Author
TL;DR When AI agents run the evaluation, contact sales pricing gives them nothing to compute. The agent carries the edition as an unknown or drops it from the ranked set. A human evaluator resolved that unknown by filling in a form; an agent resolves it by ranking what it already has. The reflex fix, publishing a number, exports the pricebook to every party that can read it, including customers holding parity commitments. A legible shape fixes it: a named value metric, a stated boundary, a defined license grant.
Why contact sales pricing costs you the shortlist when AI agents evaluate you
An agent-run evaluation ranks a field of vendors on what it can compute. Contact sales pricing, read by AI agents, is a price field with nothing in it. The agent either carries the edition as an unknown and ranks the vendor on the editions it can price, or it drops the edition from the ranked set. The loss happens before anyone at the vendor knows an evaluation was running, which is the shift the agentic buyer describes at the market level. This piece takes one edition and reads it.
What an agent does with an edition it cannot price
The agent takes a configuration, projects a year of usage, and needs an annual cost to rank. The contact-sales edition returns no cost. So the agent computes the vendor on the editions below it, which usually means comparing an enterprise buyer against a mid-market edition, or it treats the whole vendor as unpriceable and moves on. Neither outcome involves the vendor.
How did the contact-sales edition behave when the evaluator was human?
A human evaluator read “contact sales” as an invitation. They filled in the form, took the call, and heard the price with the argument attached. The edition worked because the form started a conversation in which the vendor could explain what the number meant. An agent has no form. Its shortlist is drawn before the conversation could begin, and the sales motion behind the edition now sits downstream of a ranking the vendor never saw.
What a buying agent reads, and where the pricebook goes silent
The read boundary applies to the contact-sales edition as it applies to any edition. The one difference: this edition usually publishes less than the others, so the silence starts sooner.
What the contact-sales edition can publish without publishing a price
Four facts, at no price. The unit the edition is denominated in, the entitlements it grants, the boundary condition that moves a buyer into it, and the limits that apply inside it. Each of those is a fact about the edition’s shape. An agent can rank on shape when the shape is stated, because it can tell whether the buyer’s configuration belongs in this edition at all. The architecture-before-format argument behind that is machine-readable pricing, and this piece does not rebuild it.
What only the pricing decision layer can answer
The decision layer is where a software company’s licensing, packaging, and pricing choices are made, distinct from the runtime layers that execute them. It answers the three questions a page cannot: which Customer Group this buyer belongs to, what realized price this configuration lands at, and why. What AI buying agents look for in software pricing works through all three. Those are judgments, and an edition that leaves everything to judgment has no shape for an agent to read.
Which Customer Groups does the edition serve?
Customer Groups are clusters of customers who derive value from a product in similar ways, regardless of size or vertical. A contact-sales edition is usually a proxy for one or two of them. When the edition’s boundary is stated in terms of what those buyers do, an agent can see whether its principal belongs there. When the boundary is stated only as “enterprise,” the agent sees a label.
Talk to an Expert About the Pricing Problem in Front of You
Describe what you’re facing and a pricing expert will reply with a concrete read on your licensing, packaging, and pricing architecture.
Why publishing a number is not the fix
The reflex answer to an unreadable edition is to publish a price. That move solves the agent’s problem and creates two of the vendor’s.
The price an AI agent can read is a price a customer can claim against
A number an agent can read is a number every existing customer can read. Customers holding parity commitments in their agreements have contracted for a defined relationship between their price and what others pay. Publishing the edition’s price turns a readable number into a claimable one for that population.
The constraint is a structural fact about what a published price is. How far a specific commitment reaches into a specific base is diagnosis work against the paper and the pricebook together, and a pricing page settles none of it.
What happens to realized price when the pricebook becomes public
The realized price is what a deal lands at once commitment, term, and every reduction that reaches a buyer are applied. Where discounting has been a sequence of decisions rather than a discipline, list and realized price sit far apart. Publishing list into that gap publishes a number the market already knows is fictional. Price comparison at scale squeezes a vendor whose unit reads as everyone else’s and rewards one whose unit is distinct. Publishing a price without a distinct unit behind it puts a vendor on the wrong side of that comparison.
Does full transparency win the agent evaluation?
Only when the architecture behind the number holds. An edition with a published price and no named unit is a number an agent cannot compare to anything. An edition with a named unit, a stated boundary, and a defined grant is legible at zero price disclosure. Transparency and legibility are separate properties. The evaluation rewards the second.
The pattern from the corpus is the pandemic price cut. In 2020 a number of software companies lowered their enterprise prices to hold deals that were stalling, and the reductions were public enough to be read as the new number. The existing base then read them too. Customers holding most favored nation clauses had contracted for exactly that relationship. The cut that was meant to win a quarter of new business became a repricing event across accounts that had never asked for one.
The published number did not create the exposure; the agreements already held it. Publication is what let every party who could claim against it see it at once.
Giving the contact-sales edition a legible shape
Shape is what an agent can rank on. Every element below can be published without a price, and each is a decision the vendor has already made somewhere, whether or not it has been written down.
Name the unit before you name the number
The value metric is the unit the licensing model selects, and it comes before any price in the sequence. The pricing model is a rulebook applied to that unit. An edition that names its unit tells an agent what a buyer’s usage would be measured in, even with no rate attached. An edition that names a rate with no unit tells it nothing it can multiply.
Publish the boundary, not the bill
The boundary condition that moves a buyer into the edition is the most useful fact an agent can have about it. Stated in the buyer’s own terms, it lets the agent answer the one question it can answer for this edition: does my principal belong here. Peer-reviewed research on option presentation finds that how a set is displayed changes which option is chosen, and a blank cell in a comparison is a display choice with a result. Where an edition lands its boundary is design work; that a boundary exists and is stated is the frame.
Where agent-driven usage sits in the license grant
A license grant is the set of rights a license conveys. It names who may use the software, in what unit, for what purposes, and under which policies. A buyer’s agent will ask, in effect, whether usage driven by an agent counts inside the grant or outside it, because the answer changes the projected cost. A grant that anchors to humans leaves agent-driven usage outside it, and the edition should say so.
A longer sequence has to be true inside the company before any of this can be exposed. The grant question is the part an edition can answer today.
If your contact-sales edition has a unit, a boundary, and a grant that nobody has written in one place, that is a legibility problem with a bounded fix. Describe where the edition stands and a pricing expert will reply.
Tests to run against your own contact-sales edition
Put each of these to your own pricing page and pricebook, and answer from what an outside reader could see.
- Does the edition name the unit a buyer’s usage would be measured in, and is the unit defined the same way everywhere it appears?
- Can a reader tell, from the page alone, which condition moves a buyer out of the edition below and into this edition?
- Does the edition state what the license grant covers, and whether usage driven by an agent sits inside or outside it?
- If an agent priced your last five deals at this edition from the published surface, how far from the realized price would each land?
- Which existing customers hold a parity commitment that would reach a published price at this edition, and does anyone on your side know the number?
- If the edition’s price field were filled in tomorrow, which of the questions above would still have no answer?
The tests do not settle where a specific company’s edition should land. That resolution depends on the base, the paper, and the architecture behind the pricebook, which is a conversation rather than a checklist. If the tests turned up more open questions than closed ones, describe your pricing situation and a pricing expert will reply.
