Industry Guide: Marketing
How to Use Claude in Marketing: A Guide for Advertising and Brand Teams
From ad brief to live campaign, with a person approving every step.
This guide is for marketing leaders, performance and brand teams, and advertising and martech companies looking at how organizations put Anthropic's Claude to work on the campaigns, creative, and reporting that marketing teams run every day. It is not about one marketer's personal workflow; it covers what named marketing and advertising companies, and Anthropic's own growth marketing team, say about using Claude in Anthropic's own customer stories and blog posts.
Knowing how to use Claude in marketing means pointing it at the work that eats the most hours for advertising and brand teams: generating and screening ad creative, setting up and launching campaigns across ad platforms, answering plain-language questions about live performance, and drafting ad copy against brand guidelines. Marketing and advertising companies such as Advolve and Supermetrics have built Claude into their own platforms to automate parts of this work, and Anthropic's own growth marketing team uses Claude Code internally for creative and copy production. A person still approves every live campaign change and every piece of creative before it ships.
Published
Free lesson · no account neededContext Is EverythingStrategic Prompting5 minStart the free lessonKey Takeaways
- Marketing and advertising companies use Claude most for ad creative generation, cross-platform campaign setup, live performance queries, and ad copy drafting, with a human approval step before anything goes live.
- Advolve uses Claude as the central orchestrator of a platform that generates ad creative, sets up campaigns, and allocates budget across ad platforms for its clients.
- Supermetrics built an MCP server so marketers can ask Claude about live campaign performance and, with explicit human approval, pause or adjust campaigns in the same conversation.
- Anthropic's own growth marketing team uses Claude Code for an ad-creative Figma plugin and a slash command that drafts Google responsive search ad copy for human review.
- Figures from Anthropic's customer stories and its own growth marketing blog post are specific to those companies and workflows; measure results against your own campaigns.
Copy and Paste
Prompts You Can Use Today
Ad creative variation brief
You are a senior creative strategist. Using the campaign brief, brand voice guide, and three approved past ads pasted below, propose five distinct ad creative directions for this campaign. For each, give a headline, a one-line concept, and the audience segment it targets, and flag any claim that is not supported by the brief instead of inventing a benefit. Brief and assets: [paste].
Screening concepts against the brief before anyone touches design is what makes an AI-assisted batch fast to review.
Google responsive search ad copy
Draft a Google responsive search ad from the landing page copy and brand voice guide below: 15 headlines and 4 descriptions, each one checked against our style guide for tone and against the approved claims list for accuracy. Flag any headline that repeats another headline's core message, and export the result as a plain CSV-ready list for manual review before upload. Landing page and style guide: [paste].
Nothing uploads automatically; a person reviews the export before it reaches the ad account.
Cross-channel performance question
Using the connected ad and analytics accounts, tell me how last week's spend, clicks, and conversions on our two highest-budget campaigns compare with the prior four weeks, called out by platform. If any campaign looks like a candidate for a budget change, describe the change you would propose and wait for my approval before touching anything live.
The model answers and proposes; a person still approves any change to a live campaign.
Budget reallocation proposal
Review the campaign performance data below and propose a reallocation of this month's remaining budget across the five campaigns listed, with your reasoning for each move and the expected effect on overall return on ad spend. Present it as a table of current budget, proposed budget, and reasoning, and do not change anything until I confirm the plan. Performance data: [paste].
A proposed change table, not an executed one, is what keeps a human decision in the loop.
Agency client report consolidation
Combine the attached exports from our ad and analytics platforms into one client-ready weekly report: a one-paragraph summary of performance against last week, a table of spend and results by platform, and three recommendations for next week with the reasoning behind each. Flag any number in the exports that looks inconsistent with another export instead of smoothing it over. Exports: [paste].
Flagging inconsistencies between source exports is what keeps a fast report trustworthy.
From Anthropic
What Anthropic offers for Marketing
Anthropic's Growth Marketing Claude Code Workflows
A January 2026 post by a growth marketer at Anthropic describes two internal Claude Code workflows built by the company's own marketing team: a Figma plugin, built in 45 minutes, that generates ad creative variations in one click, and a "/rsa" slash command that drafts Google responsive search ad copy checked against Agent Skills for brand voice and accuracy, exporting an upload-ready file for manual review before anything is published.
