CRM Pricing: Why the Seat Price Is One of Four Meters

Every CRM pricing page answers a question nobody is actually asking. The seat rate is the one number vendors publish clearly and the one number that almost never causes a budget problem. Anybody comparing pricing across a full go-to-market stack needs a different figure, and the usual attempt at it is where the whole conversation goes wrong.

The received wisdom says the sticker price is a fraction of the true cost, and it usually arrives with a number attached: the real spend is two or three times list once you count integration, admin time, and idle licenses. That claim is repeated on dozens of vendor blogs and attributed to a major analyst firm. Follow the citation and it does not resolve. Every page repeating it cites another page repeating it, and the underlying report does not exist in any locatable form.

So this post does not use it, and it does not substitute a cleaner-looking invention. The problem with the multiple is not only that the source is missing. It is that a multiple assumes a base, and most go-to-market stacks have no single sticker price to multiply. Replace the multiple with a count and the argument gets checkable.

The short version

• There is no credible published figure for a stack's true cost as a multiple of list price. The widely quoted one traces to nothing.
• A normal GTM stack bills on four meters: seats, consumption units, database volume, and one-time gates.
• Only the first meter is on the pricing page in a form you can budget against.
• Three tools in a mainstream sales stack publish no rate at all. Gong, Outreach, and Salesloft list zero dollar figures on their own pricing pages.
• HubSpot's product catalog requires a one-time onboarding fee of $1,500 on Sales Hub Professional and $3,500 on Enterprise. Most comparison posts never mention it.
• Salesforce's own pages put an agentic seat between $300 and $550 per user per month: Enterprise at $175 plus the $125 Agentforce add-on, up to Agentforce 1 Sales at $550.
• Zylo's 2026 data names the actual source of surprise charges, and it is not the seat price. It is 78 percent tied to AI features and consumption pricing.
• Zylo also puts unused licenses at 46 percent. Nearly half of what a company pays for is nobody's login.
• Call it the meter count. Fewer meters, all published, beats a lower headline rate on three of them and a phone number for the rest.

What the tools in a GTM stack actually charge

Every figure below came off the vendor's own pricing page or legal product catalog in July 2026, not off a comparison site. Where a vendor publishes nothing, the row says so, because that absence is a finding rather than a gap in the research. Prices are United States list, before any negotiated discount.

Read the right-hand column rather than the price column. That is where the actual finding sits. Most of the rows a buyer can forecast are seat rows, most of the rows a buyer cannot forecast are consumption rows, and three vendors do not give you a row to reason about at all. The prices are checkable against Salesforce's sales pricing page, its Agentforce pricing page, and HubSpot's product and services catalog, which is the primary document for the onboarding fees and the credit rate.

The meter count

The meter count is one question asked of a stack rather than of a vendor: how many distinct things is this stack metering, and how many of those rates are published? Four meters show up over and over, and they behave differently enough that a single budget line cannot hold them.

1. 1Seats. A per-user rate that changes when headcount changes. The only meter most buyers track, and the only one on the comparison chart.
2. 2Consumption units. Credits, actions, conversations, tokens. The rate is published, the volume is not knowable in advance, so the bill is a guess until the month closes.
3. 3Database volume. Contacts, events, records stored. It grows because the business grew, which means the bill rises fastest in a good quarter.
4. 4One-time gates. Mandatory onboarding, implementation, and migration fees. Not recurring, so they vanish from monthly comparisons, and they land in the quarter with the least slack.

Count them across a real stack and the number is not four, it is four meters spread over five or six vendors, several of whom bill on two at once. A CRM on seats plus credits, an enrichment tool on actions, a marketing tool on seats plus contacts plus credits, a sequencer with no published rate, and a conversation-intelligence tool with no published rate. That is roughly the shape of a normal 2026 stack. Nobody running it can answer what it costs next quarter, and it is not because they are disorganized. There is no number to answer with.

This is why the multiple framing fails on its own terms. To claim the true cost is some multiple of the sticker price, a sticker price has to exist. For the stack described above it does not, and the missing pieces are not small line items. Sequencing and call recording are core spend, and both come from vendors who publish nothing.

Blu Agent

Run the count on your own stack before you run any comparison. List each tool, write its billing meter next to it, then circle every meter whose rate you cannot find on a public page. The circles are your forecasting problem, and they are usually a bigger number than the license you were arguing about.

Where the surprise charges actually come from

Zylo's 2026 SaaS Management Index answers this directly, and the answer is more specific than the folklore. In the past year, 78 percent of IT leaders reported unexpected charges tied to AI features or consumption-based pricing. Another 61 percent cut projects because of unplanned SaaS cost increases. That second number is the one worth sitting with: the cost of an unforecastable meter is not only the overage, it is the work that got cancelled to pay for it.

