Supermetrics today extended its integration with Anthropic's Claude from reporting into campaign management, allowing marketers to create and edit campaigns on Google Ads, Meta, Microsoft Advertising, TikTok, LinkedIn and Snapchat without leaving a chat window. Campaigns created that way stay paused until a person approves them, according to the Helsinki-based company, which said more than 6,000 of its customers have used its Claude connector since it launched in February.

In Short

Supermetrics, a company that collects marketing data from many ad platforms in one place, now lets people ask Claude to build or change ad campaigns on six platforms, including Google and Meta. That matters to advertisers and agencies because the same chat that explains why a campaign slipped can now draft the fix. New campaigns do not go live by themselves: they start paused, and edits can be checked before they take effect.

What Supermetrics announced

The announcement, dated November 3, 2026, positions the update as a move "from analysis to approved campaign action." Since February, the company's pitch for its Claude connector has rested mainly on analysis: pull numbers from many platforms into one conversation and ask questions about them. The new part is the step after the answer.

According to Supermetrics, marketers can investigate performance through Claude using data from advertising platforms, CRM systems, SEO tools, ecommerce platforms and data warehouses, then create or edit campaigns on six advertising platforms in the same conversation. The six are Google Ads, Meta, Microsoft Advertising, TikTok, LinkedIn and Snapchat.

Three controls sit around that write access, according to the company. Newly created campaigns are paused by default before activation. Proposed edits to existing campaigns can be reviewed before they are applied. And customers can set approval requirements for different accounts, so an agency could, in principle, apply stricter sign-off to one client than to another.

Mikael Thuneberg, co-chief executive and founder of Supermetrics, framed the change around the gap between insight and action. "AI has made it dramatically easier to ask questions, but marketers still have to bridge the gap between getting an answer and doing something about it," he said. "We believe the next phase of marketing AI is about closing that gap. Give AI trusted marketing context, let it identify what matters, and then give the marketer the ability to act - while keeping the human firmly in control."

What the announcement does not specify is just as relevant for anyone running paid media. It does not list which campaign types can be built, whether budgets and bids can be changed, which objects below the campaign level (ad sets, ad groups, individual ads) are in scope, or whether every platform supports the same set of actions. Supermetrics' product page for Claude describes creating, starting, stopping and modifying existing campaigns with a single prompt, but the announcement itself stays at the level of "create and edit."

6,000 customers and a growth rate without a base

More than 6,000 Supermetrics customers have used the Claude connector since February, according to the company, with active users growing by an average of 250% month over month. Supermetrics called it one of its fastest-adopted product launches to date, without naming the launches it was compared with.

The connector arrived as part of a broader set of AI integrations. Supermetrics' February 2026 product update, dated February 23, said the company now offered official integrations with leading AI platforms, powered by a new Supermetrics MCP server, and listed Claude and ChatGPT. At that point the update described AI agents querying live marketing data. It said nothing about changing campaigns.

Two figures help place the 6,000 number. Supermetrics says more than 200,000 businesses use its products. On that base, 6,000 customers works out to 3% of the company's reported customer count (PPC Land calculation). The comparison is imperfect, since "customers who have used" a connector at least once is a different measure from paying businesses, but it shows the integration is still a small share of the installed base nine months after launch.

The growth rate is harder to read. The announcement gives no starting number, no monthly series and no definition of an active user. An average of monthly growth rates is sensitive to the first months after a launch, when a small base produces large percentage gains. Read as a 3.5-fold increase each month and sustained over eight months, roughly the span from February to October, growth of 250% a month would multiply a user base about 22,500 times (PPC Land calculation). The figure therefore cannot be converted into a count of current active users, and Supermetrics did not supply one.

Paid-media reporting is "currently the dominant use case," according to the company. That matches what other vendors have reported once marketers gain chat access to ad data. When Adform published its first usage figures for its MCP server on October 28, campaign reporting made up almost a third of the queries. Reporting is where the early demand sits; campaign changes are what both companies are now pushing.

The Layer example

Supermetrics' main customer illustration is Layer, a Norwegian digital marketing agency. According to Supermetrics, Layer reduced a bespoke client-reporting task from up to 10 hours of manual spreadsheet work to approximately 20 minutes using Supermetrics and Claude. That is a reduction of about 97%, or a 30-fold speed-up (PPC Land calculation), measured against the upper end of the range.

A customer story Anthropic published on its own site adds detail. It describes Layer as a 25-person agency that had quoted a client 10 hours of billable work for one report drawing on five data sources, and says the agency recorded no hallucinations during a multi-week validation of live marketing data. Both accounts come from the vendors involved. Neither describes an independent audit, and neither gives the agency's error-checking method.

