Earlier last year I wrote about how siloed, manual data analysis by analysts, data scientists and lack of right causal data is adversely affecting quality and timeliness of decisions in marketing and trade promotions, causing large-scale inefficiencies (last year’s blog). Automation and Augmentation of data-driven analysis in these processes that consider the impact of all market factors at the time of decision making will have a huge financial impact.
We are now entering 2018 after a year of greater awareness gained from the evaluation and experimentation with Applied AI products by Enterprise buyers. A number of companies are looking at data initiatives as pre-requisites for AI success e.g. building a new data lake on the cloud, which is actually a long drawn out project with an uncertain ROI. Having been in the data, BI, analytics and AI Industry for long enough now, I have been surprised at how data rhetoric around AI is similar to BI.‘’AI is as good as the data you have! ‘’15 Years ago, that slogan was ‘’BI is as good as the data you have!!‘’ In the process, enterprises are buying more data and cloud solutions rather than AI!!
In many ways, business and technology transformation is an iterative, agile project and the idea of building a new, shiny, data infrastructure ( e.g. on the cloud) first before you should adopt AI is actually putting the cart before the horse.
1) A data warehouse or data lake may not be the best place for high velocity and variety data, which is where the Industry 4.0 data paradigm is moving.You may draw this data you need for AI directly from existing enterprise systems or business process, and it's more efficient and economical.
2) Another issue is the rapid loss in the relevance of data. 30% of data in marketing decay every year. It is important to recognize why most recent data in near real-time and lowest level of granularity could be more important than historical months/ years data for quality and timeliness of decisions.
3) A lot of data that you may need for AI systems that augment decision making is going to be external, market, customer data on top of your usual internal, operational data.
4) Establishing AI use cases for business benefits is the best way to determine what data is crucial, where is the data challenge, and what is the ROI of investing in a data initiative.
I believe, it would be important for companies to run their enterprise and data transformation initiatives driven by AI adoption and strategy first ( rather than vice versa).
So, I would strongly recommend that enterprises focus on AI adoption and follow a lean methodology like below:
Identify high impact use cases for optimisation, efficiency and business growth e.g. Trade & Retail promotions optimisation, Personalised marketing, Intelligent Demand forecasting and supply planning.
Plan Proof of Value, Learn and Expand:
Scope Adoption at limited scale ( Prioritise Category, Channel, Area)
Understand ROI, data, business process, and people challenges
Define a better strategy to magnify competitive advantage with AI and unique data
Choose a strategic AI partner that brings deep knowledge of AI and Data technology as well as the product with a deep understanding of your use-case.
An Early AI adoption can shape your business and data organisation in the right direction leading to rapid ROI and long-term competitive advantage.
What do you think?