Why you need to Centralise your Data Analytics Team

Nick Creagh is an experienced data & analytics leader. He has a PhD in Mathematics and spent over 10 years in management consulting before moving into leadership roles at leading education sector companies, Pearsons & TES. His core strength is creating teams that deliver value, insight & data products that enable powerful strategic shifts. He does this by pulling together siloed analytics teams into one cohesive unit and then driving that team to deliver company-changing Analytics.  Nick wrote this article as an amalgamation of several popular articles he wrote on LinkedIn that aimed to share insights from his wealth of experience helping companies build out their data analytics capabilities.

 

 

Alphabetti Spaghetti

You’ve heard of the C-Suite? The plethora of titles that make up the Board of every FTSE company, and more than a few others as well.

Ok, let’s play a game. Go through the alphabet and see how many letters you can match to a title in the C-Suite…

Done? Ok, so how many did you get? CEO, CFO, they’re the obvious ones. CMO? CTO? In the more advanced companies, a CIO, CSO and a CDO?

Then there’s the CCO, CHRO, CPO, COO etc. I’ve even heard of an XO (Executive Officer) – the second in command to a Commanding Officer (CO) in the military.

The list can go on, and I’ve got a new one for you.

A CAO…

 

The Chief Analytics Officer

So what do they do? On one level, it’s obvious. They’re the board level champion of analytics across the company. So the more important question becomes – why do we need one?

Answering that is the purpose of this article. I’m going to argue that not only does your company need an analytics team, but that team also needs to be centralised and have a leader that sits at the highest level of the company. A Chief. Not a Director, not a Manager, not a Head of. A Chief.

Making the argument daily for analytics to drive performance, performance measurement, product development, customer interaction and the bottom line. And if you don’t have that, you’re going to be outclassed by those who do.

But first…

 

What can a data analytics team do for me?

I’m prepared to bet, regardless of the type of company you’re in, you’re swimming in data. You may think you have it under control, and you’ve dashboards that update your key metrics on a weekly, daily or even on a live basis. You may even have people who can predict the future with advanced models.

But is that just for you, your department, or your entire company? And what about your customers – could analytics be integrated into the product/service you provide?

With the right data analytics team in place, you can have all this, and more. Not just the right metrics, but the right decision-making, the right predictions and the right products that bring the right customers to you, ready to buy.

Analytics products can do anything from:

• Analyse customer behaviours. Not just to show what your customers are doing now, but what they think and feel about your product, and what they want it to do. Harnessed properly, this tells you what your customers will do next. I’ve harnessed customer segmentation theory to transform an eCommerce business, changing how product teams viewed customers and allowing them to focus on the 10% of the market that leads the rest. Couple that with a strategic view of future markets, and your Strategy team can see into the future!

 

• Develop new product lines. Previously I have built recruitment products that increased the candidate pool by up to 25% in a key market, turning a marginal add-on product into a key, profitable part of the customer offering. Couple that with an empowered sales team, and your go-to-market strategy lifts off!

 

• Identify profit and loss-making areas. Using market basket analysis (a well understood but underused algorithm), I identified profitable product lines as well as separating loss-making from loss-leading ones, for a publishing company in nearly a dozen international markets. Couple that with advanced analytics techniques in the warehouse management team, and your Supply Chain team are supercharged!

 

• Fill gaps in your data. I’ve implemented tools that more than doubled the number of reachable customers for marketing campaigns by deducing likely field values for customers from their on-site behaviour. Couple that with data-driven campaigns that invite customers to complete auto-suggested fields and you have a strategy that Marketing adore!

 

• Listen to your audience. Understand how customers are talking to each other about your product, not just to you. Using network theory, I have helped teams understand which customers are leading the conversation, and how to take advantage of a ‘ripple effect’ to increase the power of marketing messages and customer communication without an increase in spending.

 

In order to build and implement the right analytics products for your business, you need a team that understands your data, knows your strategy, experiences your business and works across the business.

That’s more than just number-crunching, that’s a truly transformative team, a powerhouse working across your organisation, at all levels, with all stakeholders, across all product lines. Couple that with a marketing team that understands what makes messages go viral and you have a method of transforming Customer Communication.

That’s an analytics team.

And getting there is takes climbing a ladder. And you’re probably already on the first rung.

Christophe Ferron | Unsplash

 

The Analytics Ladder

Hopefully, you already have people who are number-crunching as part of their day-job, and not just in finance. Your digital team will be using Google Analytics to check the performance of your website, your logistics or supply Chain team will be running numbers to manage stock flows and your customer-facing teams will be working to daily, weekly and monthly metrics.

So I’m not going to try and convince you that you need people who know how to handle numbers – that’s been true for centuries, and why the original definition of a computer was one who computes, i.e. a person, not a machine.

I’m also prepared to bet that your company already has people who are number-crunching as their full-time day-job. You might even call them analysts. If you do, you’re already on the first step of the analytics ladder.

So let’s look at those other steps of that same ladder.