Notice what those findings do not say. They do not say seats got more expensive. They do not describe a hidden markup on list price. They describe a bill that moved after the contract was signed, on a meter the buyer was not tracking, which is precisely the failure the meter count predicts. Zylo's own framing of the mechanism is that pricing now changes based on usage patterns and tier adjustments that occur long after a contract is in place. The same dynamic ended the standalone signals category, documented here, for a related reason: a line item that cannot be tied to an outcome is the first one cut.

Zylo is the right source for this specific question because auditing it is the company's actual business. The 2026 SaaS Management Index is in its eighth year, built on more than 40 million licenses under management and more than $75 billion in discovered spend. The scale is worth stating plainly because it cuts both ways, which the next section gets into.

One more figure from the same report explains why consumption meters are spreading rather than settling down. Spend on applications where AI is core to the product jumped 108 percent year over year, and use of applications in the broader AI category grew 181 percent, the fastest expansion anywhere in Zylo's dataset. High Alpha's 2025 SaaS Benchmarks Report, which Zylo cites alongside its own data, found 92 percent of SaaS companies had launched or planned to launch AI features. Vendors rebuilding around agentic features are rewriting pricing at the same time, and consumption is the model they are rewriting toward.

The seats nobody is using

Zylo puts unused licenses at 46 percent on average, driving roughly $19.8 million in wasted SaaS spend per year across its dataset. Nearly half of every paid seat is nobody's login. It is the largest single cost in a stitched stack and the only one a buyer can fix unilaterally, without a renegotiation, a migration, or a vendor conversation.

The dollar figure needs a caveat that most posts quoting it skip. Zylo's dataset averages 305 applications and $55.7 million in annual SaaS spend per organization, so $19.8 million in waste is an enterprise number. A 20-person company will never see it. The percentage is what travels. Forty-six percent idle applies to a five-seat sequencer license as readily as to a five-thousand-seat CRM, and on a five-seat license it is the difference between two people using a tool and paying for four.

The reason seats sit idle in a stitched stack is structural rather than sloppy. Every tool needs its own seat for the same person, so one rep consumes a CRM seat, a sequencer seat, a dialer seat, and a conversation-intelligence seat. When that rep leaves, four seats have to be reclaimed in four admin panels by whoever remembers all four exist. This is the same mechanism behind reps not updating the CRM: the work of maintaining the stack lands on somebody whose job is not maintaining the stack. The honest map of which of that work can be handed to an agent instead of a new hire is a separate question, and a narrower one than most vendors admit.

Zylo's shadow-IT data shows where the rest of it hides. In 2026, shadow IT averages 45 percent of an organization's applications, and 51 percent of software purchases are miscategorized in expense reports. In large enterprises, expensed applications make up 41 percent of the software stack while accounting for under 1 percent of total spend, which is the shape of a long tail of cheap tools nobody has counted. Those are the tools that never make it onto the comparison chart and always make it onto the credit card.

What one seat-based price changes, and what it does not

The honest claim for a single platform is narrow. It is not that $49 per seat beats the sum of the point tools, because for a team using four tools lightly it might not. It is that the meter count collapses and every rate ends up on one public page, which converts a guess into a budget line.

Intempt publishes four tiers: $0, $24, $49, and $99 per seat per month, with annual rates at $19.20, $39.20, and $79.20. Each tier publishes its event ceiling, its credit allowance per seat, and its seat limit. Sends are pay as you go on every tier, which is a consumption meter and should be named as one rather than hidden behind the seat number. So the count is not one meter. It is a seat meter with published caps plus one consumption meter, from one vendor, with no mandatory onboarding fee and no tier that requires a phone call to price.

Compare that to the agentic path through the incumbent stack, using Salesforce's own published numbers rather than an estimate. Sales Cloud Enterprise is $175 per user per month, and the Agentforce add-on for Sales is $125 per user per month, which puts a working agentic seat at $300. Agentforce 1 Sales, the packaged edition, is $550 per user per month billed annually with 2.5 million Flex Credits included per org per year. Past that pool, Flex Credits run $500 per 100,000 and conversations run $2 each. The $300 to $550 band is not an outside estimate of what agentic CRM costs. It is arithmetic on the vendor's own price list.

That gap is the reason the agentic GTM platform is a pricing argument as much as a product one. When the enrichment, the sequencing, the call analysis, and the deal all sit on one profile, the same rep does not need four seats, and there are not four admin panels to reclaim them from. The Data Analyst reading the pipeline is reading the same records the outbound ran on, and the Lifecycle Marketer working a nurture track is not billed as a separate login for the same person. That is a product claim, and it is also why the seat count stops multiplying. The structural version of the argument is in GTM platform vs. CRM, and the enrichment-layer version is in GTM platform vs. Clay.

