Many businesses have been incentivised to invest in AI, encouraged by the possibilities of automating repetitive tasks, producing deeper analytical insights, and increasing efficiency.
In the retail and e-commerce industries, AI is often considered as the path to creating a dynamic, personalised, and frictionless customer shopping experience on the web and in physical stores.
To make it happen, businesses first need to generate a massive amount of data on their customers, employees, products and services, and internal operations. Without continually feeding their AI software healthy doses of high-quality raw information, businesses risk failing to use their new tools to replicate the successes that steadily increase revenue while building customer loyalty.
AI shouldn’t be tantalising because it’s trendy; it should excite business owners because of its potential to unlock real value across business sectors by more efficiently capturing, managing, and analysing the vast and ever-increasing amounts of data. As you scramble to be in the majority of companies now implementing AI components to their business models, you should first assess what your actual business goals are—and then ask yourself four critical data-driven questions.
Regardless of which type of AI software you implement—reactive, limited theory, the theory of mind, or even one that strives to be self-aware—your AI solution will only be as good as the data it’s fed. If you put bad data into the system, you will receive bad analytics, which can quickly lead to costly and regrettable business decisions.
The first step to having good data is to have clean data—and the first step to that is always to know where your data lives. As companies spread out and begin to scale, they often end up collecting data in multiple systems: CRM, ERP, even individual spreadsheets.
Data may get flung across the organisation to serve an immediate business user need, but as it grows, it is more likely in the aggregate to become contaminated with conflicts, redundancies, trivialities, and duplications. If this data remains unsorted and cleaned, it will lead AI to make conclusions based upon errors. But if the information is stored in a “data hub,” it will be crisp and clean and ready to serve AI properly.
They don’t call it “big data” for nothing. While big data might now have a dozen definitions, its lowest common denominator can still be traced back to what Gartner’s Doug Laney first identified in 2001 as the “3 V’s” of data management: data volume, velocity, and variety.
How does this apply to today’s businesses? Because our digital systems produce and store such large amounts of data, the inability to capture and collect all that you need could greatly hinder your data capabilities—which could have disastrous effects on your AI.
Simply put, if your AI is relying on a limited data source, its conclusions may be incomplete, and even its accuracies will be inadequate and unhelpful. This doesn’t mean you should mine your customers and employees for every bit of personal information. That’s counterproductive at best and, in a world where starting May 25th the EU’s sweeping new General Data Protection Regulation (GDPR) will affect companies worldwide, financially crippling at worst.
What it does mean is that if businesses want to optimise your AI by using predictive analytics, to avoid faulty conclusions, you’ll need more than a tiny slice of data. For instance, if you want to know more about your customer’s purchase history, as well as the details of those products, you’ll need either a robust loyalty program or troves of third-party credit card data. If you have multiple customer channels and touch-points, you’ll need contemporary collection on all of them, to ensure you aren’t excluding the most critical data.
It sounds like a tall order, but the good news is that, in today’s digital world, even small businesses can use big data.
At its most sophisticated, AI today is about pushing machines to complete tasks that are the reliable domain of humans, like driving cars and recognising objects. But, as analysts are quick to point out, how useful these tools will ultimately be is a dark unknown.
This counterpoint is equally important for businesses considering implementing AI. Just because AI is dominant does not mean it is better. While there are plenty of things AI can do more quickly, accurately, and safely than a human—search the web, translate languages, create a scientific theory, diagnose certain diseases, survive deadly environments—there are plenty of things it cannot, like read for comprehension.
You may trust a chatbot to handle routine customer queries—one European telco found that chatbots could resolve 82% of calls autonomously—but would you trust it to speak with a dissatisfied customer? What about a customer whose needs simply fall outside his or her data points?
When you’re thinking about implementing AI, consider how it could be a tool to complement your shrewdest and most insightful business practices, not a cheap replacement for a valuable system or employee. Don’t go AI for AI’s sake.
A business practice could be defined as a complex set of social and economic arrangements meant to deliver a product or service to a paying customer. A technological tool greases the arrangement, allowing it to happen swiftly or slowly, lightly or cumbersomely. Even with the rise of devices like software-as-a-service (SaaS), a company continues to be driven by its business goal, underpinned by a unique set of business values (could link to Salah’s FTC “Third-path company”, once published).
If you fail to consider the business case for implementing AI, you may be surprised to find that your AI fails you. This could have an adverse ripple effect throughout the organization, with employees wondering why they took on so much technical and intellectual debt to power a shiny new AI tool that either under-delivered or delivered frustrating outcomes.
Just as businesses in the age of big data have had to ask themselves what they’re going to do with all this new information, businesses today and in the future are going to have to ask themselves how the possibility of AI alters their visions. If you’re not choosing an AI solution that works best for your needs, the best AI in the world won’t optimise your supply chain, deliver better customer service, and reach new markets. AI works best when given a specific set of rules. It can do great work, but it needs the company to point it in the right direction.
Author: Salah Kamel, founder and chief executive officer at Semarchy
Image credit: Fotolia