About this episode:
In this episode of Behind the Data, host Matthew Stibbe interviews Mirela Mart, CFO of Articulate Marketing, discussing her insights on data management, the impact of AI on data cleaning, and the importance of stakeholder engagement. Mirela shares her career journey from mathematics to accountancy, emphasizing the need for a structured approach to data. The conversation also covers the significance of data storytelling, ethical considerations in data representation, and a case study on a revenue tracking project, highlighting lessons learned and the importance of aligning stakeholder expectations.
AI-generated transcript
Matthew Stibbe (00:00) Hello and welcome to Behind the Data with CloverDX. I'm your host Matthew Stibbe and today I have a very special guest on the show, somebody who I know very well who is in fact the CFO at my company, Articulate Marketing. ⁓ Mirela Mart, welcome. Thank you for being on the show today.
Mirela Mart (00:17) Hi Matthew and thank you for having me.
Matthew Stibbe (00:20) So before we dive into your world and your experience with data, I'm really interested in knowing what you're geeking out about right now.
Mirela Mart (00:29) Well, a lot of things I'm geeking out about when it comes to data... but I think the latest that comes to mind is about the AI and what it can do with data. So, recent examples relate to data cleaning
Stuff that in the past maybe, you know, either you used to spend a lot of time in Excel doing that or, you know, using a little bit of script with Python or pandas. Right now you upload something into ChatGPT and it does quite a good job of cleaning it. But ⁓ while this is kind of a nice possibility, it also brings challenges with it and challenges about data security.
You know, is your data confidential? ⁓ if you upload it into ChatGPT, what type of subscription of ChatGPT do you have? You know, what happens with your data? Does it train the model or not? So there are a lot of, ⁓ you know, data security, ethics around it. So it should be used with care and, and with diligence.
Matthew Stibbe (01:40) Yeah, sure. And I know you spent quite a lot of time learning some data science techniques and learning Python and pandas and things like that. And do you think that if you were starting that now, you would be taking a different approach because of AI or would you even start it today?
Mirela Mart (01:58) Well, I think there is value in ⁓ doing a data qualification or learning about how you can do with data and the possibilities because with all the tools, you kind of need to know what the tool is capable of. But you also need to learn how to structure your data and what you can ⁓ get out of it ⁓ and how to kind of create your infrastructure. I think for me, the
benefit of doing the data science qualification was that mindset, that structure, that architecture, that, you know, if you're not doing such a qualification or if you're not doing personal research on knowing about that, or if you're not structured by nature, you cannot bring into the big picture. AI is a tool. I mean, you ask it to give you something, it will give you something, but does it make sense?
Is it the best thing out of it? Can you challenge it? Can you ⁓ kind of see if it did not hallucinate, if it not give you something that is completely wacky or something that is not necessarily useful?
Matthew Stibbe (03:10) There's a real, there's a danger and an opportunity there. And I know from your ⁓ background as an accountant and a mathematician originally, you have quite a rigorous mind in that. Can you tell us a little bit about how you got from mathematics all the way through to accountancy at Articulate Marketing? Tell us a little bit about your career.
Mirela Mart (03:31) Well, it's not necessarily mathematics, but, you know, I was, ⁓ I used to like math when I was in school. So I did computer science in high school. This is where I started and it was good in terms of embedding that structure about, about that logic, because in order to do even the basic programming, you need that logic, ⁓ you know, structuring of information and of steps so that you can avoid ⁓ the loops.
The infinite loops, which were the biggest scare. But then I did economics and economics is a lot based on math and that logical thinking. And then I've decided to do ⁓ an accounting qualification. And of course, accounting and especially management accounting relates quite a lot on dealing with structured data and information and trying to make sense
of ⁓ what is there. So it's a lot of mathematics applied everywhere, so you need to like math to kind of be successful in this.
Matthew Stibbe (04:40) And I, you know, knowing you as I do, this sort of very rigorous, logical approach to information is admirable. You know, having trained as a historian, it's all a little bit gray area for me. But you know, the numbers don't lie for you, do they?
