Are you deploying the right models? 

Now that consumers expect speed and hyper personalization in all things, marketers have to find innovative ways to meet demands and maximize their budgets. To do this, marketers are turning to artificial intelligence and machine learning. In fact, there is a new term just for this – AI Marketing.

Customer expectations have never been higher. Amazon, Netflix and Google have set the standard for what customers have come to expect from technology and marketing. Amazon takes your order in one click and delivers it next day. Netflix wades through years of your entertainment choices and immediately suggests the next shows you’ll want to binge-watch. Google corrects your spelling, programs Alexa to tell you when that Amazon package is arriving and provides you with instant answers to the most obscure bar bets.  

AI Marketing, as shown in these examples, leverages technology to collect data, develop customer insights, anticipate next best actions, and make automated decisions about marketing efforts. If your goal as a marketer is to drive revenue, help lower costs through efficiencies, and drive customer engagement and satisfaction, AI marketing can help you accomplish all of those things. 

Let’s explore 3 areas where AI Marketing can be helpful and what you should know about each area before starting any project. 

For each project, we’ll briefly explore what it is, how it works for marketing, and any pitfalls – technical or cultural — that you might need to be aware of in applying it.  

1. Marketing Data Management: 

What it is. Marketing data management is the process of collecting and handling marketing data, competitive intelligence and market research information. This function should not occur in the IT department – this is at the heart of what marketing does. Determining who the best buyer is for your product or service is clearly a marketing function. Collecting and managing the data associated with your buyers is marketing’s first consideration. What do you know about your customer? How many of them do you have? How do you describe a customer? Which ones buy which products or services? How large is the entire market for your product or services? All of these important marketing questions are answered through marketing data management.  

How it might work for you. The use of AI and machine learning in this area can be applied both at the macro and micro levels. At the macro level, you can deploy AI and machine learning models to understand how your entire customer base segments into specific buying groups. At the micro level, you can predict a product’s lifetime value and associate it with individual customers. This micro-level data analysis helps you determine which customers or prospects are the best to pursue with which products. Accumulating data from these efforts only helps to make your models stronger and more accurate.

Accumulating data also requires that you manage the quality of the data you collect. Machine learning can be deployed against large datasets to deduplicate records or provide adjustments to standardize fields like zip codes or addresses. ML is also useful in helping to organize datasets for use in other AI applications.  

Other uses of machine learning include techniques like web scraping. This process is handy when trying to understand your competition. Each competitor’s website usually contains information that can be accumulated via this method such as new products available, customers mentioned and special programs. This is all public information, and with the right algorithms, data scientists can glean basic information about existing, as well as emerging, competitors. 

Read next: Why we care about AI in marketing

Things to look out for.  There are hordes of tools and consulting agencies in the market that want to help you with marketing data management. Tools include a wide range from Google Analytics to SAS, each providing a particular capability. Understanding what you want to accomplish – market segmentation, competitive analysis, etc. – will help you decide on tools or agencies that can support you. Getting your marketing operations lead involved is also a good idea.

When beginning marketing data management projects, consider first the purpose for managing your data and then look for the tools that are best in doing those identified tasks. When engaging consulting agencies, look for those that have experience in your area of need.  

2. Customer Intent

What it is. Customer intent data is sales and marketing information derived from observing the actions of the customer when accessing online content, looking at competitors, registering for events, contacting analysts, or engaging in any number of social media activities – from searching the web to posting on LinkedIn. Nearly every marketing organization today depends on this type of data to some degree, but it often doesn’t work for all marketers. 

How it might work for you. From the data collected about each customer’s interactions with your brand, website or staff, statisticians and data scientists can make inferences about the interests of the customer and their intentions to engage and purchase from your company. These inferences can be helpful in positioning to customers the right product at the right time. 

Once an algorithm is developed for identifying these customers, it’s imperative that you also gather input on the output of the AI model from the sales teams who will use this information, as well as from the marketers who might be applying it to online campaigns. Test the output of the model, but also test how sales and marketing is using it.   

Things to look out for. Sources of data are most important in determining intent. You already have good information about what your customers purchase, when they purchase, who they buy from and what type of company or individual is buying.  But intent data relies also on the actions that your customers or prospects may do before the actual purchase.

For example, this may require your AI algorithm to make connections between an inquiry on your competitor’s site and your prospect or customer list. There are firms that can provide contact-level intent data that identify an actual person taking an action. This information is helpful but must be used cautiously to avoid the “creepy” effect.   

Also, when using intent data, remember that it is only directional – it’s not specific or actual.  If your sales team uses intent data, they will need training on what the information actually means. For instance, Identifying a CIO who is likely to purchase an ERP system in the next 30 days may only mean that the CIO has begun a year-long process to identify a system. Giving this to sales as a lead without the explanation could be a blow to your marketing organization.

3. Opportunity and purchase prediction.  

What it is. Forecasting is a way of predicting what will happen in the future. For example, you can forecast what the sales of products and services may be in any given period.

Sales forecasting helps management plan for expenses, business growth or economic downturns. It’s the crystal ball that sales managers use when predicting whether they will make their targets or not. Sales forecasting is usually fairly accurate because it uses past sales transactions to predict future ones.  

How it might work for you. Marketing can use predictions in their work as well. For example, Norway’s tourist department uses AI methodologies to predict how many tourists will visit the country. Although not a sales figure, it is an important KPI for Norway tourism. AI or advanced statistical analysis can also help predict attendance at events, numbers of people who will take you up on a special offer made on your website, or the number of qualified leads that will make it through to purchase. 

Things to look out for. Forecasting can be very rewarding, but it is only useful if it proves to be accurate. Here are a few tips:  

• Consider more than just last quarter’s numbers. Good sales forecasting has at least 18 to 24 months of company performance data.  Working with that much data allows you to be more precise in your forecasts.  If the data is not available, avoid forecasting.
• Account for change in your overall business. Good forecasting accounts for the sale of the same product and service over time. Acquiring new products to sell, divesting of products and changing pricing or strategy all effect your ability to accurately forecast sales. Also, if you are predicting other marketing events, one of the variables that is often important is the budget allotted for an activity. If that varies greatly from quarter to quarter or year to year, then it may be more difficult to forecast, or you may need to allow for these variances in the model.
• Don’t try to forecast sales into new markets with new customers. No matter how tempting it may be, you need performance data to forecast sales. Leave this forecasting to your sales teams. This is often considered business development, and these sales teams know how to evaluate whether a customer will purchase or not.  For marketers, this is a matter of collecting the information from the sales team, developing a profile of a good customer and then applying “look alike” analyses to other prospects.

These are only a few of the key areas of marketing for applying AI and machine learning techniques. As you explore more in this world, you will find that opportunities abound especially in helping marketing to streamline the myriad decisions they make each day.

Opinions expressed in this article are those of the guest author and not necessarily MarTech. Staff authors are listed here.

About The Author

Theresa Kushner is passionate about data analysis and how it gets applied to today’s business challenges. For over 25 years she has led companies – like IBM, Cisco Systems, VMware, Dell/EMC – in recognizing, managing, and using the information or data that has exploded exponentially. Using her expertise in journalism, she co-authored two books on data and its use in business: Managing Your Business Data: From Chaos to Confidence (with Maria Villar) and B2B Data-Driven Marketing: Sources, Uses, Results
(with Ruth Stevens). Today, as the Data and Analytics practice lead for NTT DATA, Theresa continues to help companies – and their marketing departments — gain value from data and information.