Klaviyo Pricing in 2026: A Calculator Whose Answers Are Not in the Page

Klaviyo's pricing page is a calculator, and the calculator's answers are not in the page. That is the whole of the pricing story. Everything below is the evidence for it, the figures the vendor does publish elsewhere, and what to do about the gap.

What the page actually serves

Read from Ashburn on 8 October 2026, the page returns 601,147 characters of markup containing thirty-two dollar amounts ^[2]. None is the price of an email tier. Twenty-three are the free plan's two five-dollar allowances. The others are Composer at twenty dollars, the Flex rate of ten dollars per thousand credits, and a nineteen-dollar first-month discount.

The only profile counts anywhere in those 601,147 characters are 250 and 500, and both belong to the free plan. The payload names pricingCalculator, sliderLabel and sliderTooltip fifteen times each, and priceAmount four times, so the widget exists and is wired. Its output is produced in the reader's browser and never leaves the server as text.

## The test that settles it

Request the page at ?contacts=5000. Request it again at ?contacts=25000. Compare the dollar amounts. They are identical in every value and every count ^[3].

A parameter that looks like it selects a rung changes nothing the server sends. The consequence runs past curiosity: no price above the entry tier is published to a reader, a crawler, or an assistant that does not run the page's JavaScript to completion. If you have ever asked a model what Klaviyo costs and received a vague answer, this is why.

The only published price, and where the vendor put it

One price for the email product exists on the marketing site, in the terms of a discount: "Applicable on paid $20 Klaviyo Email plans only (not applicable on any SMS-only or Email+SMS plans, or any other Klaviyo product)" ^[4].

Twenty dollars a month, scoped by the promotional banner to five hundred profiles. The nineteen-dollar credit makes month one a dollar, and the account then "will automatically renew at the then-current standard monthly rate until cancelled" ^[4]. The offer runs through 30 November 2026 and requires that the account has held no paid Klaviyo plan in the previous twelve months.

Read the forfeiture clause before taking it: "Offer is valid only one time and will be forfeited if the account is downgraded, paused, or terminated during the Discount Period" ^[4].

## The rungs that are published, and what they were published for

Figure	Source	The question it was answering
$80 a month, 35,000 emails, $1.14 per thousand	February 2026 FAQ [6]	what a flexible-sending overage costs
$100 a month, 5,000 profiles, 50,000 emails	changing-plans article [8]	how a mid-cycle upgrade is billed
$140 a month, 6,500 profiles, 65,000 emails	changing-plans article [8]	the destination of that upgrade
$4,765 a month, 1,000,000 profiles	billing article [5]	when a usage fee beats upgrading
$9,100 a month, 2,000,000 profiles	billing article [5]	the tier that example declines

Every one is a real published price. None was published as a price. The documentation needed a number to explain arithmetic, used a true one, and moved on. If your band is not one of these five, your figure is not on the open web.

The $100-to-$140 example is the most useful of them, because it shows the shape of the ladder rather than a point on it. Forty dollars buys 1,500 more profiles and 15,000 more emails ^[8]. That is the local slope, and it is the only slope the vendor publishes for the email product.

The other five lines of your invoice

Product	Entry price	Unit
Customer Hub	$20 a month	up to 10,000 active profiles
Helpdesk	$10 a month	up to 50 shopper-initiated tickets
Customer Agent	$50 a month	up to 75 resolved conversations
Marketing Analytics	$100 a month	up to 13,500 active profiles
Composer	$20 a month	AI credits; Flex at $10 per 1,000

The first four are published in four consecutive sentences of one support article ^[5]. Together with the entry email plan they total two hundred and twenty dollars a month. That is the figure to carry into any comparison, not the twenty.

Each meter counts something different. Composer bills on AI credits consumed within a billing cycle; Reviews bills on orders placed within a cycle; Helpdesk on tickets opened; Customer Agent on conversations resolved ^[5]. Only the first is really under your control, and the account usage page shows each over the last three, six or twelve cycles.

Two thresholds change the shape of the account rather than its price. Flexible sending for SMS is not offered below the first: "Note that flexible sending for SMS is available for plans priced $490/month and higher" ^[6]. Card billing gives way to invoicing at the second, when "Your Klaviyo plans total $2,000 or more per month" ^[9].

## What the extra two hundred dollars buys

The add-ons are not trim. Each is a product another vendor would sell on its own, which is why the invoice grows the way it does.

Composer is the AI marketing agent, described by the vendor as bringing together "what it knows about your business, Klaviyo's expertise, and the tools you already use to find opportunities, decide what to do next, and help you act on it" ^[14]. It bills on AI credits consumed in a cycle, with the $10 per 1,000 Flex rate when the allowance runs out.

Customer Hub is a logged-in area for shoppers, carrying rewards, offers and loyalty balances alongside reorders and recommendations ^[12]. It bills on active profiles and, unlike the email plan, moves down as well as up on its own.

Advanced Klaviyo Data Platform is the piece that justifies the CRM framing. It adds memberships, appointments, subscriptions and other custom data to profiles for use in segmentation, resolves returning visitors to known profiles, and gives AI tools governed access to the data ^[12]. It is also the only product whose ladder the vendor publishes two rungs of, at $4,765 and $9,100.

Helpdesk and Customer Agent are the support pair, billed on tickets opened and conversations resolved. They are the two that cannot be capped.

