In this series of articles I’ve been talking about the 5 steps of digital marketing attribution. The last article went over Valuation as it applies to marketing channels, and this built on the articles about Attribution modelling, Pathing and Classification. For this final article I want to give you a brief overview of how to make this information actionable within your business.
Before we get going it’s important to add a heavyweight disclaimer and say that all businesses are different, and the precise steps you need to take to optimise your marketing will be entirely dependent on your own business. That means this article should not be treated as a proscriptive approach to marketing optimisation. Without understanding the details of your business it’s impossible to suggest how to improve it! The only thing I can say for sure, is that if you follow the first 4 articles in this series, and then don’t do anything else, you’ve wasted a whole bunch of time.
With that disclaimer out of the way lets talk about optimising our marketing channels. There are a few key concepts that we need to get out of the way first. Let’s start by recapping a few measures that we can use to evaluate marketing performance. Probably the most familiar term to a digital marketer will be CPA/CPC/CP?, for business folks it’s going to be ROI or the more marketing focused ROAS.
In the last article we talked about a number of different types of conversion (lead generation, account creation, individual purchase). Once we’ve picked a metric that we want to measure, we can work out our cost per conversion using the following simple formula
CPC = cost / conversions
This is the most common metric people use in online marketing but it misses out on a fundamental element of the conversion; it’s value. It’s good to know that an advertising channel is working at a CPC of £20, but if each of those conversions is only worth £1 then you have a problem.
ROI stands for Return on Investment and is commonly used across all aspects of a business to evaluate business decisions. A lot of digital marketers prefer to use the a metric called Return on Advertising Spend or ROAS. The formulas are similar and can both be used for fine-grained or course-grained analysis.
ROI = 100 * (revenue - costs) / costs ROAS = revenue / spend where revenue is the return from a campaign, spend is the cost of the campaign, and costs includes other spending within the organisation such as agency fees
Let’s look at the difference in these numbers for campaigns A, B and C which all cost 10K, but returned 5K, 15K and 25K worth of business, with additional agency fees of 5K.
ROI A = 100 * (5-15) / 15 = -67% ROI B = 100 * (15-15) / 15 = 0% ROI C = 100 * (25-15) / 15 = 67% ROAS A = 5 / 10 = 0.5 ROAS B = 15 / 10 = 1.5 ROAS C = 25 / 10 = 2.5
In my view ROI is much clearer; any marketing campaigns that fall below an ROI of 0% should be canned. ROAS is simpler to calculate day to day and can be used to optimise campaigns within a single channel. But it’s usefulness is limited when you compare different types of advertising channel. In this case fixed costs are likely to be different, and of course some advertising channels don’t require advertising spend anyway.
Imagine the scenario above for an SEO team where they’ve created some campaign web pages to tie into a major event. The ROAS will be *infinite* in all cases, as there is no spend. The ROI will give you an honest picture, and a value you can compare against other channels.
There are two main types of channel optimisation that you should be doing in your business. The one you’re most likely already doing is within-channel optimisation, or intra-channel optimisation for want of a snappy name.
For single channel optimisation we’re effectively considering similar advertising approaches with different vendors. Let’s go back to car examples. Imagine that I work for BMW and I’m advertising on two websites that review new car releases. I can do some research to estimate the audience that each website gets, but ultimately I need to spend some money and measure the marketing performance to work out which website appeals most to my target audience.
For most companies this means looking at the number of clickthroughs that I get for each website. You can obviously do much better than CPC by looking at leads captured, individual conversion value and lifetime value.
The nice thing about this type of optimisation is that the display banners are the same, and the channel behaviour should be similar between equally successful websites. If you register a different level of return then it’s relatively easy to modify your spend in response.
So this is the grand goal of cross channel attribution, and it’s not the easiest thing to get right. I’ve spent a few years looking at cross-channel attribution statistics, and channels can show results in very different ways when you start comparing them.
You do tend to see the outliers fairly clearly though, and that’s where you should start with optimising your marketing. Of course, the first thing to do is validate what you’re seeing in the model, as most often the dramatic trends tend to be data issues. Once you’ve got that out of the way then you need to fully understand the behaviour of the channel in terms of where it acts in the sales funnel, the type of leads being introduced, etc..
Eventually you’ll feel confident enough to modify your spending on a particular channel, and then you need to monitor the impact. It’s worth bearing in mind that many channels have strong interdependencies, so a boost in spending on one channel may lead to a boost in attribution for lots of other channels. Similarly a drop in spending could also have unforeseen consequences!
Overall though, it’s better to monitor this and change your spend, than to sit with your eyes closed and your fingers crossed. Good luck.