As Marketers, being data-driven has become part of the job. It’s part of the revolution that has been happening for the last few years.
At the center of this process, Analytics measure conversions and attribute sales. To help them, Attribution Models tie together conversions and traffic sources by attributing credit to the latter.
The consumer’s buying journey has evolved in the last few decades. Analytics struggle to understand the flow between all touch-points and fail to consider all the factors involved in a conversion.
As a result, budget is invested where it shouldn’t be, because data is flawed by current Attribution Models. Your marketing plan is not as effective as it could be.
What does the future hold? What do you need to improve your marketing activities? What about Big Data and Machine Learning? Can they help?
Problem with Current Models
Analytics are pulling out an incredible amount of data. You can know who is doing what and what led them to do so.
If you don’t know much about the various Attribution Models, I recommend that you read this article: Multi-Channel Attribution Modeling.
Attribution Models are incredibly helpful: they help us determine the traffic sources’ role in each conversion. However, weighting conversions – as they do – is shortsighted. They don’t take into consideration the countless factors leading to a conversion.
Analytics have been made to take conversions into account, however they don’t focus on the big picture of how several conversions interact together. People sign up for newsletters, share content, and purchase later on. How are these conversions interlaced? Are they impacting each other?
The purchase journey is a complicated one because nowadays people investigate and seek more information than ever before. They look for info and compare prices, and purchase six months later. These behaviors are left out of analytics software with the lookback window (90 days max on GA). Such a setting shows how the current situation is nowhere near an accurate portrayal of consumers’ behaviors.
Data collection is focused on on-site events – you don’t collect what is happening outside of your website. Social signals (eg: Likes, Retweets), or Display Ads (except for AdWords), are simply ignored. Maybe the ad you saw on Facebook is the one triggering your visit…
Google added the ability to create custom models and it’s a great way to tailor attribution to your specific needs. However, the workload it takes to study your audience and understand their behavior is massive. On the other hand, small mistakes can damage the model and lead you to make incorrect decisions.
Assisted conversions are an important step towards a better understanding of how people are interacting with your brand. However, these models fail to understand the different roles of each channel in each conversion. For example, they understand that Twitter played a role in the conversion, but they can’t understand the exact value of its role. As a marketer, you don’t know what truly led to a conversion; you just know that a channel impacted it.
Attribution models are great but their evolution is slow. Technology can now do much more, and Attribution 2.0 should come into play quickly.
What do we need?
If Attribution Models were to evolve, what features would they have? What would help marketers to make more informed data-driven decisions?
What do we need? Smarter software powered by algorithms. They collect everything, crunch the data, figure out the role of each touch-point, and what led to the conversion.
With the amount of data that you collect every day, comparing several purchasing behaviors is possible. You should be able to look at each purchase journey and find out which one is the most effective.
Here are some examples:
- Blog + Facebook + Purchase: Conversion rate 8%
- Blog + Newsletter + Purchase: Conversion rate 15%
Being able to understand what is truly having an impact on conversions will be an incredible step towards better data-driven marketing.
Analytics will also understand that different content leads to different actions and therefore allows you to better understand which kind of content you should produce in order to reach your goals.
When people buy products or services, their engagement with brands accelerates as they get closer to the moment of purchase. Attribution Models should take that into consideration and identify the elements that led to that particular increase in engagement.
It’s important to understand how your activities are impacting consumer’s position on the funnel. Algorithms will be able to determine which activities push users through the funnel and how to model the transition from each stage. You’ll be able to invest more resources into some specific activities and move more prospects from one step to another.
Analytics will also include users’ data to draw conclusions about who should be targeted and how each demographic can be more easily reached. You could, for instance, discover that 18-25 convert better when they come from Facebook.
To sum it up, here are the main ideas:
- Automatic weight attribution
- Compare different buyers’ journeys
- Identify the content leading to a purchase
- Understand the journey as a whole (not just the last 90 days)
- Consider how micro & macro conversions interact together
- Consider buyers’ engagement within all touch-points
- Stage identifications and how to push users further
- Demographics analysis
Having the ability to gather all this information could enhance the way marketing works and help us to design better funnels. Marketers will know exactly what led to the purchase. They’ll be able to replicate their findings to fine-tune their marketing activities.
Are We There Yet?
As avid readers, you are sensitive to the data-driven approach. For you, data is now one of the pillars of marketing. However plenty of startups don’t include data in their decision-making process. People who don’t reach that point will most likely lose business to those who did.
As consumers’ journeys become increasingly complex, marketing professionals are starting to touch the limits of current Attribution Models. We look at alternatives but unfortunately, they are not as good (and cost-effective) as we want them to be.
Complex pieces of software are getting there. They are doing a great job at correctly attributing conversions and learning about what drivers lead to conversions. However, these products are limited to the big players…
These big players – advertising agencies & corporations – are paying big bucks to access systems such as Analytics 2.0. They have an incredible amount of data that they need to make sense of in order to improve the effectiveness of their marketing activities.
In 2014, Google bought Adometry – which specializes in attribution solutions – and tried to help marketers to make better decisions. They integrated many features on their premium plan, which is not affordable for startups at the hefty price of ~$150k per year. Yet we can expect Google to include these features in their free plan as they see the demand growing.
Are we there yet? The major players have access to some sort of system. Startups are still not there yet and we can assume that we will need to wait for several years before having access to this type of advanced software.
In 1980, 1Gb used to cost $193k. It now costs $0.07. We can expect attribution modeling features to follow the same pattern and to see their price become quickly accessible to everyone.
Analytics companies will start to develop solutions for smaller brands and target smaller players to enable them to make sense of their data. Imagine that algorithms will soon be able to model the customer journey and analyze each event independently. You could run scenarios and mimic users’ behaviors “What happens if you remove this touch-point? Reduce / Increase budget?”
Let’s keep the dream going… They’ll help you plan your marketing strategy and automatically adapt the budgets depending on the impact of all touch-points. They’ll never be able to replace real human beings (who have strong insights about their industry and consumers), but they will be able to help you in your everyday job.
Preact or Framed.io are companies who are leveraging Big Data for Customer Success and Churn reduction. They predict the future by looking at current behaviors. Powerful systems are already here, we just need someone to build them for more general analytics.
Interesting days are ahead in terms of Attribution Models. Even as I write words I feel like I’m dreaming, but the future is coming quickly and we’ll soon have better, more accurate Attribution Models.
Although you might think they have been around forever, analytics are still quite young. We are in an era where technology transforms everything, and that is what helps us to become better Marketers.
I strongly predict that analytics will become increasingly efficient and will soon include more data than ever. Algorithms and Machine Learning will help us to make better sense of all this data and foster competitive advantage.
The future is full of opportunities for better analytics & marketing. You’ll soon be able to better measure users’ behaviors and to tweak your marketing strategy to a greater extent.
Interesting days are ahead. I’d like to hear your ideas now. What do you think? Are analytics going to move in that direction? What features would you be interested in?