction, making the model look like a ‘W’ shape. This model gives a huge weightage to key touchpoints.
This model attributes revenue to activities throughout the entire customer journey. Typically, 22.5% of the credit is given to the first interaction, lead-creation interaction, opportunity-creation interaction, and interaction before conversion. The remaining 10% among all other touchpoints. Though, there are other ways to allocate weights here too. This is the most extensive and technical multi-touch attribution model. Since it tracks every marketing effort, it helps marketer see what exactly is working. Implementing this model is however resource-intensive and is advised for scaled-up teams.
This model is best suited for companies that have a seasoned team, ample resources to track and a sophisticated measurement solution in place. The marketer needs to have an in-depth understanding of the user journey so that he can assign the right weightage for each interaction. There should be enough insights with the team to be able to identify the most influential interaction so that the maximum weightage can be given to it. If not done right, this model takes the marketing strategy down the wrong path. Though this model helps build precise measurements, it is resource-intensive and prone to errors.
In conclusion, choose the attribution model that fits your needs. Be aware of the pros and cons of each model. Attribution is complicated due to various platforms, channels and measurements systems being a part of the user’s journey to purchase. One of the core problems with attribution today is the way marketing data is measured. Due to various channels being siloed, the data is siloed, which in turn requires different types of attribution systems.
Be cognizant of how each attribution model changes the metrics and affects your reporting.
Attribution modelling has an impact on optimising marketing spends. More importantly, it also helps marketers get their marketing mix right so that the users can be guided from the first touchpoint to sales efficiently. Keep working on building insights on what the user journey is and constantly evolve the attribution model that you use.
The infographic below highlights key statistics and different types of attribution models.
Previously published at https://www.damansoni.com/post/attribution-model
Create your free account to unlock your custom reading experience.