How do marketers measure the impact direct and indirect marketing channels have on the success of their campaigns?
At a recent conference organised in London by the Institute of Fundraising we asked an audience of around 70 people, all of whom work for charities, whether any had developed a reliable view of how different aspects of their marketing spend impacted their donations. Not a single person present was able to say yes.
In the distant past, before the internet had been invented, and when mail was the only direct channel, it was a whole sight simpler; all you needed to do was to create control groups that you didn’t mail, and then measure how they performed compared to those that you did.
But in today’s multi-channel world, measuring the effect of direct and indirect marketing channels is a problem of great complexity.
Our concern is that, because of the number of different channels and influences that can precede a customer action, like making an order, marketeers may have, to a large extent, given up.
But to do nothing leaves us with a £26bn per annum unanswered question just for the UK alone.
Clearly, there are in fact two very different questions to answer:
- First how to infer the effect of non-direct media like outdoor advertising and most of TV
- And second, how to measure the effect of those channels that are direct like Google PPC, direct mail or Facebook?
Measuring direct and indirect marketing channels
Most practitioners who want to measure non-direct channels use some kind of time series modelling, and this works reasonably well when we are just looking at summarised data, such as the overall sales value of an organisation in January.
But there are big limitations in that it’s relatively expensive to develop the models, they cannot get into the detail of campaign performance rather than looking at aggregated channels, and they rely on the advertiser varying the amount of spend each month in each channel.
When looking at the outcome from a time series model there is also always a large proportion of sales whose cause cannot be explained by the model, and this has to be assumed as being due to the influence of the brand.
However, where marketeers appear to have thrown in the towel unnecessarily, is in respect of measuring the effect of direct media, looking at online and offline channels in combination.
We don’t believe that any organisation selling goods or services to identified individuals, such as home shopping companies or travel or financial services to name but three, has to give up on measuring the impact of their direct channels.
But to do this measurement one needs to work back from each order, rather than forward from the spend in each channels, to unravel what is actually going on in the real world.
We approach this by looking at all the known interactions between an organisation and a customer in a 90-day window before an order is received.
We ignore all clicks, opens, opportunities to view etc. etc. indeed anything that cannot be directly related to an actual order event, and treat these just as noise.
What then comes to the surface is much more complex that any last click proponent would like to admit; we find ourselves looking at a unified view in which emails, PPC, social, natural search, mobile SMS, direct mail, OBTM and any other direct channel employed can each play their part.
This table is a real example of just five individual orders received by a home shopping company, and counts the times each different channel played a role in the 90 days before the order:
| Customer type | Catalogues received | emails received | Google PPC | Direct entry | Phone in | Total |
| Existing | 2 | 3 | 1 | 1 | 7 | |
| Existing | 2 | 1 | 3 | |||
| Existing | 8 | 2 | 10 | |||
| New | 3 | 3 | ||||
| New | 3 | 13 | 1 | 17 |
Even in this relatively simple example the first thing that become apparent is the wide diversity of the routes taken by customers before they actually placed their order.
The good news though is that once you have joined the online and offline data together, and considered the weighting to give to different channels, and to different time intervals prior to an order being received, you have here the solid building blocks for attributing actual value to the channels being deployed.
You can then allocate the value of each order across the channels that influenced it, and end up with an overall value contributed by each channel in a particular time period.
We are not suggesting that this is rocket science, but it does need attention, and technology, to make it happen.
We have built the technology to automate this kind of attribution, and would be interested to discuss it with you if you felt it could help.
Please CLICK here for a short PowerPoint explaining in a little more detail how we do it.
UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.
Our technology focus has been to develop UniFida. Our data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.
Marketing is changing at an explosive speed, and our ambition is to help our clients stay empowered and ahead in this challenging environment.