 

Level 1 – “But isn’t that something the accountants are doing?”

As I’ve said above, I don’t think there’s anyone who doesn’t have people running at least some numbers as part of their day-job. But if your finance function is the only one doing analysis, then that means no-one’s doing sales modelling, customer forecasting or any such. Or, if they are, then it’s being driven by finance.

I don’t believe there’s anyone who’s only here. So let’s move on.

 

Level 2 – “Yes, I need some analysis done. Let’s get a few bright grads”

Ok, now we’re getting somewhere. I’ve worked in organisations like this, I’ve consulted to organisations like this and I’ve even been the ‘bright grad’ in that role. This works well if your data can be trapped on a spreadsheet (i.e. its not just small data, but really small data) as your “bright grad” can then model away.

This is the first real rung at which analytics can be said to be happening. So if you think that analytics needs to be done by only one person, and they have all the power and that you’re the kind of leader that likes to hoard information, then this is the level for you.

 

Level 3 – “Data is good – let’s manage it”

The first thing I’ve noticed organisations do once they’ve gotten used to the idea of having someone working on data, is grow small teams in different directorates. But that’s the beginnings of using analytics corporately. The first real step in moving the organisation to a corporate analytical mindset has to be the building of a central analytics team.

This can be done by having a single team reporting to someone in the C-Suite, or by creating a common working arrangement between separate analytics teams, which I’ve built and seen work in organisations that are highly decentralised.

This is also the level at which that team can own centralised dashboards (at least, the building of) so that the organisation starts to get used to seeing data in a standard way, visualised in a common tool, and begin the conversation about what to do about it. That and the other challenge of getting the analytics team to work well across departments is where the move to the next level comes in.

 

Level 4 – “Information is good – let’s talk about it”

Now you’ve got a central team. This means the analysts in that team can start to specialise. Their challenge comes in working for the whole organisation, not just the bit they nominally work in. This is the same challenge as for any back-office corporate-facing function (like HR, Finance or Tech) and, to my mind, is answered in the same essential way – with champions. Just as HR have their HR Consultants sat with the business and Finance have their Finance Leads (or Directors) in each business area, so I think Analytics needs to put Lead Analysts out in each main business area. Their role is to be part of that area’s day-to-day conversation and make sure analytics is supporting that conversation where possible.

The reason I’ve called them leaders is because their role is different from that of the lone grad sat in a general business team. They represent the Analytics team to their area and their business area to the Analytics team. This way they can bring challenges back to the Analytics team as projects to be managed, and its their role to manage the delivery of those projects. This could be anything from a deep dive into the data (aka what’s going on here?) to building a new dashboard (somethings changed and we need to track it) to building a new product (data science as business transformation).

 

Level 5 – “Debate is needed – let’s universalise it”

Now you have a centralised team that’s:

• Driving conversations across the organisation.
• Managing your metrics.
• Simplifying your suite of dashboards.
• Solving problems like there’s no tomorrow.
• Using advanced analytical techniques.

That’s the basic Analytics ladder. There are steps beyond the 5^th rung, which is where the whole ‘data-driven’ vs ‘data-informed’ debate comes in.

The reality is that a data-driven company means one in which any decision is informed by the most up-to-date analysis of the current and likely-future positions, and then a leader makes a call and is held accountable for that call.

No-one is going to let an algorithm make every decision in a company, well, not yet…

 

What kind of analytics team do you need?

Ok, so lets re-cap. You’ve worked out what kind of analytics capability you currently have and are thinking that maybe you need to do more. You’ve got problems not solved, teams not tracked, data holes not filled, opportunities not exploited. And your analysts are telling you that they can do so much more, if only they were further up the ladder.

So let’s talk about the kind of team you can build. As with any team, specialisation is key. It’s my view that there are four key components of an analytics capability:

• Analysts – Whose job it is to talk to the business in plain English, understand their concerns and work with them to make sense of the data. Often grads, these are the natural iteration of those lone grads we saw at Level 2, but now with a team structure, senior support and growth potential, they’re no longer alone, or quite so junior.

 

• Dashboard Developers – When an analyst identifies a problem, it could mean a new KPI needs to be created and tracked. Which is where these guys come in. Also on their agenda is the simplification of the existing suite of dashboards.

 

• Data Scientists – The blue sky crew, these are the people to bring in when a problem is beyond the ken of the analysts. These are also the team who build products the first time (whereas the Data Engineers build it every time after that).

 

• Data Engineers – These are the guys who knit all the data together so that others can do their jobs effectively. Often these have a dual reporting line into the CTO as well, as their skill set is that of an engineer, but their customer is the analytics team. They’re also the team that works with the data scientists to build any data products in a robust and reliable way (as opposed to the trial version a data scientist will build – it’s the difference between making a test-ready car and a road-worthy car).

