Why the Same Product Gets a Different Price for Every Customer
In B2B quote histories it is not unusual to find the same product, at similar volumes, priced double-digit percentages apart from one customer to the next. Most of that gap comes from discretionary discounts by individual sales reps. "They've been with us for years." "We need to hit the quarter." Each call seems reasonable on its own, but together they add up to price leakage that nobody sees at the company level.
Cost changes also reach quotes too late. Raw material prices and exchange rates move, but price lists are often refreshed only once a quarter, so deals won in between are signed at thinner margins.
Price has more leverage than most teams realize. A McKinsey analysis published in Harvard Business Review in 1992 found that a 1% price improvement lifts operating profit by about 11% on average. Take a company with ₩10 billion in revenue and an 8% operating margin: recovering just 1 percentage point of leakage raises operating profit from ₩800 million to ₩900 million, an increase of 12.5%.
B2B Price Optimization Is Not B2C Dynamic Pricing
Airlines and e-commerce platforms reprice in real time using huge volumes of transactions. B2B works differently:
So the goal isn't the highest price the market will bear. It's the balance point between win probability and margin.
Data and Model Design
Getting the Data Ready
Start with past quotes and whether each one was won or lost. Quoted price, final price, volume, customer tier, industry, lead time, competitive situation and cost at the time all need to live in one table. If loss reasons were never recorded or quotes are scattered across spreadsheets, this cleanup can take up half the project.
Win Probability and Expected Margin
A win-probability model (logistic regression or gradient boosting) estimates how likely you are to win a deal at a given price under given conditions. Multiply that by the margin and you get expected margin. For a product that costs ₩80:
The best price is neither the lowest nor the highest. It's ₩100. How much the win probability moves as price changes is your price elasticity.
Presenting a Price Band
Where a product has too little data of its own, group similar deals by customer tier, volume and product family, then look at the distribution of contract prices. Set the 25th percentile as the floor, the median as the target, and the 75th percentile as the ceiling. That gives sales a band they can actually work with.
Putting It to Work in Sales
How to Roll It Out, and How POLYGLOTSOFT Helps
A three-step rollout is the safest path:
POLYGLOTSOFT designs this process with you, drawing on our work building predictive and classification models and integrating ERP and CRM systems. With our subscription development service, the model and the quoting screen are refined together every month based on feedback from your sales team. If your quotes vary by who happens to write them, a price leakage diagnosis is a good place to start the conversation.