· Source
In Practice
How Marketing Teams Use Claude
According to Anthropic's published customer story
Advolve
Advolve, a B2B company Anthropic describes as "Service-as-a-Software," uses Claude as the central orchestrator of an AI platform that automates the full digital-advertising customer-acquisition process for clients such as iFood and Cogna: generating and evaluating ad creative, setting up campaigns across platforms, deploying them, and allocating and optimizing budget. Anthropic attributes Advolve's choice of Claude to code-generation quality and a lower hallucination rate than the alternatives it evaluated. Advolve reports a 90% reduction in operational work time, a 15% increase in customer return on ad spend, and reaching human-level return on ad spend on multi-million dollar budgets in under 30 days, with ad budgets exceeding $100M and 10x revenue projected by the end of 2025.
According to Anthropic's published customer story
Supermetrics
Supermetrics, a martech company serving 200,000 businesses, built an MCP server so marketers can ask Claude about live campaign performance across Google Ads, Meta, Microsoft Advertising, TikTok, LinkedIn, and Snapchat, and, with explicit human approval, pause or adjust campaigns in the same conversation; new campaigns are created paused so nothing goes live without a person's sign-off. Supermetrics cites its client agency Layer, a 25-person shop, using the connector for cross-channel client reporting and media planning: active users grew 250% month over month on average since the February launch, a 10-hour client report fell to 20 minutes, 17 of those spent pasting into Google Sheets, and Layer reported zero hallucinations across a multi-week validation.
Illustrative walkthrough
A creative agency's first-draft review cycle
- A brand team shares the campaign brief, brand voice guide, and a handful of approved past ads with Claude.
- Claude proposes several ad concepts and headline directions that fit the brief and brand voice, flagging any claim it cannot support from the brief.
- The creative lead picks a direction and asks Claude to produce copy variations and layout notes for the design team to build from.
- Legal and brand reviewers check every claim against the approved brief before any version goes to the client.
- Only the version the client approves moves into production and media buying.
Core Workflows
Where Marketing and Advertising Companies Use Claude
These are the places AI for marketing teams and advertising companies saves the most time today, based on what Anthropic's own customer stories describe.
Ad creative generation
AI for ad creative generation and screening, producing multiple directions from a brief and judging them on message fit and brand voice rather than appearance alone.
Campaign setup and launch
Setting up and launching campaigns across ad platforms from a creative and targeting plan, with new campaigns created paused until a person signs off.
Campaign performance queries
Plain-language questions about live performance across ad and analytics platforms, instead of exporting data and rebuilding the same dashboard by hand.
Budget and bid optimization
AI for campaign management covers budget allocation, bid changes, and day-to-day optimization against a return-on-ad-spend target, usually with a sign-off step.
Ad copy drafting
Drafting headlines and descriptions for search and social ads against a brand voice guide and claims policy, with a person reviewing every line before it ships.
Agency client reporting
Consolidating exports from several ad and analytics platforms into one client-ready report, cutting the hours agencies spend pasting numbers into a spreadsheet.
What Marketing Companies Report
How Marketing Companies Use AI Today
How is AI used in marketing and advertising companies right now? Anthropic's own customer stories give two detailed, named examples. Advolve, a B2B company that Anthropic describes as "Service-as-a-Software," uses Claude as the central orchestrator of a platform that automates the full digital-advertising customer-acquisition process for its clients, including named brands such as iFood and Cogna: generating and evaluating ad creative, setting up campaigns across platforms, deploying them, and allocating and optimizing budget.
Supermetrics, a martech company serving 200,000 businesses, took a different angle on ai for digital advertising: it built an MCP server so marketers can ask Claude about live campaign performance across Google Ads, Meta, Microsoft Advertising, TikTok, LinkedIn, and Snapchat in plain language, and, with explicit human approval, pause or adjust campaigns in the same conversation.
