Let’s talk some harcore nerdy stuff for the readers that like new types of information by reading more than 3 minutes instead of looking at funny GIFs 🙂
Together with our super cool data analytics partner Haensel AMS we took the last two years to set up our own data and dashboard system for our clients. For this, we use a dashboard tool that includes all online data sources and is based on a tailored attribution model, that Haensel AMS has developed.
Let’s begin with writing down what Google says actually is an attribution model:
An attribution model is the rule, or set of rules, that determines how credit for online sales and conversions is assigned to touchpoints in conversion paths. For example, the Last Interaction model in Google Analytics assigns 100% credit to the final touchpoints (last-click) that immediately precede sales or conversions.
And to add to this, these sets of rules are predefined and standardized to ensure that it’s one-size-fits-all. Let’s be honest; for most brands this appears to work pretty well, right?
However we agree to disagree here. Why?
Well, because of:
- Most european brands we work with are very well known in their home country, but not (yet) in the German market. The German market is fiercely competitive. Brands tend to spend a lot more media budget than in most other EU countries. This means you can easily burn most of your precious budget without having had any impact on your sales;
- Nowadays most Ecommerce managers have to make spaghetti out of too many channels, data sources and even different attribution models which can become somewhat complicated to manage and react in a fast way;
- And this is especially the case because Google and Facebook, being the biggest players, use different attribution models to make it extra difficult to compare data or results;
- Last but not least we had the hypothesis that there is no standardized way of allocating a sale or conversion for the simple reason that when you buy, for example, a bag of dog food (and we happen to sell dog food) the customer journey is different versus when you want to buy a €500,- mattress (and we happen to sell mattresses).
So what did we do?
For each of the brands we are working for, we started with collecting lots and lots of (GDPR-proof!) website event data. Here are the most important variables we track, weigh and value:
- the time between customer touchpoints, e.g. a visitor clicking on a Google Search ad which leads to a webshop visit. Two days later the same person is visiting the same webshop through an email campaign.
- the type and amount of actions on the specific webshop during a visit. Here you can think of the amount of pages viewed, whether the potential customer subscribed to a newsletter, whether he/she clicked on a product, and went to the check-out page, etc.
On top we further enriched the website traffic data with information from platforms such as Google Analytics, Google Ads, Facebook, ERP´s, etc. These third party data sources are integrated with API connections and automatics daily/hourly downloads.
We don’t want to give away the secret Coca Cola recipe to our model, however what we can say is this results in finally seeing fact-based which channels really matter in the three interaction phases (Initializer, Holder, Closer) for your brand and at what point in the customer journey they do (We keep in mind that most brands have different seasonalities, product launches, shopping events, etc).
On top of that we managed to pinpoint visitors that use both their mobile and other devices in the same customer journey, which means we have pretty accurate data that we can really rely on.
Why is all this pretty amazing?
We e.g. see in our portfolio that for an average customer it takes up to 28 days to buy a product. In this time frame, they visit the website up to 10 times and show different kinds of behavior every visit. And you can do A LOT with this info!
This is what our Head of Ops and BI Juan says, we can do with this kind of information:
- We can decide to invest the lionshare of our precious media budget sometimes up to three weeks before a big moment coming up for the brand, for example the launch of a new collection or a big sales campaign. We can allocate a big share of media budget in channels that are normally undermined by default attribution. Yes that feels like risky business, however the numbers tell the tale. The much needed proof for “AIDA” or “Touch-Tell-Sell” approaches than can easily feel like burning money at the wrong moment
- We can reliably predict behavior and build so-called bottom up forecasts (starting at channel level) which means we can place safer bets with mostly higher media budgets over time
- We have insights on difficult-to-measure channels like influencer marketing, brand collaborations or even PR.
- We can collect all this input of different data sources (our own website tracking, GA, Facebook, Influencers, Affiliates, Content partners, etc.) and visualize them in one dashboard tool (right now we are using PowerBI and Tableau for different clients). This means that every stakeholder -from the CEO up until a specific marketing specialist can use the same dashboard, look at the same data and compare apples with apples.
What´s next? According to our Managing Director Radboud Langenhorst it’s all about constantly improving the data quality and understanding how to best work with this data for our clients. Also with this tool we are onboarding a lot of different stakeholders into e-commerce and letting them see the true chances and opportunities through all the costs and risks.