5 Tips before start your Data & Analytics Projects
Hello Medium readers, today I’ll bring a post that I wanted to write for a while, but I was expecting for a little more of experience working as a Data Scientist. I choose to wait because I wouldn’t like to be that guy, we all know that guy, that shares information and tips without never experiencing these tips in real life, well I assure you, that’s not the case here.
Now that I’ working for the last 3 years and a half as a Data Scientist, and in the last 6 month as a Senior, I converge to 5 points that I learned in hard lessons developing projects related to Data & Analytics (DA) realm. In this period, I directly worked on 9 projects that resulted in financial cost reduction, and some others without it. Covering distinctives areas such as material forecast demand, warranty, production line optimization, product delivery optimization, and stock reduction.
Now, when I must begin to work on a new project in the Data field, I always try to use it these principals, and doesn’t matter if is something simple such as a statistical analysis or more complex such as a Machine Learning (ML) Model. Maybe, you can even apply these tips to a broader management project idea.
So, without further ado, let’s jump into:
Listen your clients, make them comfortable
If you ever developed something for someone you know how easy your clients can change their minds, doesn’t matter in Software Development or in Data & Analytics field, this always harms your timeline and stress the development team.
So, one thing that I like to do in the first meet for an eventual new project is listen all thoughts, pains, wishes, and requirements of the clients. We know that normally in these meetings, clients want to receive the sky, but most of the time, they just need to have a picture of the sky, and even that you need to give the sky, you certainly will do by phase, step by step.
Be careful about judgement in early speeches of your client, therefore, they may take a defensive stance and all communication process can be affected, and you as a developer / project manager want your client as a helper partner and not a staunch critic.
A bridge question to the next Tip that I always like to do is: “how do you get access to the data related to this process?” Believe me or not, but this question can give you a very good visualization of how complex your new project may be. I back to this in the last Tip, but I like to use when I’m ready to start to guide the discussions.
Bring your clients to reality
After the active listening is extremely important to be frank and direct with your clients, bringing them to reality, because most of the time you won’t be able to deliver everything that was requested for several reasons, and for me, it always fell in these following categories.
- Doesn’t make sense
- High effort and low additional value to the business
- Very complex, too much long project or too much risk involved
- Unavailable technology
It’s quite important for you to know exactly how you can construct arguments to explain to your clients why is not possible to deliver everything that was requested. A simple “no” without basis will weaken the belief of your client in your development, and maybe if your client has an optional to not continue if your development team, they will try to another team or company.
One example that I like to use is: “as you have goals to reach and metrics to follow, my department or team also have, so we need to prioritize what can we develop, maybe if we wait for a new technology comes, or reduce the scope of this project, or even, have a longer period, breaking in sprints, then I think we can make a deal”. Do not just say “no”, find a way to say how can you deliver them something.
In rare cases you can delivery all of your client’s requirements, but do it with phases, specially, applying the ideas presents in the Agile methodology, utilizing one of its several frameworks. On my team we apply ideas from Scrum and Kanban.
Make sure your work adds value to the business
Sometimes big companies have development teams available to be used as a on demand service, and occasionally, we, as developers, are so focused to deliver results that we want to start to code as soon as the demand arrive. I know this feeling pretty well and I suffered with this throughout my entire first year as a Data Scientist.
One thing that you always need to question is: “Will the company spend more salary with me to develop this solution than the possible gains brought with it?”. Please, Please, Please don’t forget to include the cost with infrastructure to keep the solution up and running.
Over time, I started to notice that always will exists a new problem to solve, but you need to accept that you won’t and in fact, doesn’t need to solve all of them. During my first year as a Data Scientist, I readed this phrase from Steve Jobs.
“I’m as proud of many of the things we haven’t done as the things we have done. Innovation is saying no to a thousand things.” — Steve Jobs
I admit that I thought, “wow, this sentence is iconic”, but I knew it that I still couldn’t realize it’s truly significance. It took me two years to fully understand. If you remain in this area, delivering successful projects, the more demanded you will be for more solutions. In this mid-point of your career, you must understand and develop a self-intuition where you should put focus or not, because most important to know what to do, is know what you don’t need to.
Know all of your stakeholders
Unfortunately, I had the bad experience of developing an entire project without knowing that my key stakeholder didn’t know about the development at all. You maybe are questioning how this would be possible.
In big companies here and there one department takes care about a business process locally or helps collectively to manage a process, but the real manager can be another department, in another business unit, city, state, or even a country.
At the end of the project, I discovered that one department controlled the process in the business’ unit, but another department controlled the budget of this process. Yeah, I know, it’s quite strange but the companies are full of non-optimized process, or being more direct, of bad-designed business process. It’s out of your control, just try do adapt and learn fast as you can.
So, lesson learned in a very hard way, now, before I create the first code line, or even a repo in a Git service, I spend the double of time, checking the risks of the project and know all my stalkeholders, because sometimes they are hidden. Talk with your manager and colleagues can really help to anticipate future problems in the development phase.
Know your Data
I really could create a whole post just to speak about this tip. Data is the new oil, you may already know that, and in DA projects this is crucial. Depending on the size and the technology maturity of the company that you are working, you may face completely different challenges to deal with data.
I already heard from DS working in companies which data still is registry in paper. If this is your reality, then you should consider improving the process first, before really think about Data & Analytics project. On other hand, If you’re luckier, you only find data available in databases or in a Data lake.
As the best option, always try to rely your project on data with origin in system, such as MES, MRP, CRP, and ERP. Normally, data coming from standard system existing in the company results in less effort to develop project and achieve the initial and financial goals.
But sometimes, often, you may need to rely on a excel spreadsheet, csv or even a txt file. If this is your reality, try to create with your client a data governance, basically, given to them the responsibility to keep the data up to date and in the standard format which you built your project to work with.
When this happened, I developed the project to work waiting for a very specific format of data, and ready to return to the user a warning saying that the output won’t be giving until he or she correct the data format in the input file ;).
Another possibility is when you need to rely on third parties to supply data for your projects, you really need to consider them as your stakeholders, because, if they stop to generate this data, your entire project will be affected, and this already happened with me. I had luck to find another team which supports a similar dataset for my use case but be aware about that.
Conclusion
In summary, embarking on Data & Analytics projects requires a thoughtful approach to ensure success and maximize value. By actively listening to your clients and setting realistic expectations, you can build a strong foundation for collaboration and trust.
Ensuring that your work provides tangible business value helps prioritize efforts towards impactful solutions. Identifying all stakeholders early on can prevent miscommunications and align project goals with broader business objectives. Finally, having a deep understanding of your data is crucial for the success of any DA project, allowing you to build robust and reliable solutions.
By incorporating these five tips into your project management strategy, you can navigate the complexities of Data & Analytics projects more effectively and deliver results that truly benefit your organization.
I hope you liked of my post, follow me for the next ones :)
