FE Today Logo

Data and AI for a company

Khan Muhammad Saqiful Alam | December 10, 2023 00:00:00

2022-2023 have been a couple of very eventful years, with Russia's attempted invasion of Ukraine, Twitter's takeover by Elon Musk, several large crypto businesses' meteoric crashes, rounds of downsizing in tech giants, and the very recent humanitarian crisis in Gaza. But in the business and tech world, there was one event which I believe has garnered consistent attention, and has also transformed how we think our daily work will look like in the future - the advent of Generative AI, such as OpenAI's ChatGPT, Meta's Llama, and Google's Bard.

The future of work and business is witnessing a paradigm shift these couple of years.

In order to understand this transformation, we need to first look at how the situation was before. The field of Data Science and Machine Learning was popular across the globe from the outset of the 21st century, when we figured out that our cell phones packed more processing power than a mainframe computer of the 1980s. The key concepts of Machine Learning, Neural Networks, and Deep Learning, were already out there in academic papers and theoretical possibilities, but with this tremendous processing power, these concepts were not only being tested in labs and in the annals of theory, but also were becoming potential tools and techniques to be used in businesses to boost their profitability and growth. Companies such as Amazon and Netflix adapted recommendation models to match the right product for the right customer, superstores like Walmart were using data to achieve cost efficiency in every step of their business model, and FinTech, EdTech, Delivery Tech, and Social Media industries were emerging strong, standing on the shoulders of these practical applications of data.

Even in Bangladesh, companies such as bKash, Pathao, and 10 Minute School started making the best use of data science to grow their products, and the classical manufacturing and consumer goods companies were starting to ask questions like how they could best utilise data to increase profitability in the early 2010s. As a consultant and adviser on digital and data transformation, I was helping several local companies transform physical transactions into data and generate insights, optimise distribution and sales networks to increase sales and reduce cost margins, and track business and credit fraud. Companies such as Intelligent Machines emerged to build end-to-end AI solutions for businesses.

But amidst all this progress, there was a key challenge - broad-level understanding, across different levels of an organisation, on what data can do for them. Especially, in my experience with working and having conversations with mid and large-sized companies in Bangladesh, the challenge was further aggravated by the separate aims and objectives of the tech and business teams. To elaborate, a lot of tech teams in organisations, (BI teams, Analytics teams) were very coding and system development-focused, with the key performance indicators (KPIs) of building a certain number of dashboards and models for the organisation. Normally these teams didn't have a sales or growth KPI, or any KPI that directly connects to the profit and loss (P&L) of an organisation. On the other hand, sales, distribution, and product teams had direct P&L KPIs, but a lot of these teams did not have full visibility of the data capabilities of the tech teams and also did not have a clear idea of what tools and models could best suit them. A similar challenge also plagued the senior management of these companies - they were well aware of the need to be data-driven and had a good sense that data transformation would lead to profitability, but the challenge remained that they could not visualise what successful data transformation would look like (compared to the effects of a marketing campaign, or securing a new market).

After OpenAI's unveiling of ChatGPT, the challenge for organisations to imagine and understand the right use of data can now be overcome. As generative tools like ChatGPT and MidJourney are becoming more and more commonplace, now it is possible for everyone to test out the fruits of implementation of data and models. Besides correcting our writing works, consolidating information from the net, giving us quick formats and advice in common situations, generative AI tools are capable of doing basic analysis of data and suggest context-specific solutions. Today, companies are slowly visualising the optimum implementation of data - all of the data of a company, from different sources such as production and marketing, hosted on the company servers, and a generative AI model built on top of that so that a CEO can ask questions like - "What was the ROI (return on investment) for having endorsed a cricket player in the BD team in the Cricket World Cup?", or "What was the negative impact on regional sales centres after a recent bad response to a social media campaign?", and the company's own generative AI model will be able to build a preliminary quick answer and report to the CEO! This is not a science fiction dream anymore, rather a potential possibility within the next few years. And this has been a key contributor to data adoption for a lot of companies, as now they can see what really they can expect from being data-driven.

But being data-driven is easier said than done. And given the nature of investment required in data structures, teams, and software capabilities, investments to make use of Data and AI come with high Capital Expenses. To get the best ROI out of such an investment, an organisation needs to take a broad and strategic approach, which I will be detailing in the last part of this article. This approach cannot be just a KPI for the CTO/CPTO and the product team, but rather should involve and have the buy-in from all the CXOs. There are a few parts to this approach - Data Audit, Data Strategy, and Data Culture.

A Data Audit is the first and most important step to make the best use of data. A lot of times, organisations make the key mistake of jumping to analyse whatever data they have, investing in the recruitment of analysts and tools, later to complain that the ROI of these investments was not met. The rather cautious, but almost guaranteed successful approach is to carry out a Data Audit. Despite the fancy term, the task is very simple. The company's senior management and their department heads should look into the business model and the business units, understand what they are looking for, identify the objective of being data-driven, and then identify what data they need to achieve the objective, and whether they have the data or they need to develop systems for collecting that data. In such contexts, one challenge can be that the company finds it difficult to imagine what outcomes they are looking for from being data-driven, which is where the company can avail the services of Data Transformation consultants.

The next step is to build the Data Strategy. This includes collection, storage, analysis, model building, and deployment. If the data audit is done properly, then using the objective as the end goal, the strategy for each of the stages can be built up. Usually, this is when it pays off to recruit a data team and a good Data Transformation Manager, with good project management and cross-functional work capabilities. Also, this is a good time to invest in training. With the strategy being developed for each of the stages, organisations will also have a clear idea of which teams will be required in which areas. And this will give organisations a clear idea of training and skills requirements, rather than investing in generic training courses and online platforms, as normally in cases of data upskilling, generic courses are too introductory in nature and specific team-based and need-driven training programmes have given a better ROI.

But in order to make the most out of the Data Transformation that is initiated by a Data Audit, companies will need to work on the final piece of the picture - developing a Data Culture. In simple terms, a company with Data Culture looks at driving day-to-day decisions as well as mid and long-term strategies based on hard evidence from data. This is more of a challenging step to implement, as it will again require a company-wide change of mindset. Ground level decision makers will only focus on justifying their claims and decisions with data when their line managers and senior management will ask for data to support any suggestions, claims, and plans. There are a few ways an organisation can create a data culture. One is a top-down approach, where senior management leads the way by asking at every significant decision point - "Do we have data supporting the need for this decision? Is the hunch or anecdotal evidence supported by any data?" The other is a more bottom-up approach, where from the ground level, teams can be trained in what data is relevant to their work and function and is being collected, and what nature of analysis they can do and the data team can support.

Data and AI transformation, AI implementation within an organisation, are most of the time not plug-and-play solutions. A company can go for a quick plug-and-play or outsourcing approach, but more often than not, such an approach gives a 'tip of the iceberg' return. This is because of the operational and strategic nature of how the world of Data can impact an organisation. Apart from the fact that AI tools can help us with our day-to-day work, it is evident how much a company stands to gain with the right implementation of Data and AI. In this article, it has been aimed at showing the readers how generative AI has given businesses a clear picture of how data can drive profitability, and the key things to consider to make the best of investment in Data.

Khan Muhammad Saqiful Alam is a Data and Digital Strategy, Growth and Tech Project Management professional. At present he is leading the corporate training wing of LightCastle Partners, as a Senior Expert.

[email protected]

Share if you like