Why Won't Sales Tools Show a Price? 126 of 264 Don't

Here's the trap. You have six tools on a shortlist and a budget meeting on Thursday. Two of them print a number, so you have real figures by lunchtime. The other four say Contact sales, so you fill in four forms and wait.

Three weeks later you have sat through four discovery calls, explained your headcount four times, and received one proposal. You cannot tell whether the number in it is good. The only two prices you can compare it against belong to the two tools you liked least, and you spent your bargaining position before anyone quoted you anything.

The mechanism is simple and it is not a conspiracy. A price on a page is the same price for everybody. A price given on a call is not. The vendor who can price you individually will always earn more than the vendor who cannot, and everything downstream of that follows, including the pricing pages that print a number nobody can buy at.

It works like a builder quoting for an extension. Nobody prints a price for an extension, because the price depends on your house. Fair enough. But the builder is also using the visit to work out what you can afford, and you do not get to see anyone else's quote before you hear yours.

The number on the sign is not the problem. The absence of a sign is. This page counts every tool in this directory against one rule, splits them by category, gives three sourced reasons vendors gate, shows what the law requires (almost nothing), hands you a twenty-minute check you can run yourself, and says what a gated purchase costs you in month three.

So how many of the 264 actually show you a price?

138. One hundred and thirty-eight of the 264 tools in this directory publish a price a buyer can act on without speaking to anybody. The other 126 do not.

That is 47.7% of a market you cannot put in a spreadsheet before the meeting, computed from this site's own records as they stood on 25 August 2026.

The old headline said 108, and 108 was measuring something else

This page previously led with 108 of 264. That is the count of tools with no self-serve checkout, which is a fact about the buying process rather than about the price. 114 publish no price of any kind.

Only 96 are in both sets.

Twelve print a number and still route you to a form. Eighteen will take a card without ever showing you a rate.

Leading with 108 implied the two questions were one question, and they are not.

Nothing here is a measurement. Every figure is computed from vendor pages as observed and recorded in this site's own entity records, 152 of which carry a first-party price verification date in August 2026. No invoice has been read. No contract has been seen. Where a question can only be answered by buying the product, this page says so and tells you what to check.

What counts as a price, and what "gated" means

Now the definition is worth having, because you have just felt why it matters. A price is only a price if the vendor shows it to a buyer, in a currency, for a stated period. That is the test all 264 records were checked against, and it is stricter than it sounds.

A number in a structured-data blob is not a price. A number behind a currency selector defaulting to somebody else's currency is not your price. A number that exists only in a competitor's comparison table is not a price at all.

Gated means something different: you cannot complete a purchase on your own. The button says buy, the button opens a form, and a person decides what you pay. A tool can be gated and priced, priced and ungated, gated and silent, or ungated and silent. Keeping the two axes apart is the whole trick, and it is why the recount moved.

This site has been caught by the softer version of the rule. Koncert's pricing page served five tier prices inside a container carrying display:none, with the currency span absent and the billing-period span present and empty, read 2026-08-07. The integers were in the HTML. A buyer looking at the page saw nothing, so Koncert publishes no price and that is how it is recorded.

The last row overlaps the two above it and is counted separately because it is the one that decides your week.

Eighty-five of 264 tools offer no free tier, no published rate and no checkout. There is no way to learn anything about their cost that does not begin with handing over a contact detail.

Where the gates actually are, category by category

This is the part no ranking page has, and it is the reason the census was worth running. Gating is not spread evenly. It clusters hard, and the pattern tells you what the vendor thinks it is selling.

The spread runs from 11% to 90%, an eight-fold difference inside one market. Enablement and writing tools sit at 90% because they are bought by one person with a card on a Tuesday afternoon. Nothing to scope, nothing to quote, so the price goes on the page.

The hardest products to integrate are the ones that publish

Voice infrastructure at 80% is the row that breaks the industry's own explanation. Deepgram, Cartesia, LiveKit and Telnyx are harder to wire up than most of the gated tools above them. They publish because their buyer is a developer who leaves the page if there is no rate on it, and because the unit they sell is a second of audio rather than a business outcome.

Complexity does not decide this. The buyer does.

Why is inbound qualification the darkest category on the list?

