From Aryan Vatsa | Product & Market Analysis

Klaviyo Q3 2026: Revenue Up 25%, NRR at 108%, 12 Agent Metrics, No Dollar Sign

On this page

Klaviyo's Q3 2026 earnings call carried 12 separate figures about its AI agents. Not one of them was denominated in dollars. Revenue grew 25% to $387.2 million and net revenue retention slipped to 108%. The growth the filing can actually trace sits in large customers, multi-product deals and markets outside the Americas. The agents are real. Their revenue is not yet visible.

Key takeaways

  • Klaviyo's measurable growth is concentrated at the top of the customer base. Customers above $500,000 of annual recurring revenue rose 57% to 105, while the total customer count grew roughly 15% to more than 211,000.
  • Net revenue retention fell for a second straight quarter, to 108%. It was 110% in Q1 2026 and 109% in Q2, and the CFO said a prior profile enforcement change will weigh on it through Q1 2027.
  • All 12 agent figures on the Q3 call measure usage, not money. Composer reached more than 10,000 daily active accounts and 18 million weekly credits, measured over 3 weeks, while Klaviyo deliberately gives credits away free.
  • The claim that agents help close deals has no metric behind it yet. The quarter's flagship win, Frasers Group, was explained by management as a choice for the breadth of a single data platform.
$387.2MQ3 2026 revenue, up 25% year over year. Source: Klaviyo Q3 2026 release, November 2026.
108%Net revenue retention, down from 109% in Q2 and 110% in Q1 2026. Source: Klaviyo Q3 2026 release and earnings call.
105Customers paying $500K or more a year, up 57%. Source: Klaviyo Q3 2026 release and call, November 2026.

Short answer

Klaviyo's agents are producing usage, not disclosed revenue. In Q3 2026, the numbers tied to money all point elsewhere. Enterprise cohorts grew 49% to 57%, multi-product deals reached 43% of new ARR, and revenue outside the Americas rose 32%. The agent figures cover 3 weeks, mostly lack a base, and Klaviyo is giving Composer credits away free.

What Klaviyo disclosed in Q3 2026

Klaviyo reported results for the quarter ended 30 September 2026 on 4 November. Revenue was $387.2 million, up 25% from about $311 million a year earlier. Non-GAAP operating income was $56.9 million, a 15% margin, against $45 million and 14.5% a year earlier.

The GAAP picture is thinner. Klaviyo posted an operating loss of $4.5 million, a margin of minus 1%. That is an improvement on the minus 4% it reported in Q2 2026. It still means the profit story depends on the non-GAAP adjustments, mainly stock-based compensation.

Cash generation weakened year over year. Free cash flow was $24.8 million, about 6% of revenue, against $47 million and a 15% margin in Q3 2025. Klaviyo also reports an adjusted free cash flow figure of $40.1 million. Earnings Whispers attributes part of the gap to a $15.3 million purchase of intangible assets.

Klaviyo headline metrics, Q3 2025 against Q2 and Q3 2026.
MetricQ3 2025Q2 2026Q3 2026
RevenueAbout $311M, up 32%$370.6M, up 26%$387.2M, up 25%
Net revenue retention109%109%108%
Customers above $50K ARR3,563, up 36%4,477, up 36%4,725, up 33%
Total customersMore than 183,000More than 205,000More than 211,000
Non-GAAP operating margin14.5%14%15%
Free cash flow$47M$82.9M$24.8M

Q3 2025 figures are from the Q3 2025 earnings call as transcribed by The Motley Fool. Q2 2026 figures are from the Q2 2026 release. Q3 2026 figures are from the Q3 2026 release and call. Free cash flow is seasonal, so the quarter-to-quarter swing is less informative than the year-over-year one.

Guidance carried the deceleration forward. Klaviyo guided Q4 2026 revenue to $421 million to $425 million, about 21% growth at the midpoint. It raised the full-year range by $9 million to $1.537 billion to $1.541 billion. CFO Erica Smith joined in early September. She also declined to give the early look at 2027 that Klaviyo had provided on previous third-quarter calls.