AI for marketing agencies shows up clearest in Supermetrics' account of its client agency Layer, a 25-person shop that used the connector for cross-channel client reporting and media planning rather than building its own tooling from scratch. That is how marketing teams use AI most successfully right now: inside an existing platform or workflow, with a clear point where a person signs off before anything goes live.
Across ai for marketing teams, and generative AI in marketing in particular, these two examples follow a similar shape: automation sits inside the platform the company already runs for clients, reporting is treated as a trust problem as much as a speed problem, and the step where a human approves a live change is treated as a feature, not a gap to close later.
Anthropic's Own Playbook
How Anthropic's Growth Marketing Team Uses Claude
Anthropic has also published how its own growth marketing team, not a customer, uses Claude day to day. A January 2026 post by a growth marketer at Anthropic describes a Figma plugin, built in 45 minutes, that generates ad creative variations in one click; the team describes work that previously meant building ten or more variations by hand for a single batch, and a single ad that took 30 minutes to produce now taking roughly 30 seconds to draft.
The same post describes a slash command, "/rsa," that drafts Google responsive search ad copy: Google requires up to 15 headlines and 4 descriptions per ad, and the command checks drafts against Agent Skills tuned for the team's brand voice and accuracy before exporting an upload-ready file, with a person reviewing every export before anything goes live.
Anthropic's post also cites results from other marketing sub-teams at the company: influencer marketing freed up more than 100 hours a month, customer marketing cut case-study drafting from roughly 2.5 hours to 30 minutes and saved about 10 hours a week overall, digital marketing reported a roughly 5x year-over-year gain in output per person, product marketing saved 5 to 10 hours per launch, and partner marketing spent about 40% less time preparing for trade shows.
These are one company's internal numbers, not a benchmark every team will hit. They are useful mainly as a map of where a marketing team's own time goes: creative production, ad copy formatting, case studies, and event prep are all places worth measuring before and after introducing an AI-assisted workflow.
Responsible Use
Keeping a Person in the Loop on Live Campaigns
Is AI safe for live campaign changes? Supermetrics' own design answers this directly: campaigns built through its Claude connector are created in a paused state, so a budget increase, a new audience, or a new campaign only goes live after a person reviews and approves it in the same conversation where Claude proposed it. That single design choice is worth copying into any workflow that lets a model touch a live ad account.
The same split applies to creative and copy. Anthropic's growth marketing post is explicit that manual human review happens before any ad is uploaded, even after a draft has passed an automated brand-voice and accuracy check. A model drafting against a claims policy is not the same as a person confirming the claim is still true for this specific campaign and market.
Treat reporting the same way. Supermetrics' client agency reported zero hallucinations across a multi-week validation of its own reports, which is a strong result precisely because someone validated it against the source data for weeks before trusting the workflow. Build that same validation period into your own rollout before using an AI-assisted report as the basis for a client or budget decision.
Choosing Tools
Claude vs ChatGPT for Marketing
The Claude vs ChatGPT for marketing comparison is most useful run on your own work: a creative brief you already built by hand, a client report you already finalized, and a campaign budget decision you already made. Compare how each tool handles your actual brand voice, your actual claims policy, and your actual mix of ad platforms, not a generic demo.
For marketing and advertising companies specifically, the decision usually comes down to what each tool can connect to. Supermetrics and Advolve both built their Claude integrations around their own platforms' data and actions; whichever model a marketing team or agency adopts, the deciding factor is often which one its existing ad, analytics, and creative tools can be connected to safely.
Governance matters as much as capability. Who approves a live campaign change, how creative claims get checked against policy, and where client data is allowed to flow are all questions your legal, brand, and media teams should answer before any AI tool touches a client account, regardless of which one you pick.
Getting Started
A Safe First Month for a Marketing Team
Start with finished work
Test prompts on a brief, a report, or a budget decision you already finished, and compare the AI-assisted draft against what you actually shipped.
Load your brand voice and claims policy
Put your brand voice guide, approved claims list, and past approved ads into a project so drafts start from your own standards, not generic copy.
Pick one high-volume task
Choose one repeated task, such as responsive search ad copy or weekly client reporting, and measure review time before and after.
