Platform-provided analytics software doesn’t always tell marketers and sales teams the full story. Paul Savage, COO of ecommerce data specialist Conjura, warns that the old data adage ‘rubbish in, rubbish out’ still applies, and gives best-practice advice for those looking to harness all that the new generation of AI and ML powered data tools can offer.
The amount of data Facebook and its subsidiaries share with each other and with advertisers is still a contentious issue for consumers. However, the reality of what data is actually disclosed by the biggest platforms is a moot point. The truth is that platform-provided analytics software doesn’t always tell marketing and sales teams the full story. In fact, some of these tools actively skew attribution in their favour.
Like it or not, ROAS (Return on Advertising Spend) isn’t the only game in town any more – at least for those that want the truth of where their budget is going. Instead, businesses would be better advised to focus on Customer Acquisition Cost (CAC).
It’s possible to get a more objective view of advertising performance through (non-partisan) Cloud-based open-source analytics tools. The barriers to entry have fallen considerably recently to allow companies without multi-billion analytics budgets to better compete with the global players.
At least that’s the theory. The reality is machine learning has its limits and the old data adage ‘rubbish in, rubbish out’ still applies. Therefore it’s all-important to get your (data) house in order before you can expect to pull any meaningful insights from it.
Do you need to invest in analytics…yet?
Most e-commerce firms have already taken the plunge into analytics to a greater or less extent, but not all will necessarily see the value. Early adaptors across most segments are already enjoying a significant competitive edge gained from high quality analytics-led approaches though. Two of the most celebrated data practitioners come from very different disciplines, Aviva in financial services and Tesco in grocery.
To extract enough insight to make the investment worthwhile, you do need to have enough customer data/opt-in marketing data to allow for meaningful segmentation though. In real terms, this means an annual turnover of at least £10 million. In other words, businesses considering analytics should be at least in the scale-up phase and have already have cracked the code on the basics – growth strategy, product and market fit, messaging and so-on.
Data may lie behind the e-commerce arms-race, but rushing in to analytics without proper preparation will only waste time and effort. It could potentially lead to bigger issues down the line too.
We hear a lot about data lakes, but without having the means to access and understand it, these are more data quagmires. Extracting any use from data means first combining all sources into single, accessible cloud-based database and using a schema and structure that will allow it to be ‘read’ by the chosen analytics engine.
This process may well sound incredibly ‘techie’ and for this reason, data is often laid squarely at the feet of the IT team. This is a mistake given technologists will generally approach the project as an engineering problem to be solved, rather than an evolving tool that can used across a business.
Analytics shouldn’t be owned by IT, nor should it just be the domain of the sales and marketing teams. The true value in data lies – eventually – in being able to connect the dots between departments. However, if the analytics platform doesn’t meet the needs of particular stakeholders, they may take it upon themselves to build their own and the data becomes siloed once again.
From this perspective, no single department should take ‘ownership’ of data. Digital transformation is contingent on horizontal shared ownership so it doesn’t matter who takes the lead within the business. What’s more important is that the right people are involved from across departments and all are working towards shared goals.
Avoiding data for data’s sake
The next step is to find a consensus on what data you ‘really’ need. It nearly always starts with marketing spend given there is a business imperative to understand what channels and what campaigns are really driving most growth. In time, this can be expanded out to provide the bigger picture, but it’s sensible to start with the most pressing strategic priority.
From here, set further achievable goals that can be met within an agreed timeline that align to specific business needs. Once you know what audiences and what channels perform best from an acquisition perspective, the next step could be to identify what existing audiences have the greatest potential lifetime value.
The more specific and the more granular the objective the better. It often makes sense to work backwards from a particular business challenge and consider what sorts of insights will allow you to take proactive or remedial action. Let’s say you want to reduce the number of returns, in which case you ask the platform to look for patterns in age, gender, location, acquisition channel as so-on to create a profile for a ‘serial returner’.
At a later date, you analyse the same issue from a fulfilment perspective – how often does stock arrive damaged, from which warehouses, and using what courier firms etc?
Using a layered strategy, it’s possible to build out a helicopter view of a particular aspect of the business from acquisition, to delivery to customer service. Each overlay adds colour from across departments and each should offer actionable insights that will drive efficiencies and/or improve customer satisfaction.
No-one would ever suggest data is easy of course. Just when you think you’ve got a handle on it, a legislator or a technology provider will invariably throw a spanner in the works. But that doesn’t mean any business can afford to ignore it – at least once they’ve reached a certain scale.
There has been a huge increase in online sales over the past year, but once the market settles it will be much more competitive. It’s only those companies that have moved fast to nail data analytics that will continue to grow at pace. As such, there’s a limited time frame – probably 9-12 months – before an analytics-led approach becomes standard and any first mover advantage will be diminished.
Analytics will become a standard piece of the e-commerce toolset for any business that needs to grow – which is all of them. So, the question is not should you invest in data, but how much and how soon should you invest to get data right?