 

That, to my mind, is the minimum needed for a true Analytics team. On top of that, I’ve also worked with Digital Anthropologists, Data Managers, Data Governance Experts, Data Lawyers and Data Architects. But that somewhat depends on the nature of the organisation and the appetite of the C-Suite. Different organisations will need a different mix of skills, depending on their exact situation and requirements – a company with a physical supply chain will go heavy on supply chain skills in their analysts, whereas familiarity with Google Analytics is vital in any team facing off to a digital marketing department.

 

Building your central team – how and why

Oladimeji Odunsi | Unsplash

Fingers and fists – an analogy

Organically growing your team is one thing and it’s probably how you built your current analytics capability. But how do you develop a central function? And why should you?

In the Gangs of New York, at one point Daniel Day-Lewis’s character is describing how he controls his little bit of New York. The metaphor he uses is powerful. As he says:

“Each of the Five Points is a finger. When I close my hand, it becomes a fist.”

Fingers closing to form a fist. Or staying open to be an inviting palm. Or anything in between. You can do a lot with a hand. A lot more than you can with a single finger…

So what’s this got to do with analytics? We’re talking about building a data analytics team, so the question here is ‘do you want a team to focus on a single area or the whole company?’. That’s a question of maturity and reach. If you’re starting out, you might want to form a specialist analytics team – a marketing analytics team, a supply chain analytics team, what have you. A single finger.

You start with a finger – but don’t forget the fist.

The roles I described above are what’s needed to form an analytics team. Of these, I believe that only the analysts are truly unique to that particular department. It’s only the analysts, embedded in the frontline departments but with a connection back to the analytics function, who form the fingers. The rest, the common skill set, form the palm of that hand.

It’s my contention, and the point of this article that pulling those common skill sets into a central team, with the analysts embedded out in the business makes your analytics capability much more powerful.

By sharing skills, knowledge and experience:

• Your data scientists grow and learn, and can take their expertise and experience in one part of the company to another.
• Your dashboard developers can build company-wide metrics and have a hope of creating a ‘one source of truth’ for those metrics.
• Your data engineers can take an overview of all your data, not just one bit.

Not doing that, building department-specific analytics ‘fingers‘ means creating silos. If I were just talking about the engineers, I’d be preaching to the choir – after all, its standard now to centralise engineers into one place, so they can be pointed at problems, not shuffled off into different departments/teams. So why do we continue to create analytics teams in silos, when they are so much more powerful together?

Why do we build fingers, when we should be forming fists?

Debby Hudson | Unsplash

So how do we build it?

Growing a single team is easy enough, as organic growth is a matter of building according to demand. But if you already have a number of teams, then the joining of them will take consensus building and work.

The challenge here is to identify the need. This starts with finding business problems that require joint teams to solve as a project. Problems like:

• Identifying how our customers behave on our website and the impact on stock management.
• Understanding how customers change behaviour between devices, and adjusting the page mix accordingly.
• Understanding Cost to Serve models – how costs hit products at different points in a supply chain differently, turning some products into profitable products, and others into loss-making. And, due to customer purchasing patterns, other products into loss-leading (where although they make a loss individually, they act as ‘entry drugs’ to more profitable products that customers buy)
• Connecting feedback from contact centres back to customer management processes, understanding how one impacts another.

Or to put it another way, finding problems that require holistic solutions, even if they first appear in one area.  The problems that need joint taskforces to solve – that use cross-functional skill sets to solve problems hitting the bottom line. These are the problems that centralised teams excel at solving.

Of course, the HR angle is not to be underestimated. The fact that having a single team creates a team environment, allowing analysts to challenge each other, learn from each other, try out new ideas and feel part of a team will have a positive impact on retention, performance, morale and all the other team-specific metrics that will have been hurting up until now.

 

Conclusion

Up until now, I’ve been arguing that in the modern company, the need for company-wide analytics means that you need a centralised analytics function, that combines all the skill sets of the analyst, the data scientist, the dashboard developer, the modeller and the data engineer at a minimum.

Only a centralised analytics function can solve the company-wide problems that companies increasingly face.

But does such a function needs board level sponsorship? But what if you had all of that and no voice on the board? Where would such a team sit? CFO? Then it will primarily solve financial problems. CMO? Same, but marketing ones. CTO? It will solve problems as seen by engineers (although I’ve seen this work in smaller companies). Anywhere other than the CEO and it’s recreating a siloed team, which I’ve been arguing against.

And if the team reports directly into the Chief Exec, then either the Chief Exec argues the analytics case, or that leader joins the Executive team. Either way, you have a centralised board level champion for analytics.

You could call that leader a Chief Data Officer. But that’s going to mean someone who manages data not extracts value from it. So if you have a board level leader who’s championing and extracting value from data?

You’ve got a Chief Analytics Officer. Welcome to the 21^st Century!

 

 

 

 

If you would like to write for us please let us know by emailing [email protected].

Previous guest articles include:

• Dr Janet Bastiman – Do you need an AI team? If so, how do you build one?
• Andrew Jones – The Interviewers Perspective: Data Science Interviews at Amazon
• Ben Dias – What hiring managers want to see in Data Scientists’ CVs