What a buyer can actually see when they look up a price

Looking up what a sales tool costs is not one question. It is three. Does the vendor publish a number at all. Does the number cover the thing you came for. And can the page carrying it be read by anything other than a person with a browser open. We read the published pricing pages of 25 sales-technology companies to find out, and the third question turned out to be the interesting one.

The count

Twenty-five pricing pages in the sales engagement and dialing category, read on September 1 and September 3, 2026. Thirteen published a price. Eleven asked for a form, a call or a quote request instead. One published a figure for part of its range only. That is the headline, and it is deliberately a posture count rather than a price list: what a vendor charges changes the week they repackage, but whether they are willing to tell you at all moves far more slowly and is checkable by any reader in one click.

Two of the twenty-five pages did not return enough readable text for us to conclude anything on our own reading, and they are marked as could-not-verify rather than counted as a vendor publishing nothing. That is a fact about our reader, not about them. A nine-platform slice of the same sweep is published with a source link and a read date on every row in our sales engagement platform comparison. That published table, not this tally, is the part you can audit line by line today; the wider count comes from the same reading method described above.

Apollo publishes prices that a plain fetch cannot see

This is the most interesting row in the study and it is an observation about rendering, not an accusation. Read on September 8, 2026, the Compare plans section of apollo.io/pricing arrives in the server response as a loading spinner. The plan cards and their figures are fetched by the browser afterwards, so they appear in no server response at all: not in the HTML, and not in the React payload behind it. Our positive control on that fetch counted 453 occurrences of the word Apollo and 77 of the word pricing, so the page loaded fine. The figures were simply not in it.

What Apollo does serve to a plain fetch is structured data. Read on September 8, 2026, the JSON-LD on apollo.io/pricing carried four offers: a Free plan at $0, Basic at $49, Professional at $99, and an Enterprise plan whose price field holds the string "Contact for pricing". Those are Apollo's own strings on that date, quoted here because the point of this section is which surface carries them rather than what they cost. Check them at the source before you rely on them.

The consequence belongs to a rendering choice rather than to anyone's intent, and it is worth stating plainly. A crawler that does not execute JavaScript, a plain fetch, or an AI system reading the page in order to answer a pricing question sees the structured data and not the table. If you have ever wondered why an assistant quotes a plan set that does not match what your browser shows you, this is one of the mechanisms by which that happens.

The feature you came for often sits on a tier with no price

JustCall publishes real numbers, and more of them than several of its peers. Read on September 8, 2026, justcall.io/pricing showed Team at $29, Pro at $49 and Pro Plus at $89, each per user per month billed annually, and listed both Business and SalesPro as Custom, behind a Talk to Us or a Request a Quote button. Three of the five tiers carry a figure.

The observation is narrower than a complaint about quote-only enterprise tiers, which are close to universal in this category. On that same page, read on that same date, the feature line reading "Predictive dialer (up to 10 lines)" sits under SalesPro, and SalesPro is one of the two tiers with no published figure. A buyer who arrived wanting ten-line dialing can read three prices and none of them is the price of the thing they came for.

Nothing on that page contradicts anything else on it. This is a packaging fact rather than a disclosure failure, and it is the most common way a published price quietly stops answering the question a buyer actually has.

Two products, one pricing page

Klenty sells two things, and it says so. Both of its pricing pages carry an explicit Sales Engagement and Sales Dialer toggle at the top of the plan grid, and its navigation sells them as separate product lines. Read on September 8, 2026, the parallel and power dialer is an add-on to the sales engagement platform and is bundled into the Pro tier of the standalone sales dialer. Those two facts are consistent with each other. This is ordinary packaging, not a self-contradiction, and I am saying so explicitly because an earlier draft of this research had it filed the other way.

The buyer-relevant point survives anyway, and it is about comparability rather than honesty. Someone who lands on one of those two pages, compares it against eight other vendors and never notices the toggle will conclude that a capability is priced one way when the same company also prices it another. Multiply that by nine open tabs and a spreadsheet and you have the real cost of this category's pricing pages: not deception, but arithmetic that quietly stops being like for like.

Two pages that disagree with each other

Two vendors in the set publish strings that cannot both be describing the same product. I want to be careful about what that means. A website is written by more than one team, a navigation menu and a pricing table are revised on different schedules, and the overwhelmingly likely explanation is that one surface was updated and the other was not. I am reporting that the strings conflict. I am not reporting that anyone set out to mislead, and I have no evidence for that reading.