Whether that bundle is worth two hundred and twenty dollars a month depends entirely on whether you would otherwise buy four of those things separately. For a consumer brand running loyalty, a shopper portal and support on one customer record, the arithmetic is often favourable. For a business that wants to send good email, most of it is a line you can decline.

How the bill moves, in both directions

Upward movement is automatic, enforced and switched on at purchase. "You will be automatically upgraded to a profile-compliant plan at the end of each billing cycle, if you exceed your level of profile usage" ^[5], and "When you first purchase a subscription for the Profile & Email plan or the Mobile Messaging plan, auto-upgrade is toggled on by default" ^[11].

Downward movement is yours to arrange. "Your plan does not automatically downgrade when you reduce profiles; you must choose a lower tier in the Billing page after your billable profile count has updated" ^[5]. It has a deadline, a penalty for missing it, no refunds, and an invoice lock ^[5].

An auto-downgrade preference exists, and here is its catch: "auto-downgrade can be used with flexible sending but not with auto-upgrade" ^[11]. Since auto-upgrade is the default, turning on the protection against overpaying means turning off the default protection against interruption. The two are mutually exclusive, and only one of them is pre-selected.

Getting all the way back costs more than a downgrade. "You can only downgrade to a free plan by canceling; the free plan does not appear as a regular downgrade option" ^[8].

## The arithmetic the vendor gives you

Two published rules are genuinely useful, and most vendors offer no equivalent.

The first is the break-even: "Once you're using 85% or more of the next tier, it's better to upgrade" ^[5]. Below that threshold, paying the overage is cheaper than moving up. The vendor is telling you when not to give it more money, which is worth noting in a corpus where that is close to unique.

The second is the overage formula on the data platform, with a worked example. A plan at $4,765 covering a million profiles, five hundred and twenty thousand profiles over, charged at $4.77 per thousand at that tier, produces a usage fee of $2,480.40 and a total of $7,245.40: against $9,100 for the next tier up ^[5].

Flexing has its own published example on the messaging side. On a $900 monthly plan with Flex on, sending $112.50 beyond the plan's spend is billed as $112.50 once, rather than moving the plan, and the account returns to $900 at the next cycle ^[7]. You can flex more than once in a cycle without the plan changing ^[11].

Proration applies to three products and not to the rest

From 24 October 2025 the vendor prorates first purchases and upgrades of Customer Hub, Marketing Analytics and Advanced Klaviyo Data Platform: buy a $100 Marketing Analytics plan halfway through a cycle and you pay $50 for that cycle, then $100 thereafter ^[5].

Everything else follows the opposite rule, stated just as plainly: "Plans are not pro-rated if you add them mid-billing cycle" ^[8]. An upgrade within a tier is handled differently again, by charging the difference: moving from $100 to $140 mid-cycle costs $40 that month ^[8].

Three rules, three products groups, one invoice. It is worth knowing which of the three your next purchase falls under before you make it mid-month.

## What changed in 2026, twice

Billing moved twice in eight months, and the vendor dates both changes. On 18 February 2026 it introduced flexible sending and the auto-downgrade preference ^[6]. On 13 July 2026 it moved Mobile Messaging from credits to dollars: "You still purchase your plan upfront (the same commit model as today), but your plan is now denominated as a dollar amount" ^[7]. Existing subscribers moved on their first billing cycle after 13 August 2026.

Publishing the dates and the mechanics is to the vendor's credit, and we have scored it there. Needing two changes in eight months is something to price into a twelve-month plan.

What is not published at all

The price of any email tier above the entry one. The number of tiers. The profile band attached to the twenty-dollar plan, anywhere but a promotional banner. The per-message rates for mobile messaging by country, which the vendor describes as "the list (standard) price per message for a given channel, country, and number type" without publishing a card ^[10]. The volume-discount curve, published as a rule without a table. The ceiling of any plan. The tier prices for Helpdesk and Customer Agent above their entry points. And any annual price at all, because the site sells month to month and publishes no yearly alternative.

## What customers report paying, against what the vendor documents

We coded a hundred Trustpilot reviews by theme. Billing appears in thirty-one of them and thirty of those are one or two stars; downgrading or cancelling appears in seventeen, and all seventeen are one or two stars. Only five reviewers in a hundred say the product is simply too expensive ^[20].

That distribution is the useful part. The grievance is not the price. It is the shape of the ladder: "They bill you based on outdated subscriber list sizes, so even after you clean house, you're stuck overpaying for profiles that don't exist anymore." Another reports being "charged almost three times my normal monthly bill because of profiles that were obviously generated during development/testing", which is what billing on eligibility rather than engagement means when a staging environment writes to production.

None of this contradicts the documentation. It is the documentation, experienced from the paying side.

## What to do with this

Get the price of your band in writing before you commit, and get the band definition with it: the count that bills you is the one at the end of the cycle, not the average across it. Decide at signup whether auto-downgrade should replace auto-upgrade, because you cannot have both and the billing default is the one that raises the bill. If your list is seasonal, model the month after your peak rather than the peak itself. If you run test or development traffic into the same account, expect it on the invoice, since a profile is billable when it is reachable rather than when it is engaged. And if you expect to use Helpdesk or Customer Agent, note that neither can be capped, so your ceiling on those two is whatever you remember to switch off.