Two tools out of eighteen. Inbound qualification publishes less than any other category here, and it is also where the billing unit itself is newest. Five of the eighteen publish a meter with no rate, and four of those five meters are per resolution: Intercom Fin, Sierra, Crescendo and Salesforce Agentforce all charge for a resolved conversation.

A per-resolution meter is a definitional argument dressed as a price. What counts as resolved is set by the vendor, and this site has documented a vendor billing a resolution when the customer simply stops replying. A rate you cannot see, attached to a unit the seller defines, is not a price with a missing number.

It is two missing numbers.

The eleven inbound tools publishing nothing at all are mostly the enterprise deflection platforms. Where the vendor is selling a reduction in your support headcount, the number it wants to discuss is your current cost per ticket, and any published rate would anchor the conversation somewhere below that.

Reason one: they are pricing you, not the product

This is the reason buyers suspect and vendors rarely put in writing. It is worth establishing with evidence rather than asserting, because the honest version is more specific than the cynical one.

What buyers say on the record

On Hacker News item 33425443 (2022-11-01, 195 points, 168 comments), a buyer asked the exact question this page answers: where do you find ballpark pricing for a dozen enterprise SaaS products without spending a month in sales calls. The replies are unanimous that no such source exists.

One commenter wrote: "I always assumed 'contact us' pricing meant 'tell us who you work for so we can google their funding level before suggesting a price to you'." That is one person's characterisation, not a finding of fact. It is the hypothesis, stated cleanly.

Another in the same thread put it in the vendor's voice: "'contact us' pricing revolves around understanding the customer's use case deeply enough so that you can demonstrate maximum product value to them and then extract the most possible money." On item 26144706 (2021-02-15, 77 points, 79 comments), titled Is "contact us for pricing" a dark pattern?, a commenter compressed it to eleven words: "It's to give worse prices to customers that are not good at negotiating."

What a filing says that a pricing page will not

None of that is data. The data is in a filing. ZoomInfo's FY2025 Form 10-K, filed 2026-02-12 under accession 0001794515-26-000012 (CIK 0001794515), reports revenue of $1,249.5 million, states its software is used by over 35,000 companies, and separately discloses 1,921 customers with $100,000 or greater in annual contract value at 31 December 2025, up from 1,867 a year earlier.

That is 5.5% of the customer base.

The same filing states how the number is set: "We price our subscriptions based on the functionality, users, and records under management that are included in each product edition." Three variables, none published as a rate card.

A single vendor running contracts from a free community edition up past $100,000 a year on one product line is not evidence of dishonesty. It is evidence that a list price would be wrong for almost every customer, which is precisely why there is not one.

The theory has a name and a canonical write-up two decades older than any SaaS pricing blog. Joel Spolsky's Camels and Rubber Duckies, published 2004-12-15, works through why software has near-zero marginal cost, therefore no natural price, and why segmentation rather than cost sets what you pay. A commenter on item 26144706 cited it directly.

Reason two: the price is a weapon and the page is where it gets taken

The second reason has nothing to do with you. A published price is readable by every competitor, permanently, at zero cost, and competitors read pricing pages far more carefully than buyers do.

On item 20191005 (2019-06-15, 70 points, 69 comments), a commenter worked the game theory from the seller's side: a reseller who published would win business briefly, and then "pretty soon the competition would catch on and embrace the same sales approach. Now everyone have a smaller profit margin on the product or, even worse, might get dragged in a price war." A commenter on item 33425443 said the same thing in nine words: publishing "would be even more of a race to the bottom."

Those are practitioner opinions rather than measurements, and they agree. The corroboration is again in a filing, and it is unusually direct. ZoomInfo's FY2025 10-K risk factors warn that competitors "may be able to offer more flexible pricing models that better align with the actual usage and realized value obtained by customers, including pay-as-you-go models or other bespoke pricing packages." A company that names pricing-model flexibility as a competitive threat is telling you the pricing page is a competitive surface.

Sometimes the silence is not the seller's choice

On item 20191005 a commenter noted that reseller silence is often "an end run around contract clauses that tell them they can't advertise pricing until there's a qualified lead." One person's account of one industry, uncorroborated here. It matters because if a partner cannot quote, the manufacturer's terms may be the reason, and pushing that partner is wasted effort.

