Matt Allington is the founder of Excelerator BI, which focuses on consulting and training for Power BI users. He’s the author of several books on Excel, Power BI and the DAX programming language. He spoke with us recently about how he moved from being a Coca-Cola sales rep to a business consultant.
This interview has been edited and condensed.
How did you get started with Excel and related technologies?
I taught myself to program on an old HP basic computer in school. In my work life, we had one PC in the building. I worked as a sales rep Monday to Friday and came in on the weekend to teach myself Lotus 1-2-3.
My ability to get results was always based on data and getting access to information that other people didn’t have the skills to get. After working in sales at Coca Cola for 15 years, I eventually moved into IT, becoming the BI director for Asia Pacific. I was responsible for delivering business intelligence to that territory, and I discovered the technologies we were using had a lot of untapped potential, including Power Pivot for Excel.
This was about five years ago, and I was coming up to 25 years at Coke. I felt I needed a change and was so impressed with Power Pivot that I set up a consultancy.
What was the focus of the consultancy?
I thought I’d be doing customized, traditional Excel. Instead, I ended up doing almost all of my work with data and Power Pivot. Whereas I started off thinking it would be a broad-based Excel consultancy, it ended up being a self-service, business intelligence consultancy, leveraging Power Pivot and Power Query for Excel.
Then three and a half years ago, Power BI became available. Two and a half years ago, it became what I considered to be a suitable, quality product that could be used for business operations.
What did Power BI bring to the table?
Business moves very quickly for salespeople and finance people. They have KPIs and need to analyze data. They have reporting requirements that are very short-term. Enterprise BI doesn’t suit that type of analytics. What’s changed in the last five years, thanks largely (if not solely) to Microsoft, is that Microsoft has merged their Excel heritage and their SQL server heritage.
They brought the people that built those products together with the business user in mind. They’ve built a robust self-service business intelligence capability that your Power BI Excel user can do themselves. If you search Google Trends for “Power BI,” you see exponential growth. There is enormous interest in this product because it’s good; it’s been built by Microsoft, and it’s completely flexible.
When customers reach out to you, what are the typical things that they’re seeking help with?
One area would be technical help with the DAX language for data modeling. They’ve been able to progress to a certain extent, but then they get stuck. People call me when they’ve reached the limits of their self-learning capability and they need help. Typically, they don’t know whether they’ve done a good job or a bad job.
I prefer to work with clients that want to help themselves. I’m not a typical outsource-your-BI-needs-and-I’ll-do-it-for-you guy. I’ll help you get the skills and knowledge so you can look after it yourself.
The other area is more strategic advice about how to move forward — people wanting to make a good design decision about what type of infrastructure they need, what sort of processes should be put in place and how to go about building out these solutions.
You talk about three core components in Power BI. What’s core about data modeling, acquisition and visualization?
Microsoft started by building the data modeling piece first. Then they built the acquisition piece to address how to get the data into the modeling engine. Then they built the visualization piece last. You need all three of those things to be a good self-service BI tool.
Microsoft basically grabbed all these Excel experts and said, “How would you like to start it from scratch using modern-day technology, taking everything that you’ve learned over the last 15 years with these other technologies and build something fantastic from the ground up?” And that’s what they did, and they got it right.
Microsoft acknowledged pretty early on that it couldn’t build all the visualizations required by end users. It open-sourced its visualization layer so anyone can build their own custom visuals. I think that was a masterstroke, allowing people to adopt the tool when it didn’t meet the visualization requirements, because they could always get something built custom for their needs.
In my view, the data modeling is the most important piece, and really that’s my primary focus. Data acquisition is the second most important. Of course, the visualization is important, but lots of people just want to see tables of numbers, Excel-style. So I think that’s the least important of the three.
What are the horizon issues for BI?
The BIG thing I see happening more and more in the future is the integration of artificial intelligence and machine learning directly into the tools. Instead of users having to decide what reports they need, the tool will suggest reports based on the data. Instead of users having to know how to “clean” the data, they will just show the tool what “good” looks like, and the tool will work out how to clean it. Expect a lot less of “push this button to do task A;” instead, there will be more of “show the tool what you want, and the tool will work out how to do it.”
In some ways, you’re really a teacher.
It’s what I’ve always loved doing. When I was at Coke, people would ask me for technical help, typically on spreadsheets or databases. I spent a lot of my spare time on forums, helping people for free. I saw my consultancy as an opportunity to do that full time and get paid for it. I do a lot of training in-house, live, video and online. That’s my core. It’s teaching.
Excel Chat provides on-demand help from vetted experts to solve your most pressing Excel needs. Try it now to see how we deliver Excel solutions in seconds.