Pricing strategy can influence customer behaviors. Data Scientists must evaluate their pricing models from a systems perspective. Let me explain with a few examples.

When I first started my business, I did a lot of traveling to clients. After a couple of years, I wanted to spend more time at home, so I created a pricing strategy to incentivize clients to let me work from home.

I offered on-site and off-site hourly rates. On-site was my standard rate, and off-site was 50% higher. In 3 months, clients realized I was just as effective off-site as on. My traveling frequency dropped from monthly to quarterly.

I created a tiered pricing model that forced clients to feel cost when I did. Before that, there was no difference for my clients when I was on-site vs. off-off site. Perception of value was the only driver of on-site demand.

I could have tried talking to clients and convincing them to see things my way, but pricing did the talking for me. Why increase on-site instead of decrease off-site prices? You must take a systems view to understand my reasoning.

If I had dropped prices, I would have reinforced their belief that I was more valuable on-site than off. By raising prices for on-site, I told them, "It costs more for me to be on-site, and I'm passing those costs on to you."

It's critical to align my brand message with the pricing strategy. The message has to fit my prices, or changes in pricing strategy can have unintended consequences and significant downsides.

Pricing strategy is a powerful tool if businesses use it with a systems view but will erode margins when pricing is implemented tactically. Models will reinforce the bias of partial data in the same way. How?

A model trained with only sales data is biased towards the sales team's objectives. Success is defined 1-dimensionally, so the model is optimized for sales at the expense of all other success metrics. Here's an example.

I worked at a slot machine manufacturer. Their sales team would offer discounts at the end of each quarter to make their sales goals. Can you guess what happened?

They trained customers to wait until the end of the quarter to buy. Sales-centric pricing sent the message that slot machines were overpriced at the beginning of the quarter.

Margins took a hit based on the 1-dimensional optimization but even worse was the increased costs of manufacturing. Orders came in all at once, so they had to be built and delivered all at once.

The manufacturing line was slow for 2 months every quarter and overloaded for the last few weeks of every quarter. That led to staffing issues, overtime costs, and supply chain instability.

Field technicians had to install all the machines before the end of the quarter so they would be included in the number of units sold and installed. Both drove the company's share price.

Pricing sent investors and analysts an unintended message too. The business is overpromising on its projections and is in danger of missing every quarter. Pricing was making sales and strategy leadership look inept.

Using periodic discounting, retailers like Macy's have done similar damage to their margins. They trained their customers to wait for clearance sales. That impacted sell-through and margins.

Customers got the message that merchandise was overpriced until it was put on sale. As Macy's tried to build a brand image like Nordstroms, its pricing strategy was undercutting the perception of prestige.

Stores looked more like JC Penny's or Sears than Nordstroms during sales. That's what customers started associating their brand with.

Pricing is interconnected with every part of the business, making pricing strategy a complex system. In Data Science, our pricing models must pull features from nearly every aspect of the business to optimize pricing.

High ROI Data Science

Pricing Can Train Your Customers, For Better…