Define the sign-off step
Agree who approves a live campaign change or a published ad before any workflow is allowed to touch a real ad account.
Limits
Where Marketers and Agencies Must Decide
Claude can misjudge a creative direction, misread a performance trend, or draft a claim that sounds plausible but is not accurate for this product or market. Treat any AI-drafted output the same way you would treat a junior team member's first draft: useful, but checked before it reaches a client or a live budget.
What are the biggest AI use cases in marketing right now, based on the public record? Ad creative generation, campaign setup and optimization, cross-channel performance queries, and ad copy drafting, in that order of how much detail Anthropic's own customer stories give. Each one still ends with a person reviewing the result against brand, legal, and budget standards.
Client data, campaign budgets, and brand reputation are all on the line in marketing work, which is why the strongest examples here, Advolve's platform and Supermetrics' connector, both keep a human approval step at the exact point where a change becomes irreversible: before a campaign goes live, and before an ad is uploaded.
Build Reliable AI Habits for Campaign Work
Getting trustworthy drafts from AI is a practiced skill: grounding in your own brand, checking claims, and knowing where to keep a human sign-off. The free Nightschool AI curriculum is hands-on from the first lesson.
Frequently Asked Questions
How is AI used in marketing and advertising?
Mainly for ad creative generation, campaign setup and optimization across ad platforms, live performance queries, and ad copy drafting. Anthropic's customer stories describe Advolve orchestrating ad campaigns and Supermetrics connecting marketers to live performance data, and Anthropic's own growth marketing team uses similar workflows internally, each with a human approval step before anything reaches a live account.
Can AI manage ad campaigns?
It can set up, launch, and propose optimizations for campaigns, as Advolve's platform does for its clients. Supermetrics' connector goes further and lets a person pause or adjust a live campaign inside the same conversation, but campaigns are created paused by design, so nothing changes without explicit human approval first.
Can AI generate ad creative?
Yes. Advolve uses Claude to generate and evaluate ad creative as part of its automated acquisition platform, and Anthropic's own growth marketing team built a Figma plugin that generates ad creative variations in one click. A person still reviews creative before it ships in both cases.
How much time can AI save in marketing?
It varies by task and company, so measure it on your own work. Anthropic has reported results ranging from large cuts in operational work time at one customer to a single-digit-minute drop in ad production time on its own growth marketing team; neither figure is a guarantee for a different team's workflow.
What are the biggest AI use cases in marketing?
Based on the public record, ad creative generation, campaign setup and budget optimization, cross-channel performance queries, and ad copy drafting against a brand voice guide. Agency client reporting is close behind, especially for shops juggling several ad and analytics platforms at once.
Is AI safe for live campaign changes?
It can be, with the right design. Supermetrics' connector creates new campaigns paused and requires explicit human approval before any pause, budget change, or adjustment takes effect on a live campaign, which is the pattern worth copying into any similar workflow.
Can AI write ad copy?
Yes. Anthropic's growth marketing team uses a slash command that drafts Google responsive search ad copy and checks it against brand-voice and accuracy guidelines before export, and a person reviews every export before any ad copy is uploaded to a live account.
Claude vs ChatGPT for marketing: how should a team decide?
Run both on your own brief, report, or budget decision rather than a generic demo. For marketing and advertising companies, the practical deciding factor is usually which tool can safely connect to the ad, analytics, and creative platforms your team already runs on.
Keep going
Related Guides and Courses
Related reading
- AI Training for Work
- Free AI Prompt Generator for Work Tasks
- How to Use Claude in Real Estate
- How to Use Claude in Education
- Claude for Marketing
- Claude for Sales
- Claude for Writers
- Private Equity Due Diligence With Claude: Data Rooms to Findings
- How to Do CIM Analysis With Claude: Screening Deals and Drafting the IC Memo
- LBO Modeling With Claude: What It Builds, What You Still Have to Check
- Venture Capital Deal Sourcing With Claude: A Thesis-Driven Workflow
Claude in Marketing, by role
Draft Faster, Keep a Human in the Loop
Hands-on lessons for marketing and advertising teams.