How do you use data to solve problems?

Even in the advent of Generative AI and other technologies the value of asking “what is the exam question” does not go away, in fact becomes even more relevant. Next key aspect is value of ability to fuse domain knowledge with advanced data and analytics skills. Third is making sure you can implement and embed the solution, because a great idea is still just an idea if it is not applied to the business problem and is embedded into day to day life of the organisation. Fancy tech with five people has very little RoI compared to less fancy tech used by 1000 people.

In my experience asking the right question at the start of any data initiative really sets the project up for success. That is also often the hardest thing to do, as it works in any decision making, the 5 why? approach is what works best in this case as well. If you get the same answer on why you need "data", and for "what", and if you keep getting the same answer, bravo you are there.

the journalist question list applies here Who, What, Where, When, and Why? Who needs the data? What data needs to be collected? Where is the data being collected and distributed? When is the data/reporting needed? Why is it being collected, the end use or goal of the reporting? Applying this will set clear steps, guidelines for the project

I use relevant data to build a mathematical model of the system . Then I can use that model to make predictions which can be tested for validation.

I would like to add one more thing which is data sufficiency check. Often there is a disconnect between business teams and data teams and the business teams are not aware of the data availability and quality so once we start gathering the data we often run into scenarios where either the data is not available in the details that we wanted or the quality of data is not fit for any analysis. These two things pose the problem of data not being sufficient to answer the intended business problems. For example if you want segment wise ROI from your market mix models but if your media data is only available at a brand level it would not be possible to provide segment level ROIs

This is a matter of scope, what is essential to completing the project? it is easy to get buried in data overload, so very clear limits on what is collected with clear guidelines for sources, specific data points to be used and what is to be done with nonconforming data

Primary data sources collected via interviews, questionnaires could have bias and noise that can affect the accuracy of the data. So data analysis is a very crucial stage. I recommend categorising data into FOG (Facts,Opinions,Guesses). Only rely on the facts and numbers. The opinions and guesses can be taken under conscious assumptions and used accordingly.

In this step of the process there is enough data to roll out a test in normal environment. Since there is still room for human error at some point, a test may need more adjusting before initial launch. If the test is successful it is more proof to the stake holders to roll out. If it is not successful revising the plan and pushing back launch can be justified.

SWOT analysis is my favorite because of the strategic nature of this action. In practice we employ a SWOrT - M technique Strengths Weaknesses Opportunities Risks and Threats and their Mitigations

A trap at this stage is to only present the solution that best appeals to the project team, before a final report, reach out to stakeholders, present the alternatives and get their feedback. After working the project, a tunnel view of the best solution (built on bias of expected/desired outcome) can restrict best interpretation of the data. Bring in fresh eyes to break the fog.

Communication must loop back to the original question. I’ve been in many presentations where the focus has been on a beautiful dashboard or immediately dives into a complex method. Loop back to the question that was originally asked. Present the hypothesis you tested and give a high level view of what you found.

Data can be useful but it will mean nothing if you don't tell a story with it to the right audience. This is where those Customer Success skills shine! How do you leverage data to tell the story your audience cares about? My boss won't care if I tracked 20 metrics, but they will care if I show why those metrics mattered and how their progression affected our revenue. With customers, they don't care about our satisfaction rates, but they'll love a story of how satisfaction levels allowed them max usage and benefit from the product. What's the story and who are you trying to tell it to?

Know your audience. What is the preferred structure of the reporting to make it digestible? Are they a 1000 feet view type? Just want the highpoints? Are they looking to dig deep and have the raw data available for them to confirm the answers? Multiple variations of the reporting may be needed for all stakeholders to utilize it to their satisfaction

Some projects seem never-ending as the fixes can bring other issues to light, this is where scope and clear boundaries come into play. The results may trigger new projects that will need to utilize the data, here is 'need to know' access is important, confidentiality requirements are often overlooked but not all data is for wide distribution and to be used outside original parameters may need to be scrubbed to keep privacy requirements.

Having worked with customers, helping them make decisions based on data, i have seen one thing stand out often, Data presents the thin line of difference between process and people, and it's often the people. Do think through on #ChangeManagement right from start, even if you are doing a MVP level initiative around potential use of data or at the start of your #DigitalTransformation journey. Any impact across at organisational level can only happen when the change happens under business as usual scenario. Isolated projects hardly see light at the end of the tunnel. They remain a checkbox item.

In collecting and sharing data, it’s important to remember to ensure data privacy and adhere to ethical considerations when working with data.

I think, Data is the most available resource, the major thing is convert that data to information. To do that you should have a better view of your destination. For example your destination is to invest in a stock that is listed in sensex. The question is what data is required. 1. Determine the stock 2. It's performance in market 3. It's financial data which is generally called fundamentals. 4. Relativity with market 5. Finally market sentiments These are the data which are available but the right question is required here. "My thoughts"