The consequence for you is narrow and useful. A vendor withholding for competitive reasons usually will quote fast once you are in the funnel, because the cost it is avoiding is publication rather than disclosure. That vendor is worth an email. The one avoiding disclosure will still be qualifying you on call three.

Reason three: some of them genuinely cannot print a number

The third reason is the honest one and it is real for a minority of the 126. Where the product is an implementation, the price depends on the implementation, and no page can carry it.

The best statement of it is on item 33425443, from a commenter answering a comparison with SpaceX's published launch prices: "Enterprise SaaS is the exact opposite. The product that is listed on the website is just a starting point. There are teams of sales reps, account managers, solutions architects and more who will customize it exactly to the customer's liking, and all that needs to be priced." Take that as a fair statement of the seller's case, not as a fact about any particular vendor.

The census supports it for one category and undercuts it everywhere else.

Vertical voice agents sit at 25% buyable, and those products do get wired into a dental practice management system before they work. But data and research sits at 71% on six-figure contracts.

Where the argument does hold, it holds visibly. Five9's FY2025 10-K (accession 0001288847-26-000023, filed 2026-02-20, CIK 0001288847) reports remaining performance obligations of $1,190.5 million on contracts originally longer than one year, and expects to recognise about three quarters of it over the next 24 months. That is a book of multi-year, individually negotiated agreements. Five9 still publishes $119 per seat per month (read 2026-08-12) and still routes you to a form.

The twelve tools that print a price and still make you call

These are the ones that cost you a fortnight, because the number reads like a price and behaves like an anchor. Each of the twelve publishes a figure and none will let you act on it.

Read the seat prices against each other and the trap is obvious. Salesforce at $25 and Regie.ai at $180 look like different budget conversations. They are not, because one of them carries a minimum. This site documented that mechanic separately in the seat-minimum census: a per-seat figure with a floor attached is a readable number and an unbuyable one, and the floor never appears next to the price.

The last two rows are a different failure. Neuphonic and Sieve publish a $0 entry point and no rate above it, so the record holds a price of zero and the buyer holds nothing. Any census counting "has a price field" scores these as priced. They are not.

Every published per-minute rate here is a partial price

The most common form of hidden pricing on this site is not a blank page. It is a number that is real, published, in dollars, per minute, and incomplete. Six voice platforms here carry a full component-level cost stack, computed from each vendor's own published rates for the pieces its headline excludes. All six exclude at least one component. No vendor in the set has an advertised per-minute rate that is the rate you pay.

Those all-in figures are modelled from stated assumptions, not measured. Each is the vendor's advertised rate plus the published list rate of every component the vendor says it excludes, against a component floor of $0.0465 a minute. No traffic has run through any of these accounts. The advertised rates were read between 2026-08-05 and 2026-08-11.

Millis is the one worth staring at. The headline is $0.02 and the computed floor is 3.33 times it. That is not our inference against the vendor's wishes. Millis states in its own docs that the rate is in addition to model, synthesis and transcription fees, and its own worked example totals $0.6615 for ten minutes. The vendor published the finding. Nobody reads that far down the page.

Vapi is the same shape with a blunter label. Its comparison table is headed Excludes Model Provider Costs, and its own calculator totals $0.1103 a minute on default inputs, 13% above the floor computed here. Two independent arithmetic routes landing within a hundredth of a cent of each other is the strongest confirmation available without an invoice.

44 tools tell you what they charge for and not how much

This is the state most censuses miss, because a scraper hunting for a dollar sign scores it identically to a blank page, and a buyer reading it feels almost informed.

Forty-four of the 264 publish a billing model with no rate attached: twenty per seat, eleven usage, seven flat, four per resolution and two per minute.

Outreach and Salesloft both tell you they charge per seat and neither tells you what a seat costs (both read 2026-08-11).

Gong publishes a per-seat model with no rate and no first-party verification date on record here, which is marked unverified rather than papered over.