The example is also about reporting, not about the campaign creation announced today. No figures were offered on how many campaigns customers have created or edited through Claude, how often proposed changes are rejected at the approval stage, or how campaign results compare with those built manually.

Anthropic has published a similar internal case. Its own marketing operations team described cutting a weekly report from up to two days to up to two hours with Claude Cowork, in a July 2026 post. Time savings on reports have become the standard proof point for chat-based marketing tools; evidence on campaign outcomes is still scarce.

Context Layer: business rules carried into every chat

The second element of the announcement is a product Supermetrics calls Context Layer. According to the company, when Claude requests campaign information, Supermetrics returns normalized marketing data, removing the step where marketers export and reconcile figures from different platforms by hand. On top of that data, teams can define targets, reporting rules, client-to-account mappings, naming conventions and events in natural language, at the team, platform or account level.

Those settings then apply across a team's conversations. The practical effect, as Supermetrics describes it, is that a marketer does not need to re-explain which ad accounts belong to which client, what counts as a conversion, or what return a campaign is supposed to hit every time a new chat starts.

"An AI model on its own doesn't know your business, how your marketing data fits together or what 'good' performance looks like for you," Thuneberg said. "That context is what turns a general-purpose AI assistant into something genuinely useful for marketers."

Supermetrics argues that the contest in marketing AI will turn less on access to more powerful models than on access to the trusted marketing context those models need. It is not the only vendor making that case. Typeface extended enterprise brand rules into Claude and ChatGPT on September 22, 2026, carrying brand guidance into third-party chat tools rather than asking users to restate it. Dreamdata, a B2B attribution company, launched an MCP server on September 1, 2026 built so the AI model never recalculates its numbers itself, a response to inconsistent revenue answers from chat tools.

The common thread is a division of labor. The language model handles the conversation; the vendor supplies the numbers and the rules, and tries to keep the model from improvising either. Supermetrics did not say where Context Layer settings are stored, who on a team can edit them, whether changes are versioned, or what happens when a team-level rule conflicts with an account-level one.

Paused by default

The approval design is the part of the announcement with the most direct financial consequence. A reporting error costs time; a campaign error costs money. Supermetrics' answer is that new campaigns are created paused and that edits can be reviewed before they apply.

"The goal isn't to remove marketers from the process," Thuneberg said. "It's to remove the manual work between identifying an opportunity and being able to act on it."

According to Anthropic's customer story, an approval page lays out each change before it goes through. The announcement's wording leaves one question open: because customers "can set approval requirements for different accounts," it is not clear whether an account can be configured to apply edits without review. The company did not say.

The capability is also less new than the framing suggests. Supermetrics' own guide to connecting Claude or ChatGPT to marketing data, dated September 22, 2026, already said that campaign changes were available in supported Claude workflows and that the capability was not then available through Supermetrics for ChatGPT. The same guide said human review should stay in the loop before changes go live. Today's announcement does not mention ChatGPT at all, even though Supermetrics markets its AI chat products for Claude, ChatGPT, Copilot and Gemini.

Others have adopted the same pattern of AI-drafted changes with human sign-off. Amazon presented its Ads Agent at unBoxed in November 2025 with a rule that campaigns only launch after advertisers review and approve the recommendations. X took a different path in August 2026, when it released an Ads MCP server with 23 tools, ten of them write tools that act on live accounts.

Write access through these connectors also changes the security picture. PPC Land reported in July 2025 that tool poisoning attacks target the communication layer between MCP clients and servers. A read-only connector that is manipulated can leak data; one that can create campaigns can also spend budget. Prompt injection, where instructions hidden in content steer a model, is the related risk for any agent that reads external material and then acts. The announcement does not address either.

Platforms are building their own doors

Supermetrics is not the only route from a chat window into an ad account, and the platforms it now writes to have been opening their own.

Meta announced its Ads AI Connectors in beta on April 29, 2026, letting AI agents create and edit ads, ad sets and entire campaigns through Claude and ChatGPT. On July 16, it opened its ads MCP server to any developer with a Meta app. Snap unveiled an MCP server for third-party AI agents in June 2026. Amazon put its Ads MCP Server into open beta on February 2, 2026, and Amazon Ads is not among the six platforms Supermetrics can write to.

Google has moved more cautiously with its own tooling. The open-source Google Ads API MCP server it released in October 2025 was read-only, with two tools. Supermetrics did not say which interfaces it uses to create and edit Google Ads campaigns on behalf of customers.