Mirela Mart (04:57) The numbers can lie, but that's why you need that logic, that structure, that curiosity to kind of go a layer below and peel the onion until you get to the right core. I mean, you can... ⁓
manipulate data and we can talk for hours about data manipulation and all of that, but no manipulation necessarily in the good sense, but kind of ⁓ skewing the data to tell a different story or something like that. But it's very important to kind of get a sense and a feel of the data and you need to like data to be able to be
⁓ a good analyst because otherwise it's like, you know, if you're scared of data then you go nowhere.
Matthew Stibbe (05:46) Give us a little potted biography. Where have you worked before Articulate?
Mirela Mart (05:52) So before Articulate I worked for quite big companies. I started with Unilever, then I worked for the World Bank ⁓ and then I joined GE Capital and in my previous job before Articulate I used to work for Visa Europe.
Matthew Stibbe (06:09) Apart from the extra zeros at the end of all the numbers, what's the difference between, from a data perspective or reporting perspective, between working in a very large company in a multinational and a smaller business like Articulate, if any?
Mirela Mart (06:26) I mean, there is a difference because ⁓ in a large company you deal with huge data. I mean, you know, thousands and thousands, tens of thousands of record that you might need to analyze and make sense out of it. When a small business, everything, you know, it's something that can fit into ⁓ maybe an Excel spreadsheet rather than using a tool. That's kind of the...
⁓ the difference between the data, but in the small company you have more flexibility of implementing your own logic, your own structure without going through alignment, perhaps with other regions, without having to kind of involve a larger number of stakeholders. You have more freedom in terms of decision-making of implementing what you want to get out of it, provided that you know what you want to get out of it.
Matthew Stibbe (07:26) Well tell me a little bit about your experience of dealing with stakeholders because I think that this is a fundamental part of accountancy and financial reporting and working with data. It's all very well to have the data but you have to do something with it. So what's your strategy for dealing with stakeholders?
Mirela Mart (07:46) Well, I think a lot of times, ⁓ especially in finance, because my background is mainly finance, but I've also worked in business performance management. A lot of times people think that data belongs to finance or belongs to IT. So there is sometimes little interest from even senior stakeholders in understanding, you know, what
data is available in the company and what can be done with it and what investment in the systems and infrastructure and alignment in terms of methodologies ⁓ need to happen. So I think it's very important to have top of the company support in terms of building your data architecture, in terms of
restructuring your systems, integrating your systems, making sure that all your methodologies are aligned. So it's a huge project and sometimes people dismiss it because either they don't want to deal with it or maybe they don't understand or they think it's somebody else's problem.
Matthew Stibbe (08:52) Hmm.
So there's one part about executive sponsorship. Also, I know when you communicate with me and Articulate, you take pains to, I won't say dumb it down, but translate the financial thing into a language that we can use. How do you go about communicating with data to people who are not finance or data minded people?
Mirela Mart (09:28) I always believe that if you can explain a concept to like a five year old and that child understands what you are talking about, I think you're doing a good job. So, ⁓ if you're an expert in something that you are doing,
Matthew Stibbe (09:38) or CEO.
Mirela Mart (09:51) it shows more skill and more expertise into your field of expertise if you can explain something into very simple terms. And sometimes you do analogy that is related to ⁓ nature or food or like, you know, everyday life, because I think, you know, finance and data, they are surrounding us and you can have something which is translated to everyday life.
Matthew Stibbe (10:20) And that's part of the expertise, I think, of reporting and analysis is finding the way to communicate it. I was just thinking of that Groucho Marx film where he's saying, this treaty is so simple, even a five-year-old child could understand it. Quick, bring me a five-year-old child. ⁓ What's the role of data storytelling?