Twenty-six of the 44 are gated. The remaining eighteen will take your card without ever having shown you a rate, which is the strangest state in the census. Zoom Contact Center and RingCentral RingCX both sit there, as do Pylon and Default in inbound qualification, and Copilot Studio, whose consumption rate lives in a separate licensing document rather than on the product page. A meter without a rate is not worthless. It tells you the shape of the bill, which decides whether cost scales with headcount, call volume or ticket volume. It does not tell you whether you can afford it, and the two questions are routinely conflated.

Do bigger companies publish more? No, they publish less

The intuition is that a mature public company has settled packaging and can therefore print a price, while a startup is still guessing. The census says the opposite, with the caveat that the sample is small and the classification is ours.

Eighteen entities here are publicly traded or are the product line of a public parent: Salesforce, HubSpot, Five9, NICE CXone, Twilio, RingCentral, 8x8, Zoom Contact Center, Webex Contact Center, LivePerson, Verint, ZoomInfo, Bandwidth, Sinch, Vonage, ElevenLabs, Agentforce and Copilot Studio. Ten of the eighteen are gated, against 108 of 264 directory-wide. That is 56% versus 41%. Eight of the eighteen publish a usable price, against 52% overall.

Eighteen is a small sample and the public-parent classification is a judgement made here rather than a field in the data. Treat the direction as a signal and the magnitude as noise. The acquisition cut points the same way: eleven entities carry an acquisition record and seven of them are gated, 64% against a 41% baseline. Acquired products get folded into somebody else's enterprise price list.

What a published price looks like inside a filing

Three public vendors in this directory sit at three different points on the gate, and their own disclosures show what the difference does to the business. All six figures below are from FY2025 Form 10-Ks filed in February 2026.

HubSpot's average subscription revenue per customer has stayed inside a 0.6% band for three years, $11,384 in 2023, $11,343 in 2024, $11,414 in 2025, while its customer count grew 41% from 205,091 to 288,706. That is what a published price produces in a filing: one stable per-customer number the company is willing to disclose annually.

ZoomInfo discloses no equivalent figure, and the reason is visible in the structure. A base running from a free community edition to 1,921 contracts above $100,000 has no meaningful average to publish. Neither company is behaving badly. They have chosen different pricing architectures and the filings show which is which, which is more than either pricing page does.

The twenty-minute check: is this vendor's price a real price?

Run this on one vendor before you fill in any form. It takes fifteen to twenty-five minutes, needs no account, and produces a yes or a no rather than an impression.

The pass mark: you can compute a unit cost for your own volume from published numbers alone, and the checkout agrees with the page. If you cannot, write meter not published on your shortlist and move that vendor into the gated column, whatever number is on its homepage.

Step three is the one that catches careful people

Three of this site's own corrections came from reading served HTML. Koncert's five suppressed integers. Cal.com printing the same figure in two sections with only one of them stating the annual commitment, confirmed in a rendered browser rather than inferred. Lusha's credit slider, where the seat charge is identical on the struck-through and the discounted price, so the advertised sale excludes seats. All three read as ordinary pricing pages in a browser. All three were read on this site between 2026-08-06 and 2026-08-12.

What this check cannot tell you

The procedure above verifies that a published price is real. It cannot tell you what you would actually pay, and on a gated vendor those are different questions. It cannot detect the discount you would be offered, which is set by your headcount, the vendor's quarter, your competitor set and how obviously ready to sign you look. It cannot detect a seat minimum kept in a contract rather than a FAQ. It cannot detect a renewal uplift clause, because published terms are the opening position and the redline is where the term lives.

It also cannot detect the case a commenter on item 33425443 named: "Some companies make you sign an NDA before they tell you their prices." One person, one sentence, no corroboration here. Worth knowing the state exists, because an NDA on a price ends your ability to benchmark that quote against anything.

One route reaches an actual signed number: the federal award file

Five9 publishes $119 a seat and will not sell you one. The federal government publishes what it paid.

A USAspending query for Five9 contract awards, run 2026-08-25, returns award 75P00122P00119, Department of Health and Human Services, $47,144.00 for the period 2022-09-26 to 2023-09-25, described as monthly subscription services with one-time fees and a telecom deposit. Award 12326720P0034, Department of Agriculture, records $13,635.00 against a line reading agent seat, 100% recording MRC, concurrent user.

