Your pricing page publishes which buyers you won’t separate.

Someone will ask whether to put the prices on the website, and the room will split into two camps that both think the question is about disclosure. It is not. A pricing page almost never shows every price or hides every price: it shows a number to some buyers and a form to the others, and the line between them is the actual decision. What gets published is not a price. It is a partition: the list of buyers you have decided not to separate.

That has consequences somebody has measured, and they do not run the way either camp expects.

Why do B2B pricing pages reject the simple public versus gated binary?

I coded the pricing pages of the 64 companies in the BVP Nasdaq Emerging Cloud Index, a published and checkable list, in English and in German, on 9 August 2026. The frame was fixed before any page was opened, every constituent gets a row, and the whole thing: vendor list, code book, scripts, and every rendered page: is published alongside this piece.

Of the 55 vendors whose English site could be measured, 38 have a pricing page reachable from their own navigation. Of the 33 of those that show a price at all, between 58% and 79% also carry a contact-sales route on the same page. The range is not hedging: the low figure counts only a contact prompt sitting within about 1,500 characters of a price, which is roughly “inside the tier block”; the high figure counts one anywhere on the page. The truth is between, and I will come back to why I cannot narrow it further. Seven of 38 pages, 18%, show a price and no contact-sales CTA anywhere on the page.

So the modal pricing page is not transparent and not opaque. It is sorted. HubSpot’s German marketing page carries ten distinct euro prices, a free tier at zero, and an invitation to contact sales sitting among them. That is not a company that failed to decide. That is a company that decided twice, for two different kinds of buyer, and published the result.

This matters for reading everyone else’s numbers, including the ones you have been shown. A recent and careful study of 304 vendors reports that 23.8% “run a contact-sales wall instead” of showing prices. The word doing the work there is instead. On this corpus, for most vendors, it is as well as.

What operational commitments are enterprise sellers actually deciding when publishing prices?

Set the disclosure framing aside and the decision has a name.

This is the fourth decision in a sequence this journal has been working through, and it is the only one made in public. Where the number sits is set by the comparison your buyer already has in mind; the unit it is charged per is a decision about who carries usage risk; what you give away against it outlives the deal that won it. Those three happen inside the company. This one happens on a page anyone can load.

A published price is not a commitment: vendors discount published tiers all day, and if you publish a list and take 40% off it every time, you have given up nothing. That gap between list and pocket is real and it is the subject of the discount piece above. What a published price does is narrower and harder to escape: it constrains how far you can separate buyers before the separation becomes visible. For the tiers carrying a number, you have posted a reference point that every buyer can see, and moved yourself toward charging them all the same thing. For the tiers carrying a form, you have kept the freedom to charge each buyer separately, which is precisely why most pages keep some.

Economics has a name for that too: a move toward uniform pricing, and away from price discrimination, and a large literature predicting that it cuts both ways depending on the shape of demand. What it mostly lacks is measurement in markets like this one, where prices are negotiated between businesses rather than posted to consumers.

There is one good measurement, and it is worth knowing exactly what it found.

What did empirical research reveal regarding public pricing and seller surplus?

Matthew Grennan’s study of coronary stent purchasing (Grennan, 2013) is a structural model of bilateral bargaining fitted to what 96 US hospitals actually paid four manufacturers, month by month, from 2004 to 2007: 10,098 observations of negotiated, transacted prices in a business-to-business market.

He then asked what happens if those prices are made uniform. Isolating the competitive effect: holding the buyers’ bargaining strength at its observed average, so the only thing changing is the end of discrimination: prices rose about 1.7%, manufacturer profits rose 8%, and hospital surplus fell 1.4%. Buyers were made worse off by everyone being charged the same.

The mechanism is the part to carry away. When buyers are asymmetric: some prefer one supplier, others prefer another: a move to uniform pricing softens competition, because each seller retreats to the buyers who were always going to choose it. When buyers are symmetric, all wanting the same supplier more, uniform pricing sharpens competition instead. The direction of the effect is set by whether your buyers want different things.

Two things this does not license, and I would rather say them than have them found.

The number does not travel. Stents are not software, 2007 is not 2026, and Grennan’s counterfactual imposes uniformity on the transacted price, not on a web page. What transfers is the structure of the decision, not 1.7%.

And Grennan himself declines to make the argument about transparency that his result invites. A real analysis of it, he writes, “would require a model of how these variables influence bargaining ability as well as data to identify how much. Such an analysis is beyond the current theoretical frontier and also beyond the data available here.” His paper does contain a dramatic figure: buyer surplus falling by nearly half, but that is a scenario in which hospitals cannot negotiate at all, at prices he says are “well outside the observed range of data.” It is quoted around; it should not be. The modest number is the one that was measured.

Which form of pricing transparency actually benefits enterprise software buyers?

Here is the part the debate collapses. Seven years later Grennan and Ashley Swanson measured the other kind of transparency, not sellers posting prices, but buyers finding out what their peers paid: in the Journal of Political Economy.

Hospitals gaining access to peer-price benchmarking data saved 3.3% on physician-preference items, 3.9% where they were buying in high volume, and 1.6% on commodities. The savings went almost entirely to buyers who had been paying too much, and the authors are careful that most of the sample saw little. They also model, and weakly detect, a countervailing effect: because a negotiated price becomes visible to future buyers, sellers gain a credible reason to refuse their lowest prices, so some prices rise. They call that result fragile and say one of its predictions is not borne out in any specification.

