Not One Single Attribution Model Gets It Right. This Is How To Fix It.
Remember the time when we spent our advertising budgets without any worries about attribution, and just trusted the gospel preached by Google and Meta?
What a wonderfully carefree time. And how woefully naive.
Because if there is one big upside to privacy laws and cookie regulations having changed so significantly in recent years, it’s that we started to really think about attribution.
Oversimplified online attribution models (last click, anyone?) are no longer unquestioned. And that’s a really good thing.
“HALF THE MONEY I SPEND ON ADVERTISING IS WASTED; THE TROUBLE IS I DON’T KNOW WHICH HALF.”
The most important thing is to understand that no single attribution model or data source will tell the whole story.
First click, last click, lineair. They’re all wrong.
So that’s why we need multiple data sources and different types of reporting so they can together tell a story you can actually learn from.
Combining Multiple Data Sources
We use 5 different types of insights and data to put a sensible picture together.
First off, there’s the reporting of your advertising channels. Meta and Google track the behaviour of the users that aren’t cookie shy, and they attribute conversions accordingly.
The issue is of course that these channels can only attribute value to the part they play, not anything else. That’s not surprising, but problematic nonetheless. And of course behaviour of those who opt out of tracking isn’t in the picture either.
Still, this in channel reporting can be very valuable. Not only does it provide direct feedback about your ad creatives, but it can also work to prove campaign effectiveness. An increased CTR in search during your radio campaign could for instance provide a first clue in your campaign analysis.
Market volumes, traffic and sales
The second thing is to keep track of market volumes or search volumes. They will show when an increase in interest in your brand or product category might occur.
Then there’s user behaviour for your website or store. Are you seeing extra visitors among a target segment you’ve been after in a recent campaign? Or an increased interest in a type of product you’ve been promoting? These are all things that have their place in judging your advertising effectiveness.
Obviously you don’t wanna leave out your sales figures. On their own they are often hard to relate to your marketing spending in a very direct sense, but you can’t do without them if you want to get close to a complete picture.
Qualitative Data
Finally, there’s qualitative data taken from interviews and surveys. A good example in terms of judging advertising effectiveness are acquisition surveys. So upon purchase, you wanna ask your customers how they’ve come to learn about your brand, and what made them consider buying the category.
Of course, this is not the most accurate source of information (more about that here). But remember that the main point is that none of the data sources listed here are completely accurate. That’s exactly why you wanna look at all of them combined.
The added value of an acquisition survey is that it can report on marketing channels that otherwise would never come up in any other reporting or attribution model. Think of free publicity, business partnerships or fairs and events.
A survey will never teach you how big of a role any of these activities have played exactly, but it does confirm that they in fact played a role to begin with. And without anecdotal evidence like this, you’d never have any proof of the value of such channels or activities.
Long and complex sales cycles
If you’re dealing with long and complex sales cycles I probably won’t have to tell you why it can be helpful to combine these 5 different sources of information. Perhaps you’re selling software to businesses, or maybe you’re selling holiday trips to consumers, like I am.
It takes months to get from an initial idea to a booked holiday. A multitude of options are considered by different people, each using multiple devices as well as offline means. There’s no way your online attribution model is ever going to even come close to painting a realistic picture.
Your happy go lucky traveller might have seen online advertising, downloaded a brochure, got some emails, then visited the website again multiple times (of course clicking on search ads on their way there). Afterwards they will be retargeted with display, video and social. And who knows, maybe they just happen to follow one of the influencers you hired to address any final doubts. And of course they’ve seen at least one of your ads on TV or on the street, and they’ve heard a few on the radio.
A first click model might attribute the booking to Meta, while last click might say Google brought in the sale. Both of course would be far from the truth. Between these two touch points, you’ve spent a whole lot of additional marketing budget and marketing fire power to secure this booking.
Oh and finally, you shouldn’t underestimate how external factors can play a role as well. You could have a conversation with your neighbours today that might spark months of casual dreaming about your next travel destination. Or perhaps it’s a TV show, or an article in next weekend’s newspaper, or a childhood memory. You’ll never know if you never chat to your customers.
SO, one SINGLE MODEL WON'T CUT IT
So if like me you can’t rely on econometrics or a really sophisticated mixed marketing model at the moment, you wanna remember one thing. There’s so much going on in the buying journey of your customer, that not any one attribution model will be of much use on their own.
By looking at a range of different data sources and types of information, you’ll be more successful in painting a picture of the marketing activities that play a role in the path to purchase, and of the overall effectiveness of your campaigns.
Tip: read more about how we analysed the results of our first above the line campaign which combined radio, out of home and online advertising. It offers a practical example of how we combined all these different types of reporting to be able to confidently draw conclusions about the positive business impact of our campaign.
THE SHORT VERSION FOR SLEEPYHEADS
There isn’t one single attribution model that will help you understand the effectiveness of your marketing mix or campaign.
Instead, you wanna use a collection of different data sources and reporting to understand effectiveness.
Combine in channel reporting with market or search volumes, traffic, and sales figures, and add to that some qualitative insights.
Stop looking for a perfect one size fits all attribution model, and start combining all these imperfect ones to make them work together.
Find out how we did it for our first above the line campaign.