Where Klaviyo's growth actually shows up in the filing

The cleanest way to read this quarter is to ask which customer groups grew faster than revenue. That isolates where the incremental dollars came from, without relying on how management chose to describe them.

Klaviyo Q3 2026: the bigger the customer, the faster the growth Year-over-year growth, quarter ended 30 September 2026. Total customersabout 15% Revenue25% Customers $50K+ ARR33% Customers $250K+ ARR49% Customers $500K+ ARR57% Source: Klaviyo Q3 2026 release and call. Total customer growth computed from "more than" 183,000 and 211,000.
Every cohort above $50,000 grew faster than revenue. Growth is moving up-market, and that is a sales and packaging story before it is an agent story.

The $250K and $500K cohorts carry the quarter

Klaviyo ended Q3 with 390 customers paying $250,000 or more a year, up 49%. Of those, 105 paid $500,000 or more, up 57%. Customers above $50,000 rose 33% to 4,725. Total customers grew only about 15%, from more than 183,000 to more than 211,000.

A simple ratio shows the same shift. Annualised Q3 revenue divided by customer count comes to roughly $7,340 per customer, against roughly $6,800 a year earlier. That is about 8% more revenue per customer, computed from rounded counts, so treat it as directional rather than exact.

Multi-product deals and international markets

Smith said customers using 3 or more products now represent 22% of ARR, up 2 points in a quarter. Fernandez said multi-product deals were 43% of new ARR, up nearly 8 points on a year earlier. Personalisation ARR grew more than 100%. Revenue outside the Americas grew 32%, and EMEA excluding the UK grew 40%.

None of these figures is attributed to agents in the release or on the call. They describe a vendor selling a broader suite to larger brands in more countries. That is a credible growth engine, and it is the one the numbers actually support.

Klaviyo net revenue retention at 108% and why it slipped

Net revenue retention compares what existing customers pay now with what they paid a year ago. It nets out expansion, downgrades and churn. Klaviyo reports it on a trailing 12-month basis. It read 110% in Q1 2026, 109% in Q2 and 108% in Q3.

Smith gave two reasons. Text messaging expansion and cross-sell supported the figure, while lapping last year's profile enforcement offset it. Profile enforcement refers to billing customers for the active profiles they actually hold. It lifted revenue once and now flatters the comparison less. She said that effect "will continue to weigh on the trailing 12-month NRR through Q1".

Growth is stepping down a few points at a time Revenue growth year over year (line) and NRR (labels below). Only quarters with a verified figure are plotted. Q3 2025 Q4 2025 Q1 2026 Q2 2026 Q3 2026 Q4 guide 32% 26% 25% 21% NRR 109% NRR 110% NRR 109% NRR 108% Source: Klaviyo releases and calls, Q3 2025 to Q3 2026. Dashed grey line spans quarters not plotted.
The Q4 guide is the steepest single step on the chart. Klaviyo has beaten its own guides before, so read the red point as a floor, not a forecast of the result.

Management's case that NRR now understates the business

Smith argued that "as multi-product lands becomes the norm, NRR alone will increasingly understate the health of the business". The logic is sound in one respect. If more revenue arrives in the first contract, less is left to show up as expansion in year two.

I do not accept it as a reason to look away from the metric. A company that lands bigger and then expands less is still telling you expansion is slowing. NRR and the large-cohort counts now pull in different directions. Both belong in the read. Retention across other software filings is compared in the 12-filing NRR spread from the Q2 reporting season, where Klaviyo sat at 109%.

The 12 Klaviyo agent metrics, sorted by what they prove

Klaviyo sells two agent products. Composer is a marketing agent that plans and builds campaigns, and Customer Agent handles customer service conversations. Klaviyo also runs an MCP server, which lets outside AI systems such as Claude and ChatGPT read and act on Klaviyo data. Co-CEO Andrew Bialecki and Fernandez gave 12 figures across the three.

