Personalized Marketing is Not Selling; It Requires Scale


CREATOR: gd-jpeg v1.0 (using IJG JPEG v80), quality = 100

How many marketing technology vendors say they help you personalize your marketing to customers? Hundreds? Maybe all of them? But different systems are made for different purposes. When it comes to personalized marketing, it’s important not to try fitting a square peg in a round hole.

Selling vs. Marketing: A Matter of Scale

Simply put, selling is a 1:1 interaction with each potential customer for a sale, one person at a time. Marketing is about communicating to a market, that is, to many potential customers at once. The goal is to move potential customers into the selling process.

“Personalizing the customer experience,” — for example by monitoring and responding to a customer’s browsing, inquiries, abandoned cart, or other behaviors during the buying process — is really selling. These platforms attempt to automate what used to be a personal interaction with a salesperson by interjecting messages or content during a sales engagement with the hope of encouraging a sale. While machines are nowhere near as good as humans in detecting customer interests, concerns, and objections, it’s better than leaving customers completely on their own. But it’s still one customer at a time.

Truly personalized direct marketing means individual messaging and recommendations to each of every customer targeted in a campaign. That’s the Holy Grail because it lifts your communication above the din of the over-communicated, static messaging barraging everybody. In our experience over many hundreds of campaigns, it’s proven to move more customers into the selling process. It works because each individual’s product recommendations, offer, or message plays to their individual propensities. The trick, of course, is accurately calculating the propensities of each customer.

The Demands of Scale

It’s not so hard to analyze an individual customer and compare them to others to arrive at predicted buying, loyalty, or product purchasing propensities, usually expressed as probabilities, e.g., probability to buy a certain product in next 30 days. Similarly, there are reasonable techniques to decide which customers should be more interested than others in a certain specific product. But determining the top six or ten products to recommend from among thousands of products to each and every one of thousands or millions of customers targeted in a direct marketing campaign — whether or not they’ve recently engaged in a sales process — that requires some sophisticated data science.

At that scale, something automated and repeatable for regular campaign scheduling is needed. Because the analytics is complex, most companies, and marketing systems, lapse into a segmentation approach. They gather a lot of information about each customer and try to identify clusters or groups with common attributes that appear to have similar interests. The biggest costs lie in all the data that needs to be managed and analyzed. It’s also stereotyping: It can be useful, but not everyone conforms perfectly to the stereotype. Truly personalizing product recommendations means treating customers as individuals. To do that, you need to analyze them individually. For accuracy at scale, you should use a process or platform built specifically for this purpose.

The Magic of Probability

Calculating accurate propensities is critical because it only works on a large scale. It’s not a question of predicting who will buy and who will not. Nothing actually predicts the future. It’s about calculating a probability, such as a “10% chance of making a purchase in the next 7 or 60 days.” In this case, the overwhelming odds are that this customer will not make a purchase during that period. But this is just one customer. In a group of ten customers with a 10% chance of buying, one can be expected to buy. And all the customers with a 10% chance are twice as likely to buy as those with a 5% chance.

That’s powerful information.

It allows marketers to target communications at the most likely, or least likely, buyers, or those on the fence, or those with the highest probability of churn, or those expected to spend the most, or those most likely to be interested in a certain product. And by making product recommendations with the highest probability of interest to each individual, as opposed to a static product offer to all customers, the probability of response shoots up. It’s not that any individual buyer is likely to buy the recommended product (the probabilities are usually low), but the more people across the entire campaign that see better recommendations, the more that will respond.

In fact, all other variables being equal, it works every time if applied at proper scale. If you throw a die six times, the probability of throwing a 5 should be one time (1/6). However, it actually could come up zero or maybe three times. But if you throw a thousand times, a 5 will come up close to 1/6 of the time. If the propensities are accurately calculated, and enough people are exposed to better recommendations, they buying rate will eventually converge on those higher probabilities. Even if the probabilities for a given individual customer are really low, they are still higher than the static product being offered to a random customer. It’s just math. It works.

This is why personalized marketing is so important and so powerful. Its true power does not emerge for specific sales engagements; but at direct marketing scale, it’s magic.

Peter Moloney is CEO of Loyalty Builders, whose Marketing Lift Service offers a simple, cloud-based predictive analytics service enabling marketers to get revenue lift from more relevant communications to their customers. Request a Free Customer and Revenue Analysis here.



Tags :