Mirela Mart (10:34) Hahaha ⁓
I think it's a fundamental role if you're asking me because ⁓ in order to get people interested in data and in fact in order to understand the data you need to understand the story behind it ⁓ and that's why it's extremely important. Of course, you know,
there is a little bit of creativity involved into the data storytelling, but you need to understand your audience and how knowledgeable they are in a certain topic and how patient they are in kind of grasping everything you're trying to say to them ⁓ and how creative they are. ⁓ So I think it's very important to kind of find the right story, but...
it's also important not to use that storytelling to manipulate in the wrong way, the data, because if you are removing some bits of information to kind of adapt that story in a way that maybe will make the data look better, will maybe hide a lack of performance in some areas, that comes into the ethical way of
telling stories about data and using the data in an ethical way.
Matthew Stibbe (12:04) Of course, the accountancy profession has ethical standards and expectations and things, but actually in real life, how do you go about applying that? Even when you're translating complex data into simple concepts that even five-year-olds and CEOs can understand, how do you apply that rigor to make sure that you're not misrepresenting the data?
Mirela Mart (12:28) First of all, I'm trying to look at the entire picture, the big picture. I want to make sure that my data contains everything that it needs to contain. I mean, you have in even in audit, where assurance like that concept of representing a true and fair view. Yeah. So you kind of need to make sure that you didn't miss something from the cracks and
you have the entire set of data that you need to kind of tell the story of. You're not having something in a drawer left aside or you're not having something ⁓ somewhere else. Of course mistakes can happen, but as long as you do everything possible to kind of make sure that you have all the data, that's the first step that you need to do.
And the other thing is like when you start and do your analysis and you go through kind of all sorts of graphs and segmenting and looking at the data, you always need to kind of make sure that there is the link to the big picture. And of course, you haven't stripped out something that might skew the data and what you are doing. I think...
You know, the best learning for me was when I was at Unilever and the finance director at that time ⁓ mesmerized me with his skill of doing something back of the envelope. It's like, here is what it is with a pen and paper. This is what it is. This is what I would expect. Now go and, you know, really come and tell me if I was right with everything you have into the system. And
I've always admired the people that actually with knowledge and experience of that company, of their numbers, were capable of saying, you know, it should be around this and it should be in this ballpark. But of course, you know, I as an analyst at that time was I was able to kind of substantiate with real data. ⁓
Matthew Stibbe (14:45) This is an interesting concept actually, because you do this back of the envelope with wisdom and experience, ⁓ and you can get quite close to it. When I was learning to fly, we used to call that a gross error check. And you as perhaps at that time a junior accountant were really licking all the data to make sure it matched or make sure it... or maybe see if it didn't.
There's a concept behind that, I think, of granularity and materiality, which is a lovely accountancy word I've learned recently. How do you apply those concepts? At what point can you just go, I'm willing to take a very broad view and just do a back of the envelope based on my experience. Where do you go, I actually need that level of detail? And where do you stop going and digging and getting more and more and more detail?
Mirela Mart (15:15) Hmm.
Well, it depends on where that analysis is going. If it is used for internal purposes and you are the decision maker or it doesn't have a lot of, ⁓ let's say, reputational impact, you can live with a back of an envelope
calculation or something because anyway, you never take decisions with perfect information. But if that information goes to a board, goes to the street, it has, you know, reputational implications for the person that is presenting it, you really need to kind of back it up with ⁓ kind of proper analysis. You will never go to a board of the company with the back of an envelope calculation, or you will never report in your ⁓ annual reports or quarterly reports.
Matthew Stibbe (16:30) when you won't do it twice anyway.
And I know sometimes, and I say this with great respect, sometimes we have quite heated conversations about information and when I question things or I'm trying to understand things, how do people who deal with data and accountants being a very good example of that, manage that ability to speak truth to power?
Mirela Mart (16:31) Yeah.
Well, I think, again, data storytelling is one thing that will help you explain ⁓ all of those concepts. But you also need to kind of understand a little bit what your senior stakeholder is looking out for. Because sometimes
you know, they see something, of course they don't like it, but if you are able to kind of explain and understand why that thing happened and how you present and you spin the story to kind of tell a story that is true and accurate, it might help into the conversation, because otherwise you're just
kind of ⁓ having a conflict. And sometimes you really need to draw the line and you kind of say, no, there is no further analysis. There is no other way ⁓ I can do or I can change my analysis. This is what it is. You just have to take it like that.