Every quantified agent figure on Klaviyo's Q3 2026 call, and what each one lacks.
#Metric as statedWindowMissing
1Composer daily active accounts more than doubled to over 10,0003 weeks after October releasePaid share of those accounts
2Composer accounts are about 5% of Klaviyo daily active accountsSameSplit between free and paid accounts
3Usage more than tripled to 18 million weekly creditsSameHow many credits were paid for
4Credits per active account up 60%Since latest releaseStarting level
5Scheduled tasks tripled3 weeksAbsolute count
6Custom skills doubled3 weeksAbsolute count
7Free accounts using Composer convert to paid at about 1.5 times the rate of othersNot statedBase conversion rate, selection effects
8Over 70,000 accounts using the MCP serverPoint in timePaid share, depth of use
9MCP usage doubling quarter over quarterQuarterlyAbsolute volume
10Customer Agent has exchanged over 2 million messagesCumulativeRevenue under outcome pricing
11Resolution rates up 25 points for one customer3 weeks against prior 3Any second customer
12Customer Agent adoption in the $50K+ cohort up nearly 60% quarter on quarterQuarterlyNumber of adopting customers

Figures are as spoken on the 4 November 2026 call by Bialecki, per the Earnings Whispers transcript. None appears in the press release, which describes the Composer release qualitatively and gives no adoption counts.

Composer: real adoption over a very short window

The Composer figures are the strongest of the set, because one of them comes with a denominator. If more than 10,000 accounts are about 5% of daily active accounts, Klaviyo has roughly 200,000 daily active accounts. That is my arithmetic, not a disclosed figure, and it includes free accounts.

The weakness is time. Most of these numbers cover the 3 weeks after an October release. A product launch with free credits will almost always double usage in 3 weeks. That tells you the launch landed, not whether the habit holds.

Customer Agent: one customer and a cumulative count

Customer Agent's figures are thinner. A 25-point resolution gain for one customer is an anecdote with a number in it. Two million messages is a cumulative total with no period attached. The 60% cohort adoption growth has no base, so 60% could describe a handful of brands or several hundred.

The MCP server is the larger surface

Bialecki said "the usage of our MCP server is higher than Composer today, but Composer usage is growing faster". That is the most strategically interesting sentence on the call. It suggests many customers prefer to reach Klaviyo data from their own AI tools rather than through Klaviyo's agent. That pattern is described in the review of which SaaS incumbents now ship MCP servers.

Testing the claim that agents are closing Klaviyo deals

Fernandez said Klaviyo's largest customers are bought into its agentic roadmap and that "it's helping us close new deals". This is the claim the quarter asks you to accept. It is worth testing against what the same call disclosed about deals.

Four agent claims, four disclosure tests Graded from the Q3 2026 release and call. Dark = disclosed, light = partial, red = not disclosed. Usage figure Base given Over 1 quarter Revenue Composer Customer Agent MCP server Agents close deals Zan Digital grading of Klaviyo disclosures, November 2026. Judgement, not company data.
The right-hand column is red in every row. The claim management leaned on hardest, that agents close deals, fails all four tests.

Frasers Group and the $50K net additions

The quarter's flagship win was Frasers Group, a multi-product, multi-year agreement. Fernandez called it Klaviyo's largest new logo win in EMEA. His stated reason was not agents. "Frasers chose us for the breadth of a single data platform," he said. A Fortune 500 win was described as replacing a legacy marketing system.

The $50K cohort tells a similar story. Klaviyo added 248 customers above $50,000 in Q3 2026, from 4,477 to 4,725. In Q3 2025 it added 272, which management then called a record. If agents were materially improving enterprise close rates, this is the line where it should appear first. It went down.

My read is that the agentic roadmap is a reason enterprise buyers feel safe signing, not a reason they sign. That is still worth something. It is not the same thing as revenue, and the call presented it as if it were.