That is a real signed total for a gated vendor, published by the buyer rather than the seller. It is not your price. Federal pricing is negotiated separately from commercial list and often sits below it, and a 2022 award tells you nothing about 2026 packaging. It is still the only route in this market to a dated, non-negotiable, third-party-published figure for a vendor that publishes nothing, and it costs one API call. Pair it with a procurement broker's median, whose bias points the other way.

What the law actually requires, which is close to nothing

Buyers reasonably assume some disclosure obligation exists here. In the United States there is essentially none for business software, and the position got weaker rather than stronger over the last two years.

The FTC's Rule on Unfair or Deceptive Fees (90 FR 2066, published 2025-01-10, effective 2025-05-12) is the only binding federal total-price disclosure rule on the books. It makes advertising a price without clearly and conspicuously disclosing the total price an unfair and deceptive practice. It applies to live-event tickets and short-term lodging. That is the entire scope. A concert ticket carries a stronger price-disclosure obligation than a $200,000 contact centre contract.

The one adjacent rule that might have helped is off the books

The FTC's amended Negative Option Rule (89 FR 90476, published 2024-11-15, effective 2025-01-14), the one the press called click-to-cancel, would have imposed disclosure and cancellation duties across recurring subscriptions in any medium.

The Eighth Circuit vacated it in Custom Communications, Inc. v. FTC, 142 F.4th 1060 (8th Cir. 2025), finding the Commission's failure to issue a preliminary regulatory analysis "procedurally insufficient." On 2026-02-12 the FTC formally recodified the pre-2024 text (91 FR 6507, effective the same day), restoring the prenotification rule as it stood before the amendment took effect. So: no obligation to publish a price for B2B software, no obligation to disclose a total, and the nearest recent rule removed.

For scale on the surrounding behaviour, a July 2024 international sweep by the FTC with ICPEN and the Global Privacy Enforcement Network reviewed 642 apps and websites and reported 76.0% using at least one dark pattern and 67% omitting the cancellation notice timeline. Those figures reach this page through secondary coverage and are not verified first-hand.

Month three: what a gated purchase costs you at renewal

The licence fee is not where a gated purchase hurts. The damage arrives eleven months later, in three parts. You have no reference price. You never obtained an independent number, so when the renewal quote lands you cannot tell an 8% uplift from a 30% one without restarting the whole evaluation. The vendor knows how long that takes, because it took you three weeks to get one quote the first time.

The contract usually ratchets one way

Five9's FY2025 10-K describes its own terms: "We offer monthly, annual and multiple-year contracts to our customers, generally with 30 days' notice required for limited reductions in the number of licenses or the level of consumption or capacity. Increases in the number of licenses or the level of consumption or capacity can be provisioned almost immediately."

Adding is instant. Reducing needs notice and is limited.

That is a public company describing a standard commercial term in a filing, which makes it reportable in a way a sales conversation never is.

Retention is the metric the vendor is managed on, and you are one input to it. ZoomInfo's FY2025 10-K reports net revenue retention of 90% at 31 December 2025, up from 87%. Below 100% means the installed base shrank net of expansion. Five9's Annual Dollar-Based Retention Rate was 105% for 2025, down from 108% in 2024. Neither is a judgement about product quality. Both are the pressure your account manager is under in the quarter your renewal lands, and a vendor at 90% has a renewal team on a harder target than one at 105%.

A commenter on item 33425443 named the version that hits smaller buyers: "Whether or not your company has a procurement department should always be part of the equation when pricing enterprise deals. (Funding is an indicator of whether your company has one.)" One person's account of how they price, not a finding of fact. It is the clearest short explanation of why your funding announcement and your renewal quote are related events.

What to do on Monday when half your shortlist is gated

The mistake is treating the gated tools as the main event and the self-serve ones as the cheap seats. Reverse the order and the evaluation changes shape.

