With perspective on both in-house BI roles as “The BI Guy,” as well as consultancy expertise, Connor shared great insights on how to start your first BI project, and why it doesn’t have to be so intimidating. He also shared his thoughts on adoption and the innovations that excite him most in this rapidly evolving space.
To get some thought-starters on how to approach your first BI project, read the interview below! You don’t want to miss some of his spot-on analogies in which he compares data to painting, bowling, and cars.
Q: Can you give us a little intro to your background and career? How you got into BI, and what are you up to now?
CJ: I’ve always been fascinated by the fast growth in technology that surrounded me. Growing up, smartphones and computers seemed to come out of nowhere and have such profound impact on our environments, making things evolve so quickly.
This tech fascination led me to study information sciences and informatics at the University of Sheffield in the UK. In parallel, I also studied Business Management to get a practical understanding of how to apply information sciences to business.
My first role out of university was in consultancy for business intelligence, so I basically slipped right into the industry for which I thought, “this is the one I want to be in,” - lucky me! I’ve been a consultant for about 70% of my career, and the other 30% has been leading up the Business Intelligence teams at companies like Unbabel and the British Heart Foundation. I was “the BI guy” at both companies, and was able to build out a small team. That gave me an end-to-end spectrum of what it looks like as someone delivering a service, and as someone who needs the service for a particular company.
This has given me good perspective on how a client uses BI, what questions they might have, how they want to address it - helping me understand how to approach it as a consultant. It’s been great to have that end user experience, which has given me a wider lens on the BI topic.
As for what I’m up to now, I recently started my own company, CJ Analytics, and I contract my data analytics services to clients globally. I help onboard them onto business intelligence, their analytics platform, and manage their strategy and transformation around how they work with data. I help with all developments around their BI projects, whether that be collecting data, governing that data and holding it somewhere, and then developing reporting suites and dashboards. And then, of course, enabling teams to actually use them themselves and self-serve. It’s an end-to-end data experience.
Q: How would you define business intelligence and what is the main overarching objective?
CJ: Whenever I start any project, I always re-define what it means to be a business-intelligent company. People get so worried about how complex it is, but it really doesn’t need to be that way. To me, the definition is as simple as a business that uses data to drive their decision-making process. That’s it. There’s a lot of branches that stem from that, but it does not need to be complicated.
The main objective is to ensure that the data for all touch points across the company is clear, understandable, and consistent. So, wherever someone sees data - let’s say an individual revenue figure - wherever that data is available for the company, it's consistent everywhere, and BI helps to ensure that's the case.
Q: Many stakeholders think: “our company already invests in CRM, Finance, and Marketing software. Why should I also invest in a separate BI platform?” What would be your answer to this question?
CJ: Firstly, I can completely understand any business that says, “why would I want to spend extra money when I already have what I need [to some degree]?” - fair point, since many tools have some kind of reporting capability within it.
The problem is, if you’re looking to grow as a business, you can’t run the risk of having siloed teams. One of the big issues with data is that there’s a LOT of it, and various teams who are responsible for different sets of data. They’ll stick to the data they’re responsible for, which means it can’t scale or grow cohesively. You can’t collaborate on it, and it’s very static.
BI is there to give you a single source of truth - a place where all the data is taken and replicated into one single repository. So that the truth is always updated, centralized, complete, and usable. Once you have that in place, you can create your reporting suites from there.
But if data is split across different teams, only those teams will know and use the data they’re responsible for. Let's say you have a bunch of different teams that sit at the bottom of the operations pyramid, and you have leaders that want to make decisions based on how that's going. They will only be able to get insights into each individual branch that they work in, but there's no clear way to align on a single source of truth for everyone to look at.
That’s why you need to be learning from the centralized data that can be easily highlighted, analyzed, tracked, and fed into the leadership teams to help them with decision-making processes. At the end of the day, data is only valuable when it helps with decisions. And if you’re having to work to find the data, then you’re already wasting time in your decision-making process.
Q: To clarify, is that source of truth the data warehouse?
CJ: Yes. If I could show you a picture of the architecture, a simple way of looking at it would be: from left to right, you’ll have the individual boxes of all the places where you gather your data, whether that be from point of sales, CRM tools or marketing tools. And all of that will have one arrow that points into one big box, which is your data warehouse. And then any reporting that you do, ideally in one or two consistent spaces, will come directly from the data warehouse. So it’s just funneling multiple sets of data into one, and then you have that repository of data which you know to be true. If it’s wrong, it’s wrong in the source as well. So yes, the source of truth really should be the data warehouse. The CRM is just one of the data sources.
Q: Let’s talk about starting your first BI project as a small/medium sized business. What would your first steps be?
CJ: Most of the clients I’ve worked with are larger corporates, but with data, any successful project follows the same concept. It’s especially important for small companies to get it right to save time and money that you might not have in comparison to larger corporations.