Why Klaviyo's agent revenue is near zero by design

There is a plain explanation for the missing dollar figures, and Bialecki gave it himself. "Right now, adoption is our priority, so we are deliberately giving customers free credits to try Composer," he said. A company handing out credits to drive usage cannot also report meaningful credit revenue.

Customer Agent is priced differently. Fernandez said "customer agent is priced on outcomes, so our customer success ties to ours". Outcome pricing means Klaviyo is paid per resolved conversation rather than per seat. It aligns incentives, and it also makes the revenue line small and lumpy until volume builds. The trade-offs of that model are set out in the comparison of outcome-based AI pricing meters.

This is a defensible strategy and I would probably make the same call. Charging early for an agent that is 3 weeks past a major release would cap adoption at exactly the point Klaviyo needs data on habit formation. The cost of that choice is that investors are being asked to value usage. The same tension appears in the analysis of how hybrid pricing shows up in NRR.

A sibling comparison helps. HubSpot reported a 72% resolution rate for its own Customer Agent this year. That figure is examined in the HubSpot Q3 agent and seat test. Both companies now publish agent performance. Neither yet publishes agent revenue as a separate line.

Where this argument is weakest

The case above rests on absence: Klaviyo did not disclose agent revenue, so I treat agent revenue as unproven. Absence of disclosure is not evidence of absence. There are three good reasons to hold that conclusion loosely.

The counter-case for agents mattering already

First, the 1.5 times conversion figure is a revenue mechanism, even if it is not a revenue number. Suppose Composer users convert from free to paid plans at 1.5 times the rate of non-users. Then the agent lifts subscription revenue that shows up under email and SMS, not under any agent line. That would make the agent's contribution invisible by construction.

Second, the conversion figure has an obvious selection problem in the other direction. Accounts curious enough to try a new agent are probably more engaged to begin with. Some of the 1.5 times lift would have happened anyway. Klaviyo gave no controlled comparison, so neither reading can be settled from outside.

Third, the primary source for most of the agent figures is a third-party transcript with visible transcription errors. I have not cross-checked it against the 10-Q. The release confirms the financial figures. The agent figures rest on the call alone, and a corrected official transcript could differ in wording.

Finally, 3 weeks is short for my side of the argument too. The Q4 holiday season is the most important quarter for Klaviyo's ecommerce customers. If Composer usage holds through it, the usage case will look much stronger by February.

What would settle the Klaviyo agent question in Q4

Klaviyo said it will give its 2027 outlook on the year-end call. That call is the natural place for a cleaner agent disclosure. These are the four numbers that would move the argument either way.

Four disclosures that would test the agent claim on the Q4 2026 call.
DisclosureWhy it mattersWhat a weak reading looks like
Composer daily active accounts over a full quarterSeparates launch spike from habitGrowth stalls once the October release effect fades
Paid credits as a share of total creditsShows whether usage turns into revenueNo split given, or free credits still dominant
Number of Customer Agent customers on outcome pricingGives the 60% adoption growth a baseA low count across the $50K+ cohort
Share of $250K+ wins that included an agent productTests the claim that agents close dealsAgents mentioned in wins only after the fact

None of these needs Klaviyo to break out a new revenue segment. They are operating metrics the company almost certainly tracks. If the Q4 call carries more usage multiples and no dollars, that is itself a signal. It would fit the wider pattern in the ranking of how software companies disclose AI revenue.

Frequently asked questions

What was Klaviyo's Q3 2026 revenue?

Klaviyo reported Q3 2026 revenue of $387.2 million for the quarter ended 30 September 2026, up 25% from about $311 million a year earlier. Non-GAAP operating income was $56.9 million, a 15% margin, while GAAP operating loss was $4.5 million. The company guided Q4 revenue to $421 million to $425 million, about 21% growth, and raised full-year guidance to $1.537 billion to $1.541 billion.

