Ask a commercial team what they charge and you’ll hear the list price. Ask what they actually collect, after on‑invoice discounts, rebates, payment terms, freight and promotional allowances, and the room usually goes quiet. That collected figure is the pocket price, and the distance between it and list is where margin quietly disappears.

Why it stays hidden

The data needed to build a waterfall lives in different systems: invoice discounts in the ERP, rebates in spreadsheets, freight in logistics, terms in finance. No single team owns the full picture, so nobody sees it.

What to do first

• Assemble one transaction‑level table that allocates every off‑invoice element back to the line item.
• Plot pocket price against deal size. The scatter almost always reveals customers of similar size paying wildly different prices.
• Rank leakage elements by total value, then by variance. High variance signals inconsistent policy, the easiest money to recover.

Price realization projects rarely need a new pricing strategy first. They need a single, trusted view of where today’s price goes.

An AI agent that can read a customer’s history, check inventory, and draft a quote in seconds is no longer science fiction. The temptation is to start with the model. The better place to start is the policy it must obey.

Define the box first

• Floors and ceilings by segment, product family and deal size, set by people, enforced by code.
• Escalation rules that route unusual requests to a human with the context already summarized.
• Explainability: every recommended price carries the drivers behind it, in plain language a seller can repeat.

Then let it learn

Within a well‑defined box, models can optimize win rate against margin, flag deals drifting below target and surface cross‑sell opportunities. Outside it, they create risk faster than any human could.

The competitive edge isn’t the model. It’s the clean data and clear policy that let you trust it.

“Our customers will pay about $50.” Statements like this compress a rich distribution into one misleading point. Some customers would happily pay twice that; others walk at $40. Pricing to the average leaves money on the table on one side and loses deals on the other.

Read the shape

Well‑designed WTP research, whether conjoint, Van Westendorp or analysis of real purchase behavior, reveals clusters. Those clusters are your natural tiers, your fences and your upgrade paths.

• Wide, flat distributions suggest good‑better‑best packaging.
• Bimodal distributions often mean two products are hiding inside one.
• Narrow distributions mean price is a commodity lever; compete on terms and service.

Segment first, then price. The architecture should follow the data’s shape.

Most teams treat positioning as a marketing exercise and pricing as a finance exercise. Buyers don’t. The moment a prospect decides what kind of thing your product is, they’ve already decided roughly what it should cost.

The category sets the anchor

Call a product a “reporting tool” and buyers compare it to cheap dashboards. Call the same product a “revenue intelligence platform” and the comparison set, and the budget line, changes. Neither label is wrong; each one carries a price ceiling with it.

Three questions to answer together

• What do we replace? The alternative you name in your messaging becomes the buyer’s price reference.
• Which value do we quantify? Outcomes you can measure in the customer’s own numbers support a higher price than features you describe.
• Who is the buyer? Positioning to a department head and to a CFO lands in different budgets with different thresholds.

Test them with data

Run message testing and willingness‑to‑pay research in the same study. When a positioning variant moves the acceptable price range, you’ve learned something neither team would have found alone.

Before you change the price, check the story. Often the fastest pricing win is a positioning change.