Measuring cross channel digital marketing is basically fairly simple. You can produce loads of data form digital elements and draft a figures in excel. Getting a complete view about marketing and how well it contributes to sales is a bit more complicated task. To have the view all the time, not just after extensive excel exercise can be truly complicated. In this article I’ll give an idea how to model your marketing. You should keep the model as simple and straightforward as possible.
There is a good way to organize your channel presented by Steve Jackson in his book ‘Cult of Analytics’. Channel is divided in bought, own and earned channels:
Bought channel is a channel where new contacts/leads from third party sites are sought with money in your hand. This covers banners, bought keywords, affiliates and so forth.
Own channel is your own ‘real estate’ such as own sites, microsites etc.
Earned channel is the one you have little or no control over. Mainly social media, however this covers public media too.
A good question to ask form oneself is where does your Facebook pages belong to?
The division portrait above is vertical. In addition you can divide your channel horizontally to reach (sometimes term acquisition is used), engagement, conversion and retention (descriptions are pretty obvious):
Reach (Acquisition) shows you how well your message reaches people. These are banner impressions, searches with your keywords, visits or message reads in own channel.
Engagement shows how engaging your message is. These usually are banner / keyword clicks, pages per visit, article shares and so on.
Conversion is the easiest one; you have enticed people to do something that you wish them to do. Buy something, download, and make a support ticket on-line instead of calling.
Retention level tells you how loyal your customers are. What is their time between purchases, average order size, new customer orders per existing customer orders and so on.
You might notice that the leves are named differently than in this community, however don’t let it confuse you.
The examples mentioned previously are measures. As you progress deeper to the channel the measures are transforming more to key performance indicators (KPI). These indicators are always calculated numbers, typically relations (e.g. number of visits is a measure, new visitors/returning visitors is a KPI).
Now the digital channel can be presented as three by four matrix. When you put a KPI in each cell, you can see how effective your marketing is. You should select your KPI’s carefully. One of the biggest mistakes is to select loads of KPI’s (more the better, right?). Few KPI’s (one to three) in a cell should be enough. The general guideline is to show numbers that you need to react immediately upon change. KPI’s without actions are useless
In order to calculate return on investment you need to be able to divide revenue to all levels in your marketing structure. Dividing your channel as presented should ease this. You can give revenue to levels roughly in two ways. Attribution model or split revenue with all elements participated to sales.
In attribution model each level has a weight coefficient that determines which part of the sales revenue is given to e.g. banner. Determining the coefficient value can be somewhat difficult, especially if campaigns differ a lot.
Splitting revenue evenly with all elements that have participated to the sale is more straightforward approach. In this model revenue accumulates to all elements in all levels. Accumulated revenue is naturally a good KPI to report on each level (even in each campaign).
You can use either of these, I prefer straight split myself.
Congratulations, you have modeled your digital marketing channel and it looks beautiful! It’s time to lean back and look money pouring in, right? Unfortunately reality has a nasty surprise in store for you, my friend, off-line activities and sales. These are missing from your model entirely. Reach and engagement could be somehow handled (and maybe retention), but conversion gives you real trouble. To be able to determine what drove customer to buy, banner, blog article, magazine ad or something else, is close to impossible. So you can’t divide revenue to other levels because you don’t know which elements participated in conversion. My advice is to resort statistical methods. For example you can calculate a correlation between online and offline sales and use that to divide revenue between channels. It will be more like attribution model, but you can get idea what coefficient to use form your digital channel model. The simplest way to do the split is to split in a relation of online and offline spend. Straightforward and simple, however you need to check if it works for you.
Modeling your channels can be extremely complicated. Models are simplifications of real world phenomena. The model won’t be the exact truth about marketing for all businesses, but it will give you ‘good enough’ view on your own marketing. Having this view will help your planning, budgeting and overall comprehension of your channel performance. It will help you, as marketing professional, to speak credibly same language in boardroom with other C’s increasing your business credibility to a completely new level.
This article was originally published in Digital Doughnut (by Me)