Therefore, I created this 9-Step framework for companies to use as a starting point for the, ‘Should we implement AI?” discussion.
Step 1: The first step is choosing a department or area of concentration you feel AI might be useful. Is there a department that is lacking in data, efficiency and/or results?
Step 2: Start out by asking yourself what type of problem do you wish to solve within this department.
· Sales performance
· Marketing campaign effectiveness
· Customer Service issues
· Inventory management
· Generating real-time reports
Step 3: List out all the steps that are used today to solve that problem (e.g., the steps in the process)
Step 4: Do you need AI or is there some more direct process optimization approach that would work as well? In other words, can you look at your process and make changes to how things are done? If you feel that your processes are already streamlined and could benefit from AI, then proceed to the next step.
Step 5: Having listed all the steps in your process in Step 3, which step in the process do you feel is impacting the process negatively? We’ll call this step in the process your ‘broken link’ in your performance chain.
Step 6: Collect data on the broken link to fully understand and quantify the problem. This data will be your KPIs (Key Performance Indicators) to measure against once an AI solution has been put in its place.
Step 7: Identify what AI application is available in the market to solve your problem.
Step 8: What ‘tech stack’ (e.g., a combination of tools) will I need to implement this AI solution? In other words, what technology or applications will you need to implement this solution? Research features and benefits in your list of potential applications.
Step 9: Run an ROI calculation.
For example, let’s say you’re having a sales performance (Step 1-2) issue because your salespeople are spending too much time researching clients and writing personalized emails (Step 3). You’ve created a template to reduce the time it takes to write an email but it’s not enough (Step 4). You find out that each email is taking the salesperson 20 minutes to produce (i.e., to research, write, personalized) and you estimate that you’re losing 30% sales productivity each day which you estimate is costing you $25,000 per salesperson per month (Step 5-6). What you need is an AI application that can use information from either past conversations or online postings to construct a customer response with minimal, if any, salesperson intervention (Step 7). To do this, you’ll need: a Customer Relation Management (CRM) system to capture all your customer interactions, a Natural Language Process (NLP) application that will allow calls to be transcribed and social media posting to be collected and structured, and a Natural Language Generation (NLG) tool that will take email, phone call transcriptions and social media posting and construct customized responses (Step 8). The last and final step would be to assess how much time, money and effort will go into implementing the tech stack and comparing it against how much business you’re losing along both short-term and long-term (Step 9).
Check out: “Sales Ex Machina How Artificial Intelligence is Changing the World of Selling”, by Victor Antonio and James Glenn-Anderson