Kixie, read on September 8, 2026. The site navigation on kixie.com/pricing describes the Power Dialer as "Auto-dial up to 10 numbers in parallel". The pricing table on that same page caps its top dialer tier at "PowerDial up to 4 lines simultaneously", and the plan comparison further down reads "Up to 4 lines". Both strings are in rendered text, so a reader can find both without opening a developer console. It matters where each one lives: the ten-number string is global navigation chrome that appears across the site rather than pricing copy, and there is no tier above the four-line one.

Smartlead, read on September 8, 2026. The homepage lists parallel dialing among SmartDialer's capabilities. The product page, at smartlead.ai/power-dialer, answers its own FAQ by defining a parallel dialer as one that "dials multiple lines at once", defining a power dialer as one that "dials one number at a time", and stating that "SmartDialer is built for focused, one-to-one calls". Note the URL: smartlead.ai/smart-dialer now redirects to that page.

There is a reading that lets the homepage off, and it deserves an airing: the bullet sits under a Multi-channel outreach heading and pairs calls with emails, so parallel could be loose prose for doing two things at the same time. I do not find that persuasive, because parallel dialing is a term of art in this category and Smartlead's own product page is where I found it defined as one. But a reader should have the alternative and be allowed to weigh it.

Where we sit, including where we are behind

It would be easy to write this study so that it flattered us, so here is the part that does not. On parallel line count, meaning how many numbers a dialer rings at once, we are at the low end of this category. Our own comparison of Nooks and Orum records that Orum states up to 10 parallel lines in the copy on its pricing page, and the JustCall tier quoted above lists up to 10. Our dialer page says the cap is deliberate and shows the arithmetic behind it, because the share of connects arriving with nobody free to take them climbs faster than the extra conversations do. You are welcome to disagree with that trade. You should not have to learn about it from a competitor.

On disclosure we are in the thirteen rather than the eleven: we publish $499 per seat per month on our pricing page and in the Offer markup on it, so the two cannot disagree. That is the easy half. Publishing a number costs nothing when the number is simple, and ours is.

What we refused to print

Three facts were offered for this study and did not survive checking. Publishing the refusals is the part most vendor research skips, and it is the only way a reader can calibrate how hard we looked before concluding anything.

• A plan structure for one vendor that exists only in the data payload behind its pricing page. No reader can see it, so printing it as though a buyer could would have misdescribed the page rather than reported it.
• A widely circulated per-seat figure for that same vendor, which was not merely unverified but refuted at source: the price field in the vendor's own payload holds the literal string Request pricing.
• Seat figures said to live inside images on another vendor's pricing page. That page serves eight images and every one of them is an icon or a logo. The asset does not exist, so the figures could not be confirmed and are not printed.

A fourth thing was dropped on the day of publication. The working headline for this research was that five vendors contradict themselves. On re-verification, three of the five did not hold: one could not be verified at all because the strings are rendered by the browser, one turned out to be describing two separate products and was right, and one was true in substance but false as worded. Two survive, and they are above. Shipping the original five would have put three checkable errors inside an audit whose entire premise is that it can be checked.

An audit is only worth the claims it was willing to throw away.

How this was read

• Every page was fetched with a desktop browser user agent and no JavaScript execution. Where a claim depends on that distinction, this piece says so rather than leaving it to be discovered.
• Every fetch carries a positive control. The readable text was measured and a known word counted before anything was concluded, because a page that returns almost nothing proves nothing at all about what it publishes.
• The entire response was searched, not only the rendered text, so a price living in a data payload still counts as published. That is how the first section came out the way it did.
• Every publishable claim about a conflict rests on text a reader can see. No payload-only string supports one, which is a further reason the client-rendered plan table is reported as a rendering fact and not as a finding about price.
• No competitor figure appears here as our claim about what that company charges. Where a figure appears it is quoted as what the vendor's own page said on the date given, with the page named, and you should check it at the source.

Corrections

Last verified September 8, 2026. If you work at a company named above and a page has changed, or you think we have read it wrong, tell us and we will re-read the page and publish the correction with the date on it. A fixed page is the outcome we want, so if a claim here stops being true we will remove it rather than keep it with a they-used-to attached. The evidence pack behind every named claim, including the raw responses and the positive control on each one, is available on request.

• Playbooks
• Pricing
• Research