The positioning question for Supermetrics is plain. A marketer who works only in Meta can connect Claude directly to Meta. The case for a third party rests on breadth: one connector covering six ad platforms plus CRM, SEO, ecommerce and warehouse data, with shared rules on top. It is the argument Supermetrics built its business on with spreadsheets and dashboards, now applied to chat. Other marketing-data companies are pursuing it inside Claude too. Semrush released a native Claude connector covering 28.8 billion keywords in August 2026, and Lifesight's measurement connector, launched in June, is read-only.

Why marketers are watching Claude

Claude's position among marketers has shifted quickly this year. Basis' annual survey, published in October, found ChatGPT used by 74.5% of marketers and Claude by 72.3%, with Claude up from 17.9% in the previous edition, on a sample of 184 respondents. An assistant already in use across that share of the industry is an obvious target for data and campaign vendors.

Spending through autonomous agents remains small by comparison. In August, Magnite's chief executive put 2027 agentic ad spend somewhere between essentially no adoption and $600 million to $700 million. Supermetrics' approach, with a human approving each launch, sits closer to assisted campaign management than to autonomous buying. That distinction matters for how the spend is counted and for who is accountable when a campaign underperforms.

Who Supermetrics is

Supermetrics was founded in 2013 and is headquartered in Helsinki, Finland. According to the company, it has 350 employees across Europe, Asia and North America, and more than 200,000 businesses use its products, including TBWA, Dentsu, Dyson, Heineken, Air France, Levi's and Deloitte. It describes itself as an "AI Marketing Intelligence Platform."

PPC Land last covered the company in January 2019, when it launched Supermetrics for Excel, extending the Google Sheets product that built its business. At the time, Supermetrics reported more than 400,000 users in over 120 countries and data from over 50 marketing platforms, and said it had closed a €3.5 million funding round led by OpenOcean Capital in November 2017. The 2019 figure counted users and today's counts businesses, so the two are not directly comparable.

Seven years on, the company's distribution channel has changed. Spreadsheets were the headline destination in 2019; in 2026 the destination is a chat interface, and the data no longer only flows in one direction.

"If we're asking marketers to trust AI with increasingly important decisions, we have to be willing to use it on our own hardest problems too," Thuneberg said.

What remains unanswered

Several points that matter to advertisers and agencies are missing from the announcement:

  • Pricing and plans. The announcement does not say whether campaign creation in Claude carries a separate charge or which Supermetrics plans include it. Supermetrics' September guide says a trial is created on first login for anyone without an existing account.
  • Scope of edits. Campaign types, budget and bid changes, and objects below campaign level are not listed.
  • Approval settings. Whether approval can be switched off for some accounts, and whether approvals are logged, is not stated.
  • Other assistants. The announcement covers Claude only. Its September guide said campaign changes were not available through Supermetrics for ChatGPT.
  • Adoption base. The 250% monthly growth rate comes without a starting count or a definition of an active user.

Thuneberg described the end goal as "a system that understands what is happening across their marketing, surfaces what matters and helps them take the next action, all while keeping them firmly in control." How often that system's proposals are approved, and with what results, is the evidence the market has not yet seen.

Timeline

Summary

Who: Supermetrics, a Helsinki-based marketing data company with 350 employees that says more than 200,000 businesses use its products, and co-chief executive and founder Mikael Thuneberg. Norwegian agency Layer is the customer example; Anthropic's Claude is the AI assistant involved.

What: Supermetrics extended its Claude integration so marketers can create and edit campaigns on Google Ads, Meta, Microsoft Advertising, TikTok, LinkedIn and Snapchat. New campaigns are paused by default, edits can be reviewed before they apply, and approval requirements can be set per account. A product called Context Layer carries targets, reporting rules, client-to-account mappings, naming conventions and events into every conversation. More than 6,000 customers have used the connector since February, with active users growing by an average of 250% a month.

When: The announcement is dated today, November 3, 2026. The Claude connector launched in February 2026, and a Supermetrics guide dated September 22, 2026 already referred to campaign changes in supported Claude workflows.

Where: The announcement came from Helsinki. The integration runs inside Claude and writes to six advertising platforms; the company has staff across Europe, Asia and North America.

Why: Supermetrics argues that marketers lose time between finding a performance problem and acting on it, and that AI models need company-specific context to make useful decisions. For advertisers, the change moves chat-based AI from reading ad data to drafting changes that can spend budget, with a human approval step as the control. Pricing, the scope of editable settings, and the base behind the growth figure were not disclosed.