Attributing the ROI and ROAS (Return On Investment and Return On Ad Spend, for those who aren’t fluent in marketing slang) of digital marketing has been a challenge for many years. I have been working with FTSE 100 and FAANG companies for the past 15 years trying to solve these challenges, and it remains a challenge for even the biggest digital advertisers.
Digital marketing talks a good game about being able to track a user from ad exposure through to purchase with total accuracy and transparency. The reality is, that is far from the truth.
Anyone with a basic understanding of human behaviour knows that you don’t see an ad for a brand that you've never heard of before and instantly click through to buy the product without any other thought. But that is the basic premise of last-touch attribution, which remains the default attribution standard in digital marketing. It's most likely the method you're currently using to see channel attribution in Google Analytics, Shopify, or in your customer tracking.
Google Analytics will usually show high volumes of sales coming from Direct traffic. And the most effective paid marketing source will often be Paid Brand Search ads (when people type your brand name into Google). But we know that people don’t just wake up one morning and decide to type a brand name into a URL bar or into Google. Those searches and direct sessions are caused by something. And that is where marketing attribution gets a bit more complicated.
This complication frequently leads to misleading ROI for marketing channels. So all of your marketing efforts that lead people to making those brand searches rather than clicking on ads and buying a product straight away are terribly undervalued. Standard digital attribution values the triggers of actions, but ignores the drivers of the triggers.
Plug into our data science lab to get deep insights into what drives revenue.
Google Analytics can give you some view of the drivers through multi-funnel journeys, but this relies on a user staying on the same device or on a single Google login. Again, we know this isn’t normal human behaviour, people browse from a mobile device at home while watching Netflix to then later purchase at work on a desktop. These driver activities would be lost without an alternative attribution.
Time series data models let us see where this driver activity is taking place, and help us to better understand the value of the driver activities. These time series models can range from basic correlation models, to in depth multi-level regression models, and also work really well when paired with A/B or multivariate testing.
These statistical models isolate the correlations between increases in each marketing channel, and the change in underlying total sales data. They can also show us the correlations between one marketing channel and another to help us to see how consideration driven in one channel helps to drive revenue triggers through another channel.
This gives us the breakdown of drivers and triggers for each individual business, along with their relative strength, the level of incrementally for each channel (ie. how many of the attributed sales we would have got anyway without the marketing interaction), and a truer understanding of marketing ROAS and ROI.
Beyond this improved understanding of revenue triggers, we can also have a better understanding of the drivers that are nudging people towards the direct sessions and brand searches which ultimately lead to sales.
For example in this mock chart below, we can see how Facebook impressions lead to 21% of direct sessions. A relationship that wouldn’t be shown on Google Analytics or Shopify but allows us to reallocate some of the value of the conversions from these direct sessions back to the trigger that caused them to happen. We also see that the true value of email is being undervalued as some customers don’t click directly on the links but will later navigate to the site directly (representing 14% of direct sessions.)
We can also see statistically significant relationships between organic search traffic being driven by print and PR activity, along with the underlying brand equity being driven through branded marketing. With the high effectiveness of organic search traffic, this allows us to reallocate revenue back to print and PR channels, along with the brand marketing, all of which are typically very difficult to measure on an ROI basis.
Understanding these relationships allows us to open up more marketing channels to measure and buy on an ROI performance objective. Engagement-based marketing channels such as TikTok, YouTube, and Display ads tend to perform badly on the last touch tracked ROI and as a result, are difficult to justify for performance businesses. But being able to directly measure the sales driven through these channels but triggered through a different channel allows us to understand the real value coming from these channels to help drive growth from a larger and more diverse series of marketing channels.
If you'd like to know more about how our data science team can help drive growth for your brand, drop us an email at email@example.com. And for for more data science insights and breakdowns, make sure to sign-up to our newsletter - Cuppa - here.