Two interventions, same lead author, same market, opposite signs. Sellers publishing prices moved surplus toward sellers. Buyers learning peer prices moved surplus toward buyers. They are both called transparency and they are not the same act, and the second one is not something your pricing page can do. Your page cannot tell a buyer what your other customers paid. A procurement platform can, and sells exactly that, to them, about you.

Dimension	Grennan (2013)	Grennan & Swanson (2020)
The intervention	Price discrimination ends: every hospital pays the same	Buyers gain peer-price benchmarking data
Who gets the information	Nobody: prices are made uniform	The buyers
Prices paid	Rose 1.7%	Fell 3.3% on physician-preference items; 3.9% in high volume; 1.6% on commodities
Where the surplus went	To sellers: manufacturer profits +8%, hospital surplus −1.4%	To buyers: savings concentrated on those who had been paying high prices
What a pricing page can do	This one	Not this one: your page cannot show what others paid

How do SaaS vendors adjust pricing when multiple customer segments observe the same page?

The corpus has a second arm, because a vendor serving two countries has to make this decision twice in public.

Of the 52 vendors whose German-language experience could be measured, 34 have a reachable pricing page, but 19 of those 34 simply serve the English page to a German-language visitor. Only 15 of the 64 companies present a genuinely German-language pricing page at all.

And of those 15, seven show prices in euros. Eight show US dollars only. More than half of the vendors that went to the trouble of translating the page did not localise the number on it.

ChartMogul has a name for the distinction : cosmetic localisation, where the currency changes and the price points do not, against market-based, where the price itself moves. What has not existed until now is the base rate. On this frame, most of the vendors that localised at all did not clear even the cosmetic bar.

Read against everything above, that is not sloppiness. It is a published decision not to separate two markets, and unlike almost every other pricing decision a company makes, it is one an outsider can check.

What is the defensible state of empirical evidence on public SaaS pricing pages?

Twenty-one of my 128 observations could not be measured at all: a mix of bot walls and pages that would not render. They are recorded as missing data rather than counted as vendors hiding their prices, which matters more than it sounds: the first version of this corpus made exactly that mistake, scored 53 challenge pages as vendors with no pricing link, and produced a confident set of base rates that were wrong. I found it by auditing my own instrument, and the audit is published with the data.

That is also why the contact-sales figure is a range. The strict code and the loose code disagree, and I would rather publish two numbers that bracket the answer than one that looks decisive.

It is worth applying the same suspicion to what circulates. The claim that 72% of B2B buyers expect pricing on a vendor’s site, widely attributed to Gartner, does not resolve to a Gartner publication. A frequently cited figure putting bounce rates 38% higher without visible pricing is attributed to ChartMogul, which published no such thing. And a good deal of current advice still cites OpenView’s benchmarks as though the firm were issuing them, which it has not done since it wound down in December 2023. Four separate corpora of pricing pages have appeared since late 2024, reporting four different base rates; none of them published the list of vendors it coded, which is why nobody can say which of them disagree and why.

What practical audit of public pricing disclosures should commercial teams run next?

The literature cannot tell you whether to publish your prices, and any piece that says it can is selling you the confident version of a conditional result. What it can tell you is what the question actually is, and that the answer turns on something you know and Grennan did not:

Do your buyers want different things? If your market sorts: different segments genuinely preferring different vendors: then posting one price for everyone is likely to work in your sellers’ favour and against your buyers’, and the pressure you feel to publish is not coming from your buyers’ interests. If your buyers mostly want the same one or two vendors and are choosing on price, publishing sharpens the competition you are already in, and hiding is a delay rather than a defence.

Then open your own pricing page and read two things off it, which takes about five minutes because you already have everything you need:

Which tiers carry a number, and which carry a form. That line is your segmentation, whether or not anyone chose it deliberately. Most pages have both. If yours has a number on every tier, you have committed to not separating anyone: decide whether you meant to.

What currency the other market’s page shows. If you translated the page and left the dollars, you have told two markets they are one market. That may be right. It is currently the majority choice among vendors that translate at all, which is not the same as it being considered.

Neither of those is a forecast, and neither needs a model. They are your own decisions, already made, currently published.

Where are the empirical boundaries of pricing page research?

Boundary. The corpus shows how pages partition buyers, not which disclosure will maximise conversion in every market. Recode the current page set and connect the partition to win, loss and segment evidence.

Evidence base. The analytical frame also draws on these additional sources: legal source. The links identify the exact works; they support the mechanisms and boundary conditions discussed here, not every claim in isolation.

References

1. Grennan, M. (2013). Price discrimination and bargaining: Empirical evidence from medical devices. American Economic Review, 103(1), 145–177. https://doi.org/10.1257/aer.103.1.145
2. Grennan, M., & Swanson, A. (2020). Transparency and negotiated prices: The value of information in hospital-supplier bargaining. Journal of Political Economy, 128(4), 1234–1268. https://doi.org/10.1086/705329
3. Preisangabenverordnung (PAngV) vom 12. November 2021 (BGBl. I S. 4921), §§ 1, 2, 3, 11. https://www.gesetze-im-internet.de/pangv_2022/