Watching the CXL’s course on Attribution, I was both worried and happy at the same time. On one hand, the concept of “attribution” seemed vague and very “high tech” to me. But the truth is, as daunting as it maybe – attribution is nothing more than using numbers to inform your decisions.
In other words, attribution is the art of making informed decisions from numbers. Some people say, if you get attribution down – you’re 1% of the marketer in the world already. Not so sure about that, but the concept of attribution has definitely helped me to see the bigger picture. All the data that we spend analyzing, how does it really come together to make things better? – the answer attribution.
What is attribution?
Well – let’s put this into terms. Say you’ve spent more than $1million dollars on advertising every month (I know right, that’s a lot of money), but how do you know which channel – search, paid, affiliates is the real money driving engine? Sounds easy?
Great now let’s add some twist – let’s say all your customers don’t buy from their single visit to your store online. In fact, they see a advertising on fb, come over to your website browse and leave. Two weeks later, they search for your brand on Google and click to enter your website and leave. Finally one week after, they directly plug your website into the browser, come into your website and make a purchase – so here is the million dollar question – which channel helped you seal the deal? Fb? Google PPC? Direct?
Find this confusing? Well the reality is – this is how most customers purchase. In fact, the average consumer visits your store 4 times before purchasing. In this complex multitouch point era – having the right attribution to connect the dots to measure effectiveness of your spends is key.
Is attribution just for online?
Although online attribution is easier and more convenient these days thanks to the whole boom in online analytics for example in Google Analytics or Hotjar. But in reality, attribution can work for offline things too for example tv shows – how do you know which commercial worked and which didn’t to drive up sales? There are tons of services and companies out there these days that provide analytics services to answer these questions.
A slightly different example is POS register data to be use to analyze customer journey and decision making. Imagine if you knew exactly how moms on weekends tend to buy Kleenex at 11am – wouldn’t you move all the Kleenex to a more accessible area or sell in bulks to drive up sales? That’s the power of attribution – where you put together data from different models to drive decisions.
Key models in attribution models
There are several key models online on how attributions are used to measure marketing effectiveness:
- Last non direct click – this is when the last click minus one gets all the credit for driving sales. For example, say someone saw your ad on fb, clicks and leave, then search for it online, and finally come directly to your site two weeks later and buys some. The last conversion action is DIRECT to your website, the one before that direct conversion is searching it for google – in this model google search gets the attribution figure.
- Last interaction attribution – is where the last action before the purchase or conversion happening is given the most credit
- Linear attribution – that’s when all the customer touch point gets equal credit. If a customer sees your fb ads, clicks on google ads, and directly search your website – the $100 earned sales get attributed equally to all three channels.
- Time decay – takes into consideration the timing of the purchase. The more recent the step leading to purchase of the item, the more credit the step gets. In other words, a click on fb ads three weeks ago gets less credit compared to searching for your website on google last week
- Position based attribution – is when the first touch point and the last touch point gets the credit.
- Custom attribution model – is when you craft your own model from scratch. Meaning putting together different aspects of the journey to effectively predict and attribute success of the marketing. Often this includes using machine learning models and algos to help. A lot of marketers say this is the ‘gold standard’ of attribution. To make a model that is unique and custom to you.
What are some key questions you should ask prior to going on the attribution journey?
- What granularity of data do I need? – there is going to be a lot of data that you have to sieve through to make decisions so pairing the right level of data is going to be key. If you need data in an event level vs campaign level – then carefully crafting the web analytics team to fit that need is key in going on the attribution journey
- Do you have the competency to do it? - are they people who are competent and have the necessary skill to do attribution for your company?
How to apply attribution in your business?
Customer journey analysis – knowing which channel and landing pages that is likely to convert we get to start emulating key successes across different channels. More than that, knowing that we take for example 5 steps before purchase happens and figuring out ways to reduce that to 4 steps is where the value of attribution can help drive conversion for a business.
So now that you have the all the right understanding of attribution models – how are you going to apply it?