Data-driven marketing has a complex history. Like many marketing terms, it’s been overused, oversimplified and diluted to the point where it’s almost meaningless.
And that’s a shame, because the concept behind data-driven marketing is essential – for brands of all sizes.
In this article, we’ll explain what data-driven marketing is and how it can benefit your marketing programs (even if the term itself makes you cringe a little).
What Exactly Is Data-Driven Marketing?
First, let’s clarify what the word ‘data’ means. It’s not just numerical values on a dashboard. Instead, data are any facts or values that you can collect, catalogue and classify.
That could mean website traffic numbers from Google Analytics – but it could also mean qualitative feedback from customer interviews.
Of course, raw data are useless. To use it in your marketing, you also need to analyse it and distil actionable insights. Think of data-driven marketing as having 3 steps:
- Collection: what is the data?
- Analysis: what does it mean?
- Distilling insights: what should we do?
Now, it’s worth noting that most marketing decisions are already data-driven – it’s just that one of those 3 steps isn’t being executed properly. Below are some examples with the data source in bold.
- ‘Let’s do X because that’s how we’ve always done things’.
- ‘[Big industry leader] does X. They are large and successful, so X obviously works. Let’s copy them.’
- ‘My son/daughter does X. Gen Z/Alpha are the future, so everyone will soon be doing X too. Therefore, we should make X a part of our marketing strategy.’
- ‘I don’t do Y, so our target audience won’t do Y either.’
Those decisions are technically being driven by data of some kind, but they’re all very flawed. That gives us the first requirement of good data-driven marketing: the data you collect must be comprehensive, accurate and, ideally, drawn from a multitude of similar sources (free from information and selection biases).
All data analysis is about determining causality – how an independent variable affects a dependent one. That’s why good data-driven marketing must also control for confounding variables (that is, be free from confounding bias). In other words, you need to do your best to understand whether X is causing Y, or whether Y is also being influenced by W or V or Z.
Here are some examples of data-driven marketing being compromised by different biases:
- You run an A/B test on a web page with a statistically insignificant sample size (information bias).
- You rely on a study about how the general public likes to consume information when marketing to C-suite executives (selection bias).
- You assume that shorter articles rank better on Google because your articles that target hyper-specific long-tail keywords rank better and also happen to be shorter (confounding bias).
Therefore, good data-driven marketing can be defined as making marketing decisions based on data that:
- is comprehensive, accurate, and collected from multiple sources
- has controls for variables implemented
- can be viewed and analysed by any stakeholder (which eliminates anecdotal and experiential data).
Note: often, people make marketing decisions based on a very different kind of data: what will be best for them, rather than their company. Think about the new CMO who initiates a rebrand to secure their legacy, or the team leader who criticises a strategy because they feel threatened, or the agency that cuts corners for cost purposes. ‘Bad’ decision-making isn’t always the result of a flawed data model.
The Benefits of Data-Driven Marketing
To make accurate decisions, you need data. That’s a fact. From the moment we’re born, we begin collecting and analysing information about the world to inform our behaviours. If a marketer tells you they don’t use data to make decisions, they mean they’re relying on previously gathered data rather than new data – for example, their prior experience or anecdotal reports.
And there’s nothing wrong with that when distilling insights from new data would be too difficult, expensive or time-consuming. Sometimes, you just have to run with a hypothesis and test it. Problems arise, though, when that hypothesis is wrong, especially if the decision in question is a major strategic one.
That’s why the ability to collect and analyse new, high-quality data is critical. It’s not because you can’t possibly make good decisions otherwise. It’s because good data-driven marketing de-risks those decisions.
Think about one of the thorniest debates in modern marketing: lead and revenue attribution. If you can work out which channels and activities deliver the most leads and convert to the most revenue, you’ve solved most of your marketing challenges.
But, because many channels are so difficult to measure, attribution is incredibly complex – and has led to many other marketing problems, like an overreliance on platforms that can clearly demonstrate attribution (*cough* Google Ads). Closing the attribution gap through methods like hybrid attribution is a great example of data-driven marketing done right.
The Detriments of Data-Driven Marketing
Clearly, data-driven marketing is necessary for good outcomes. So why do so many marketers continue to criticise it? There are 4 main reasons:
- Data-driven marketing is often done badly, which can lead to worse outcomes than having no formalised model at all.
- Data collection and analysis can be expensive and time-consuming.
- Data are retrospective – they can’t help you in new frontiers.
- They’re criticising a different definition of ‘data-driven marketing’, like defining ‘data’ solely as numbers that can be measured on a digital dashboard.
The first criticism – bad data are damaging – is very true. An obvious example: relying on a last-touch attribution model. Last-touch attribution means that the attribution or credit for a lead or sale is given to the last touchpoint a buyer engaged with before converting.
Clearly, that’s a form of selection bias: Google Search and Google Ads will often dominate last-touch models, because buyers will normally search for a brand they hear about through another channel (for example, a podcast or word of mouth) in order to navigate to its website. That’s why, if you do invest in data-driven marketing, it’s critical that you stick to the 3 criteria we outlined earlier.
The idea that data collection and analysis can be costly is also valid. Being data-driven doesn’t mean you need to get new data for every decision.
Many smaller calls are low-risk enough that relying on intuition and past experience (that is, old, unstructured data) is perfectly fine. If you’re unsure, conduct a cost–benefit analysis and use your judgement.
The third critique – that data can only guide decisions in recurring scenarios – is simply untrue. As we’ve said, all decisions are shaped by data of some kind. It’s also incredibly rare – perhaps even impossible – that a business will find itself in a situation where no past data applies.
Finally, the idea that ‘data’ solely refers to numerical structured data is incorrect (and marketers are right to call it out). Data-driven marketing is much, much broader than that.
Next Steps
There is no downside to data-driven marketing. In fact, if a brand doesn’t have a formalised process for collecting and analysing new data, it’s essentially gambling with its future.
Not being data-driven means being unable to forecast revenue, know which marketing channels are working, or optimise any aspect of your messaging, strategy, or tactics.
And most brands do recognise that. They just struggle to build accurate, cost-effective data models. The key: work with a partner that’s done it before (many times).
A marketing agency can quickly define environments that should be measured and set up bias-free mechanisms that have been proven to work.
We’ve helped hundreds of brands across Australia measure and analyse everything from Search Console and Meta Ads data to customer interviews and revenue attribution. Schedule a consultation to find out how we can do the same for you.
