Why Artificial Intelligence Isn’t Always the Answer

Why Artificial Intelligence Isn’t Always the Answer

When to Apply AI, a Human or a Hybrid Approach to Improve Business Processes

While we can all agree that AI is certainly a leading technology trend being tapped to pack more power into software applications,business processes and analytics, the real challenge lies in determining how to best apply it to your business, if at all.

AI is rightfully capturing the attention of technologists today for its obvious business benefits, but it is often not clear whether you’ll want to assign a task to a human, AI or take a hybrid approach.  Factors, such as your industry, your company size, your target customer and growth plan will dictate whether you outsource help, implement within or simply purchase software to power up AI at your business.

In order to avoid getting caught up in the AI hype cycle, start by reviewing your KPIs (Key Performance Indicators) and compare your results against industry standards.  Now that you’ve identified the areas of your business that are in need of improvement, you can move on to the next step.  If you’re not reviewing KPIs on a weekly, monthly and yearly basis, I’ll have more on that topic in a moment.

Next, you’ll need to conduct a careful and thorough analysis of your business processes to determine if AI is the solution you need.  You may want to consider outsourcing an analysis and assessment to an expert to help you since AI is a constantly changing and emerging field of study.  Once you are ready to implement AI, it’s imperative to have someone internal dedicated to moving the project towards completion.

Now, if you’re not sure where to start to determine what areas of your business could benefit from AI?  Start by looking at your KPIs (Key Performance Indicators), such as:

  1. New customer acquisition rate – if you’re not looking at this, I would recommend doing so and regularly.  Also review your acquisition rate for the same period prior year and YOY (Year Over Year) to understand your growth rates.

  2. Customer churn or return rate – if you are a service based company, how many customers are you losing a month, a year, etc… If you a manufacturing company, for example, how many customers return your products (and if so, track the numbers for every single product in your CRM and why it was returned) and how many return to buy from you every month or every year?  Track all of it (down to the product category, brand or SKU), everything!

  3. How long does it take your sales team or chatbot to close a deal? If unable to close the deal, track who lost it and why.  That way, if there is a product or feature that is continually a loss-leader, you’ll have captured the insights in a lost deal report.

  4. What is your average deal size or in the case of an eCommerce online store, what is your average time spent online and spend per visit?

  5. For marketers, your team should be able to report on the cost per lead and conversion rates.  For companies that are heavily dependent on their website and content to drive inbound traffic and leads, what is the average time spent per visit, number of monthly visitors, traffic sources and best performing content?  What is the best performing lead source?

  6. For your support call center, what is the average time to resolve an issue?  What product, service or bug is driving support requests? How is each of your customer service reps rated? What is the customer’s satisfaction rate with their issue after it is solved?

The analysis and determination of whether a task or process should be the job of AI, a human or both should include financial modeling to determine which is the most cost effective (this is your hint to involve your finance colleagues). Perhaps the most effective approach to take is to assemble a team of individuals from various departments to aid in the decision-making process, rather than leaving the decision up to one person. If your company has IT governance, you’ll want that person on your team. Ideate an application, test it and repeat until you find a solution.

Before I jump into a few examples to demonstrate how AI is being used today, I think it’s worth mentioning the most common industries that AI is being applied to today. According to a new study released on July 21st from Accenture, “The introduction of AI could lead to an economic boost of US$14 trillion in additional gross value added (GVA) across 16 industries in 12 economies.” They added:

Of the industries studied, information and communication, manufacturing and financial services are the three sectors that will see the highest annual GVA (Gross Value Added)  growth rates in an AI scenario, with 4.8 percent, 4.4 percent and 4.3 percent respectively by 2035. This translates to an additional US$6 trillion in GVA in 2035 for these three sectors alone. Even labor-intensive sectors such as education and social services —where productivity growth is traditionally slow – will see a significant increase of US$109 billion and US$216 billion in GVA respectively.

In addition to a few industry examples of applied AI below, I’ve provided some examples of its application to pure functional areas as well:


eCommerce and AI are a profound marriage, especially when applied to returning shoppers on Amazon.com.  This is about to truly become interesting post-Whole Foods acquisition.  One of the reasons Amazon has built a fairly iron-clad customer retention model has much to do with its underlying AI-powered recommendation engine.  Check out this Jimmy Kimmel bit where he satirizes  Amazon for its incessant product recommendations (which clearly work), requests to submit a product review or sign up for Amazon Prime.  

Customer Service

In terms of functional areas, customer service and sales teams at many companies are being enhanced with AI across practically every industry with the onset of chatbots. Chatbots can close smaller, transactional deals entirely on their own.  And, AI out-of-the-box applications that utilize NLP (Natural Language Processing) to analyze a recorded phone call, for example, can help customer service reps and salespeople become more effective by providing feedback post call, thus improving customer satisfaction and close rates.


Keep in mind that today, digital marketing is the leading tactic marketers rely on, whether you are a local, national or global business.  Advertising and marketing automation tools are primarily executed through web-based or SaaS applications.  CRM market leader Salesforce recently acquired an AI company that extracts relevant data from the web on prospects, such as personal social media profiles, and better segments them so that marketers can serve them the right content on the right channel at the right time.  Serving up the right value at the right time to your prospects helps to bring them towards the “ready to purchase” stage.

Undoubtedly, every company today should be utilizing a CRM (Customer Relationship Management) to house their prospect and customer data. As a B2B company with the potential for national or global reach, email marketing automation is a must as well.  If you want to excel at whatever it is you do if you are this type of company, a competitive edge requires adding AI to the marketing mix right behind having a CRM + automated email and social media marketing.

Human, Hybrid Approach or AI?

These same concepts apply to the aforementioned examples of AI in business.   It will require human judgement and thought to determine whether a task or job requires only AI, a human or the hybrid approach.  It’s even likelier that the hybrid approach will be more common with automated workflows running in the background to link together processes. Consider the following from one of the world’s experts on the subject in a recent Verge article:

Professor Manuela Veloso, head of the machine learning department at Carnegie Mellon University, envisions a future in which humans and intelligent systems are inseparable, bound together in a continual exchange of information and goals that she calls “symbiotic autonomy.” In Veloso’s future, it will be hard to distinguish human agency from automated assistance — but neither people nor software will be much use without the other.

The most sophisticated applications of AI in business offering tangible benefit today reside in functions like marketing, customer service, finance and analytics, as mentioned earlier. In terms of chatbots performing sales, marketing and customer service functions, today many companies use them alongside humans to handle overflow or are used outside of regular business hours.

As of now, chatbots seem to be capable of only answering closed questions, such as “What is my balance (checking account)?” or very direct statements like “Pay” to allow your customers to pay their monthly mobile phone bill with the preferred method of payment on file.  As the the difficulty of the question or need increases, the chatbot will shut down and a human will need to step in to help.

In terms of analytics of big data, AI can do the job much faster than a human attempting to aggregate spreadsheets and convert the data into meaningful visual infographics.  A more common challenge for many businesses is the many data silos that exist as a result of the complex and multiple software applications relied on to conduct business.  

Today, humans are often performing tasks that are not purely solved by AI, but instead by a workflow to push data from one application to another.   For example, a salesperson has quoted a customer and they are ready to order.  The salesperson then manually enters in your CRM that the order has been accepted and pushes a button to release that information to billing to generate an invoice.  Instead, the customer can electronically accept the order and a workflow can approve and change the order status and send it to distribution to be prepared for shipment.

In summary, it’s safe to assume nothing except that humans and machines are dependent on each other to be more successful when working independently of each other and when working together in improving business processes. There isn’t a one-stop solution or answer when it comes to when and how you bring AI onto your team.