The honest counter-arguments

Four of them, and the first is the strongest. Fewer meters is not the same as less money. A single vendor charging one clean seat rate can cost more than four specialist tools a team uses at the edges of its plans, and consolidation pitched as savings is usually pitched by somebody who benefits from it. The same caution applies to this post and to every side-by-side on this site. Forecastability and total cost are two different variables, and the argument here is for the first one on the grounds that the second is unknowable without it.

Second, consumption pricing is not the villain here. It is what makes a genuinely useful free tier possible, and it is fairer to light users than a flat seat rate that charges a part-time admin the same as a full-time rep. The problem is not the model, it is that a consumption meter with no visible burn rate cannot be planned against. A published rate plus a published cap fixes most of it.

Third, the numbers in this post come from a dataset that does not describe a startup. Zylo's averages sit at $55.7 million in annual SaaS spend across 305 applications. A company running seven tools and 14 seats should read the percentages and ignore the dollars entirely, and should be suspicious of any post that quotes the dollars at them without saying so.

Fourth, general software-waste research does not prove the specific claim this post makes. Flexera's 2026 State of the Cloud Report, based on 753 respondents, found estimated wasted cloud spend on infrastructure and platform services ticked up to 29 percent, reversing a five-year decline, with wasted public cloud software spend up one percentage point. That is real, recent, and adjacent rather than direct. It measures cloud infrastructure, not the admin overhead of a stitched go-to-market stack, and it is cited here as context rather than as evidence.

What this post does not claim

1. 1Not that any analyst firm has published a true-cost multiple for a SaaS stack. The widely repeated 30-to-50-percent version of that claim traces to no locatable report, and it is not used here in any form.
2. 2Not that the meter count is an industry framework. It is this post's own model, built from published price lists. The prices are checkable, the model is an argument.
3. 3Not that Salesforce or HubSpot are overpriced. Both publish their rates clearly, which is more than three of the tools in the same stack manage, and the onboarding fees are in HubSpot's public catalog rather than hidden.
4. 4Not that Gong, Outreach, and Salesloft are doing something unusual. Quote-based enterprise pricing is normal. The point is narrower: a buyer cannot compute a stack list price when three inputs are missing, so a percentage of that list price is not a real quantity.
5. 5Not that unused seats are the whole cost. They are the largest checkable one, at 46 percent in Zylo's data, and the only one a buyer fixes without a negotiation.
6. 6Not that a platform is cheaper for everyone. It is more forecastable, which is a different and more defensible claim, and it is the one the table above supports.

How to price your own stack in an afternoon

1. 1List every tool your revenue motion touches, including the ones on somebody's card. Zylo's 51 percent miscategorization rate is a good reason to check expense reports rather than the vendor list.
2. 2Write the billing meter beside each one. Seats, consumption, volume, or a one-time gate. Several tools take two.
3. 3Mark every rate you cannot find on a public page. These are the rows your forecast is guessing at.
4. 4Pull the actual seat count against active logins for each tool. Expect a gap near Zylo's 46 percent, and reclaim it before you compare anything.
5. 5Add the one-time gates you have already paid and the ones a tier change would trigger. HubSpot's catalog is the model here: the fee is public, it is just never in the monthly comparison.
6. 6Count the same person's seats across tools. One rep holding four seats is the multiplication that makes a stitched stack expensive, and the per-tool math in the real cost of a sales stack gives you a number to argue with.
7. 7Now compare, using meter count and forecastable share rather than headline rate. A stack you can predict at a higher price beats one you cannot at a lower one, because the unpredictable one is what cancels projects.

Step four is the one that pays for the afternoon. Everything else on this list improves a decision you make once a year, but reclaiming idle seats is money back this month, and it needs no approval from anybody. The wider question of which tools belong in the count at all is covered in the GTM tech stack in 2026 and the best tooling for GTM teams. If a tool earned its place by promising to surface buying signals, check whether the signal reaches anything before you renew the seat.

The version that holds up

Read the price lists, not the comparison posts, including this one. Salesforce and HubSpot both publish enough detail to reconstruct a real annual number, and HubSpot's legal product catalog is the more useful of the two documents because it carries the onboarding fees and the credit rates that the marketing pricing page leaves out. Two things to watch while reading: list prices move, and the tier a vendor pushes hardest in a demo is often not the tier whose price you were quoted.

The defensible claim here is small and it does not need a fabricated statistic to land. Nobody knows what a stitched go-to-market stack costs, because four meters are running and several of the rates are unpublished. So stop looking for the multiple, count your meters, reclaim the idle seats first, and treat any tool that will not tell you its rate as a forecasting liability rather than a pricing mystery. Judged that way, one platform that publishes every rate and cap it bills on settles the CRM pricing question on ground a buyer can actually check.