Matthew Stibbe (17:59) There's, suppose, the difference between the messenger and the message, isn't there? Yeah. So I'd love to explore a specific project with you and see if we can draw out some lessons from that. So I wondered if you could share a project that you've worked on with data and tell me a little bit about it.
Mirela Mart (18:20) Well, I worked on ⁓ a few projects with data, but I think I will talk about ⁓ a recent one that kind of involved me spending almost a year fine-tuning everything. So, in a previous company, the problem that I had at the moment is that there was a big stream of ⁓ revenue
that we could not track properly using the accounting system. And the management wanted granularity to know what that stream of revenue was generating. So we didn't know. mean, overall, we knew the total revenue, but we didn't know exactly that particular ⁓ line of business. And especially being of strategic importance, people really wanted to have that data. Now, the problem was,
⁓ that we knew how we were billing that, but that was not translated into the accounting software. So you had to start from the basics. It's like, what is my list of products? ⁓ Who owns them? ⁓ What is their definition? ⁓ What are...
the billing codes related to that? Why can't I find them? And how much do I bill on each of that? What's the frequency of billing it? And then how do we map them to kind of a certain accounting code so that we can extract them out? And in that process is not only kind of identifying the product, their definition, their business owner, but you started to develop a taxonomy
for all of that and a methodology that you can apply consistently and everywhere. ⁓ And the problem was that, you know, being in a global company and being in a certain region, I chose an approach and maybe the other regions chose another approach and the guys in the headquarters, they had other ideas. So it was a lot of...
conversations with the headquarters to understand what they were planning to do, what I was suggesting, cross-checking with them, kind of my implementation, showing them how I'm thinking, adjusting my thinking so that I'm not going in a divergent way versus the headquarters. It was quite a lot of not necessarily...
analysis and implementation, but it was a lot of managing stakeholders and communicating and making sure there was that alignment. So that, you know, I am, I was facilitating the headquarters to have insights into what I was doing, but also making sure that I was aligned, but also making sure that I also influenced them if what I was suggesting
was something of very good quality, worthwhile to be implemented worldwide.
Matthew Stibbe (21:40) Without without getting into the... I understand that we want to protect confidentiality but what was the output or the end result of that I mean in terms of as a project what were you delivering as a taxonomy, a classification, a data definition?
Mirela Mart (21:56) No, what I was delivering was the possibility of actually extracting from the accounting system in an easy way all the revenue information for that line of business. And we were talking of more than 100 products and also having a sustainable method so that it can be applied for new product introductions so that you could immediately see
Matthew Stibbe (22:22) you
Mirela Mart (22:26) you know, where they were coming based on their names or their codes or something that was intuitive, I mean, without knowing too much about this project, building a solution that was sustainable for the future and could have been replicated.
Matthew Stibbe (22:46) So ⁓ we're almost out of time. The last question I want to ask you is, from that project, if you were going to go back in time to the beginning of it, knowing what you know now, what did you learn? What would you do differently? What would help that project be more successful?
Mirela Mart (23:04) I think the thing that would have helped the project be more successful would have been that ⁓ enhanced stakeholder ⁓ expectation and management. ⁓ Unfortunately, I jumped into the project when everybody wanted the data yesterday. ⁓ I wish I had some more time in terms of dealing with
maybe other regions with the global team in terms of ⁓ aligning expectations and aligning methodologies better ⁓ in the sense that we could all go and have the same approach across all regions. So that was more about dealing with ⁓ stakeholders and alignment ⁓ rather than trying to fix a problem.
Matthew Stibbe (23:51) Thanks.
It's so often the story, isn't it, that it's not the software but the people 'ware' that's hard. Well, fascinating. Thank you very much for sharing that story and that learning. And that brings this episode to a close. Mirela, thank you very much for being on the show today. It's been fascinating. ⁓ And if you'd like to get more practical data insights or learn more about CloverDX, please visit cloverdx.com, Behind the Data. Thank you very much for listening and goodbye.
Mirela Mart (24:09) Yeah.