Data-driven marketing. Stop pretending you know.

So many marketers use the term ‘data-driven’, but do they really understand what it means? Are you able to use data to formulate your strategy? Read on to find out what a data-driven stratgy looks like and how you can build one.

Read on

What is data-driven marketing?

At the moment, I think it’s safe to say that marketing in general revolves around being data-driven. But what is it? I’ve been in board meetings with directors of enterprise level companies where talking about attribution and being data-driven is as we’re talking in a completely different language throughout the entirety of the meeting. Why is it so difficult to understand attribution and the effectiveness of marketing?

Everyone’s a pro.

If you have an opinion, about anything, it doesn’t mean it’s fact. Opinion isn’t fact, let’s stop thinking that because we have an opinion, we’re correct. Instead, let’s start embracing the very thing which can give us what we need; data-driven marketing.

Recently, we’ve been working on building an attribution model to show the effectiveness of marketing. It’s ground-breaking what we’re doing. Literally being able to attribute a value to each marketing channel. Meaning, we can say £X marketing spend will get you £X ROI, in what ever way your business wants to know – selling bananas? This model will tell you how many bananas you’ve sold through each marketing channel, down to the very word that the user searched.

What seems to be the problem, sir?

The problem comes with educating stakeholders on this sort of activity. You see, holding a mirror up to someone and showing their true self isn’t an easy or nice thing to do, but often it’s needed, especially when you’re in the dark with information. So how do you educate and translate quite a complex marketing attribution model to a language your board will understand?

In my experience, having one-to-ones with key individuals makes for a better journey, or at least start of the journey of educating relevant stakeholders. When you do that, you’ll have the time to properly walk through everything, go through any questions that individual might have and so on. You see, marketing attribution isn’t just for the marketers, its for the business to see what the current state of play is. The purpose isn’t to do some fancy stuff that isn’t actionable, but to truly see what your customers are doing, how they’re purchasing or what advertising messages just aren’t working, and which ones are.

So what is attribution modeling?

Get all your data in one place. Whether that’s organic, PPC, email, display advertising and so on, all at an individual level. Using sophisticated techniques, identify all individual customer journeys. You can do this by recording the sequence, timing and details of each successful and unsuccessful path.

Once you have all of this data in one place, you then need to model it. To model your data, you apply mathematical techniques such as Random Forests, which uses machine learning to predict the possibility of success for each journey. You then need to break the journey out into sub-sets and calculate how successful each journey would be without a specific event. For example, what would happen if we removed an email from the sale of the journey?

At this point, apply another mathematical method – the Shapley value approach –allowing you to reach your goal of ‘attributing’ a fair share of value to each touchpoint, not just the first touch or the last touch of a channel. You’re then able to determine your base sales e.g. those you would’ve made without any of the marketing activity we measure as part of our attribution modelling. What you need to remember however, is with each successful purchase and positive experience, your base sales will increase.

Once this has been achieved, you’re able to see your marketing activity and make informed decisions to optimise your marketing efforts in turn getting better ROI.

Insight isn’t useful if it isn’t actionable.

So what do you do with the insight you’ve gained through marketing attribution and how do you create a strategy? It’s great seeing how channels perform, what your customer’s life time values are and how each channel works with one another, but how can you action it?

Once you have all of your data in a way you can digest it, the next part is to create segments. You want to use the data you have to help inform about who your consumers are, then you can create your segments. Usually, you’ll see the likes of highly active, active, low, infrequent and dormant customer groups. Your marketing attribution model will tell you how your consumer segments are moving, which will determine the effectiveness and ROI of marketing campaigns.

Get your data into a weekly or monthly dashboard. Now, a lot of key stakeholders will want to see analysis even on a daily basis (I’ve had that question), but that’s just not reasonable. You need to communicate that seeing attribution on a daily basis isn’t going to give you the right insight to execute your strategy properly. It boils down to a fairly simple explanation. See activity as events, if an event happened in the month of June, that could have an impact on a customer purchase in the month of September. So if you’re reporting on a daily basis, how are you going to determine what events have made an impact in what month and how each customer moves through the segments?

It also has something to do with life time value (LTV), you can’t determine the LTV of a customer based on a day of activity, you need to monitor their activity and measure their value based on a time period that you can gather enough data to tell you what their true value is. So inform your stakeholders, be as helpful as possible and encourage monthly (weekly if you must) reporting with quarterly deep dives to properly go through your attribution modelling.

If you need to show reports on a daily basis, you could always set up an online dashboard which shows top-level analytics you can obtain through the likes of Google Analytics, AdWords and so on. That way, you have something to show. Just make sure you caveat the activity and that the people your giving the dashboard to don’t take last-click as the true measurement.

Can data help brand?

I’ve had many a conversation with marketing professionals, believe it or not, about whether brand can be measured or not. You’re probably wondering why I’m even highlighting it, because it obviously can be, but I do think there is a misconception which can be addressed, or at least the opinion of a digital marketing specialist (not typically brand).

Often when I talk to people about brand, they usually think it’s something along the lines of being fluffy marketing rubbish. I can’t deny there are a lot of marketers who lean toward to more emotional side of marketing when it comes to brand, but needless to say it can all be measured. It’s fine having a ‘fluffy’ brand goal, just use data to back it up.

The sort of analytics you want to look at and monitor for brand are; brand awareness, purchase intent, brand reputation, satisfaction, value and brand quality. There are more things to measure, for example you can use a tool such as Brandwatch to monitor how people talk about your brand.

Brand awareness and purchase intent are two brilliant measurements to put together. Think about it, if you can pull your brand awareness data across the UK (or the country you’re based in) and overlay your purchase intent data, you’ll be able to see the locations with high purchase intent and low brand awareness, high brand awareness and high purchase intent and so on. This will formulate your data-driven advertising strategy and give you tangible insights to back up the activity you’re doing or proposing to do.

Get in touch

Do you have a project in mind? Get in touch and we’d love to help. Tell us what you’re looking for, the timescales and any other information you think could help us, and we’ll help to achieve your objectives.

12 + 5 =