For me, starting a BI project is all about the requirements - or objectives - and you won’t know the requirements that you want to set for any given project unless you know what your own existing problems are. So, understanding yourself as a company and what your pain points and problems are is the easiest way to then craft the right solution. The ROI will only be felt if you’re solving existing and painful problems.
To achieve this, define what your problems are in your current state, and define what your requirements are and where you want to be in the future state, and then you can start to build that roadmap in between. Because once you’ve got the start point, and the desired end point, it makes it a lot easier to draw the line in between the two points. If you have alignment on where you’re starting, your problems, and what your business purpose is for BI (end goal), then the rest will fall in place like steps on a ladder.
Q: Would you start with one use case or identify the overarching current and future states?
CJ: I think you do need to have a high-level view and strategy of what everything currently looks like, and will look like, to understand the broader goal. But when you’re actually spending money on tools, or planning how you use resources against data, you have to do it in a step-by-step procedure, which can come after you define your overall strategic goals.
You can’t run before you can walk. A good way to test this is to initially work on a small set of data to see the success story all the way through. So for example, do a test on your sales of a single shop (out of the 10 different shops that you have). You’ll test just with one shop first, to make sure it’s successful with a smaller data set. If this project is successful, then you can replicate and expand it across larger data sets - using the exact same methodology that you already took before.
When you do roll it out with larger datasets, it will be like driving on the exact same motorway, just with more cars going through it at once, rather than just one car traveling by itself.
Q: Is there a data stack that you would recommend to small/medium sized businesses?
CJ: Needless to say, your stack is going to be dependent on your individual situation. But if we’re talking about smaller businesses in particular, I’d say we can split it into a 2-pronged fork: 1) the way we collect and store our data (i.e. data warehousing), and 2) our presentation layer - the visualization tool for dashboards.
If I were to pick individual ones, Fivetran is a good tool for your data collection, and it’s very simple and cheap. It’s clean and intuitive, usable by a lot of people who don’t have the skills that an engineer might. So it’s quite good for smaller companies that can’t afford an engineer yet, or engineers aren’t dedicated to data. You’d need SQL and the concept of how data modeling works, but having an easy tool like Fivetran to centralize your data is a great place to start.
And on the visualizations side, I’d say PowerBI is works well at the moment just because it’s so cheap. You look at competitors like Oracle and it’s crazy how expensive they can be - they may be powerful, but they’re not easy to use. Usually the work needs to be outsourced to consultants (like me), whereas PowerBI has free versions to trial.
I have also done some research into Whaly and I really quite like it. It looks like it’s line with the simplicity aspect, but has a more powerful analytical engine under it that PowerBI doesn’t have. Whaly offers a packaged version where you have some connectors in the box, so that’s a big win.
Q: What about data warehouse?
CJ: The actual logic that exists behind the cloud-based data warehouses are super similar. It’s the interfaces and the language that you use that differ. If you’re running Google Suite products, stick with BigQuery, but that doesn’t mean RedShift isn’t great as well. Note that cloud might not be necessary if you’re a small business. You can also go with something simpler like SQL Server. It’s a cheap way to store your data and is very simple and effective, but of course, you’ll need to use SQL. AWS, BigQuery, RedShift, all great, and Snowflake seems great as well but I haven’t actually used it myself.
Q: What excites you most about the new innovations in this space?
CJ: While business intelligence as a whole has been around for some time, it’s actually quite a new concept to those who haven’t worked with it before. So the ceiling of the potential within it is absolutely huge - the fact that there are still gaps in a highly saturated market. That’s what excites me the most.
However, with the amount of demand that exists, it won’t be saturated for long. It’s incredible how there are new innovations all the time in this saturated market, and how they’re attacking all of these gaps. What I find fascinating is how tools are bringing more and more simplicity to the end user and customer, while still increasing the complexity and robustness in the back end. It’s one of the hardest things you can do, and yet that’s what tools are doing more and more, which is really cool to see.
I truly think we’ll get to a place where most of the complex code can just be written in the back end and users will never have to see it, and it will be flexible to different requirements as well. Static code might not exist. I’m really looking forward to seeing how this industry opens up in the ability to offer the powerful and highly technical solution that Oracle has, but at a fraction of the cost, and hyper-simplified.
Q: That’s definitely fascinating - keeping the robustness but also making it easy for non-technical users to access data. Part of Whaly’s mission is to boost adoption - you can’t really be a data-driven company and always make data-backed decisions if the “culture” is not there, and you’re not getting your business teams to access data.
CJ: Absolutely, a big change that will likely come from all this innovation is the amount of people who will learn how to use data, just because up-skilling people to using these new tools will be much faster. If there are 50 people in England that use data and the tools because they have the skills, once we have these new adoption-focused tools, in the next 10 years that might increase tenfold or more. This number will keep growing as more companies use data and need more people who can analyze it - there will be less skill needed to actually learn how to do it and a lot more analytical insight available from that.
So, I think the tools themselves are going to grow in their capabilities, but I think people are also going to grow in their skills and ability to access data. The competition will be very hot!
This is part 1 of the interview. Head over to part 2 to continue reading!