What is Klaviyo's net revenue retention rate?

Klaviyo's net revenue retention was 108% in Q3 2026, down from 109% in Q2 and 110% in Q1. The CFO said text messaging expansion and cross-sell supported the figure, while lapping last year's profile enforcement weighed on it. Management expects that effect to continue through Q1 2027 and argues NRR now understates growth because more customers start with several products.

What is Klaviyo Composer?

Composer is Klaviyo's marketing AI agent, which helps marketers plan, build and run campaigns using their Klaviyo data. A new release shipped in October 2026. On the Q3 call, management said daily active accounts using Composer passed 10,000, about 5% of Klaviyo's daily active accounts, with 18 million weekly credits used. Klaviyo is deliberately giving free credits to drive adoption.

Does Klaviyo report revenue from its AI agents?

No. Klaviyo's Q3 2026 release and call gave 12 quantified agent figures covering usage, adoption and resolution rates, but none in dollars. Composer credits are partly given away free to drive adoption, and Customer Agent is priced on outcomes. Any agent revenue currently sits inside Klaviyo's overall subscription revenue and is not separately disclosed.

Are AI agents driving Klaviyo's enterprise growth?

Not on the disclosed evidence. Management said large customers are bought into its agentic roadmap and that it helps close deals, but gave no metric linking agents to wins. The largest Q3 deal, Frasers Group, was attributed to the breadth of a single data platform. Customers above $50,000 grew 33%, and quarterly net additions to that group fell from 272 to 248.

What is agentic marketing software?

Agentic marketing software uses AI agents that take actions, such as building segments, drafting campaigns and scheduling sends, rather than only suggesting content. Klaviyo, HubSpot and Salesforce all sell versions of it. The category is young, and vendors typically report usage counts rather than revenue, so buyers should ask for paid usage and outcome data before relying on adoption claims.

How to use this before your next Klaviyo renewal

If you run marketing or ecommerce on Klaviyo, ask your account team one question this week. What will Composer and Customer Agent cost once the free credits end? And how is a credit or an outcome counted? Get it in writing before the Q4 holiday volume lands. Outcome and credit bills scale with exactly that volume.

If you follow Klaviyo as an investor or competitor, note the year-end call date. Then check it for the four disclosures in the table above. The answer to whether agents are a growth line or a retention feature will be in what Klaviyo chooses to count.

Related on this site

For the wider pattern of agent claims against filings, compare HubSpot's Q3 agent and seat test with the NRR spread across 12 Q2 filings.

References

  1. Klaviyo, Klaviyo Delivers Solid Q3 with Enterprise Growth, Multi-Product Strength, and Agent Traction, 4 November 2026. Used for revenue, margins, cash flow, cohort counts, NRR and guidance; text cross-read via the Business Wire reprint on TechIntelPro.
  2. Earnings Whispers, Klaviyo Q3 2026 earnings call transcript, 4 November 2026. Used for all agent figures and quotes from Bialecki, Fernandez and Smith.
  3. Earnings Whispers via Finviz, Klaviyo beats and raises again, but a 21% Q4 guide, slipping NRR and a missing 2027 outlook, 4 November 2026. Used for the Q1 2026 NRR figure and the intangible asset purchase.
  4. The Motley Fool, Klaviyo Q3 2025 earnings call transcript, 5 November 2025. Used for all Q3 2025 baseline figures, including the record 272 net additions.
  5. Klaviyo via Business Wire, reprinted by 01net, Klaviyo delivers strong Q2, 5 August 2026. Used for Q2 2026 revenue, NRR, $50K cohort, margins and free cash flow.

Weakest point in the source base: every agent figure comes from a third-party call transcript with visible transcription errors, and the Q3 2026 Form 10-Q was not opened. Figures last verified 12 November 2026.

MJ
Aryan Vatsa
Writes for Zan Digital on software earnings, AI pricing and what company filings say underneath the call language.

Related reading