• Price the 138 first, before you fill in a single form. Apollo at $69 a seat (read 2026-08-26) and Instantly at $47 a month (read 2026-08-06) both publish and both take a card. Those numbers are the only independent anchor you will get, and you cannot obtain one once negotiations have started.
• Keep one published comparator on the list even if you will never buy it. With every tool in your evaluation gated you have no price discovery at all. In categories at 11% and 22% buyable you may have to import the comparator from an adjacent category to get one.
• Run the seven-step check on every priced vendor before you trust its number. Twelve of the priced tools here cannot be bought at their published rate and two of them publish $0. Twenty minutes each stops you anchoring a business case to a figure that was never available.
• Ask for the number in the first email and treat the reply as data. Most will not answer. A vendor sending a range unprompted is usually one whose product does not need a scoping call, which tells you about the product and about the sale.
• Ask what the implementation involves and who performs it. If the honest answer is that their team configures it over six weeks, you have found the real reason there is no price, and the licence fee was never the number that mattered.

One group should stop here. If you are buying a single seat of a writing or enablement tool, none of this applies: those categories run at 90% published, you can price the whole shortlist in ten minutes, and the gated-vendor playbook is overhead. This page is for buyers whose shortlist crosses into voice, contact centre or inbound deflection, where a published rate is the exception.

The email that sometimes gets you a number

Telling you to ask is useless without the words, because the ask fails predictably. "How much is it?" returns a calendar link every time. What works better is making the answer cheap to give and expensive to dodge.

Three things are doing work. You named a number first, so answering costs them a yes or a no rather than a disclosure. You said there are three vendors, so silence has a price. And you gave them permission to disqualify you, which is what makes a salesperson answer honestly, because their worst outcome is a wasted month rather than a lost deal.

There is a published example of the other side of the trade. On item 37717689 (2023-09-30, 46 points, 61 comments), asking for SaaS companies that print high prices openly, the founder of Keygen replied that his pricing runs to $6,000 a month in public. He switched because he was "tired of answering the same questions over and over again for enterprise leads," and now the sales cycle "can usually be done in an email instead of a series of pointless calls to size each other up."

That is a vendor describing his own business, not an independent measurement. The same thread carries the buyer's side: "I'm not scared to pay serious money for a service, but putting your service behind a sales person is more than likely going to cost you my business." Two people on one thread is a pattern worth naming, not a survey. It is worth naming because it contradicts the standard defence, which is that buyers at that price point expect the call.

Where the census is weakest

Three limits, stated before you rely on any number above.

The denominator moves. Six of the 264 entities are recorded as unreachable at the host on file, including Air.ai, whose domain now serves an unrelated company, and Edify, which disappeared mid-sweep. A product that has stopped existing scores as publishing no price, which is technically true and analytically misleading. Deduct those six before using 264 as a market size.

112 records carry no first-party price verification date. The 152 that do were read in August 2026. The rest date from the original research pass and are marked unverified rather than given the site-wide date. Earlier sweeps here found six entities whose published prices had moved by an order of magnitude while a site-wide date implied all of them were current.

The four states are computed, not judged. A tool lands in a bucket because of its price fields and its selfServe boolean, not because anyone weighed how helpful its pricing page felt. A vendor publishing a genuinely useful calculator and a vendor publishing one meaningless number both score as priced. The census measures presence, not usefulness, and the per-minute table above is the clearest case of the two coming apart.

So why won't most sales tools just show you a price?

Because for 126 of the 264, there is no single price to show. There is a range, and where you land in it depends on what the vendor learns about you during the three weeks you spend telling them.

The gate is not hiding an embarrassing number. It is protecting the ability to charge you a different one from the company next door.

Three reasons run underneath it and they are not equally defensible. Segmentation is the strongest and the one vendors will not name; a filing reporting 1,921 customers above $100,000 ACV inside a base of 35,000 describes it without using the word. Competitive exposure is real and mostly about rival vendors. Implementation complexity is honest where it applies, and it applies to fewer categories than the industry claims.

What follows for you is a sequence rather than a shortlist. Price the 138 first and use them as the anchor. Run the seven-step check on every number you plan to quote in a business case, because twelve of the priced tools here cannot be bought at their published rate. Then walk into the gated conversations already knowing what the alternative costs, which is the one advantage you hold and the one you can lose by accident.

And when a vendor does publish a usage rate, read its definition before its number.

This site has computed that every advertised per-minute rate it holds excludes something, and that a seat licence and a per-minute rate cross over at a specific amount of talk time. A published price you have not decomposed is closer to a gated one than it looks.