How to truly understand the impact and uplift of each your channels
After helping a client review four best of breed marketing mix modelling consultancies, it was clear that there is not a lack of attribution approaches. There was a lack of data. WebFusion from Fusion Analytics captures the customer journey, linking both online, offline and time-series datasets. With this full picture we can use simpler analytical methods, across different channels, which can be provided back to you, your attribution consultancy, or analysed with our support.
Why is this important?
Digital has made it so much easier to create a new channel or platform, and quickly gain a significant audience. These platforms can influence the customer journey at different stages:
- They can genuinely create new audiences and “cold business”
- They can re-target existing buyers or site visitors to up sell to existing customers
- Or they can support the conversion of current warm prospects
As the number of channels increases, it is becoming harder to understand the impact and uplift of each channel. Furthermore there is generally not one source of data that can explain all channels. This makes it much harder to evaluate and apportion marketing budget.
What methods are used at the moment
There are various methods to attribute a sale back to a marketing channel, activity or spend.
For DM, EM, most digital channels, Door Drops or Inserts (via a response code), it is possible to look at the total activity and then link the individual that was sent the activity to those that responded. For example, if you sent an email did that person click a link and respond/buy straight away? This type of tracked campaign is called “Micro” attribution.
For FMCG, and brands that use tools like out of home, or TV advertising there are some ways of attempting “Micro” attribution. For example, you could look at the number of people that visit a website within 7 minutes of a TV campaign, or those that respond via a dedicated phone line. (See our secrets pages for other methods brands use to asses performance – such as through the use of surveys).
However, not everyone is sat in front of the TV with Google (or a pen and paper) at the ready. Therefore for most cases businesses have to use time series modelling methods. These compare what an expected sales level would be in a given time vs actual sales (which should be higher). This type of modelling is called “Macro” modelling.
Another common method for getting marketing mix insights, for digital channels at least, is via tools such as Google Analytics. Google Analytics looks at the tracked digital sources, such as social media vs affiliate sites vs search, and then assigns the conversion or sale to either the; first/last source, or by weighting it across the different channels that person used.
For every new channel, there is typically another “Micro” attribution approach to consider, plus how this will fit into the overall marketing mix. For totally new incremental business, the added value can be clear. However in most cases channels work together. Without a clear idea of how groups of people are responding to a specific marketing mix, analysts often have to rely on sifting through a series of disparate reports, making a series of assumptions; or expected values.
“If you torture data long enough it will confess”
Ronald Coase Nobel prize in Economics 1991
It does not take too long for this type of disparate reporting to become unmanageable, with so many assumptions the truth becomes diluted. Also, from a practical perspective, it is not possible to tie the different approaches together as the data is not held at the individual level.
In the case of our client reviewing four different providers, whilst each provider had an almost identical overall modelling approach, each had to create their own analytical approach to understanding the complex digital landscape. The approaches varied from; nested models, neural networks, (quite honestly by the provider) making a set of assumed values.
By tracking the whole customer journey, combining online and offline data sources, WebFusion from Fusion Analytics hopes to provide a single source of data, for understanding marketing effectiveness. The enhanced data set allows analysts to replicate any approach whilst using actual usage data rather than relying on too much analytics.
The result of this is that, we can:
- Provide a real time marketing mix including online and offline channels
- Take any given channel and review their “micro” reporting solution in the context of the overall marketing mix
- Allow third parties to include our data in their attribution methods
See what our MD Howard Thompson had to say on the subject of attribution at MTEX, by downloading the presentation.