All Things Growth and Marketing

AI marketing automation means putting AI models inside automated workflows so software can handle marketing steps that need reading, writing or judgment, such as summarizing a sales call, scoring an account against your ICP or flagging an odd drop in ad performance. Classic marketing automation follows fixed rules: if a lead downloads an ebook, send email two. AI automation can read the lead’s company website first and decide what email two should say.

It is also where most marketers are least experienced. In CXL’s 2026 survey, 65% of B2B marketers rated themselves beginner or below at building AI workflows and agents, and 42% said building agents was the AI skill they most wanted to learn. This guide covers the difference between workflows and agents, eight workflow templates B2B teams use, and the build process we teach at CXL.

Key takeaways

• Start with workflows, not agents. A workflow follows steps you define; an agent decides its own steps. Most marketing jobs are better served by a workflow with one or two AI steps.
• Most “AI agents” aren’t. Gartner estimates only about 130 of the thousands of vendors selling agentic AI offer real agents, and predicts more than 40% of agentic AI projects will be canceled by the end of 2027.
• The gains are measured. In an MIT field experiment with 2,234 people, teams working with AI agents produced 50% more ads per worker.
• Human review stays in the loop. The workflows that work send AI output to a person for approval before anything reaches a customer or spends budget.
• You can’t automate a process nobody has written down. Documenting the job is step one, and the step most teams skip.

Classic automation vs AI workflows vs AI agents

The terms get used loosely. This is the distinction that matters when you decide what to build:

Anthropic’s engineering team draws the same line in Building effective agents: workflows run language models “through predefined code paths”, while agents “dynamically direct their own processes and tool usage.” Their advice to developers applies to marketers too: find the simplest solution possible and add complexity only when it is needed.

What is an AI marketing agent?

An AI marketing agent is an AI system that is given a marketing goal and a set of tools, such as web search, your CRM or a spreadsheet, and works out the steps to reach the goal itself. Ask it to “build a research brief on Acme Corp before Thursday’s call” and it decides what to look up, in what order, and when it has enough.

Be careful with vendor claims. Gartner calls the practice of rebranding chatbots and simple automations as agents “agent washing”, and estimates that only about 130 of the thousands of agentic AI vendors offer the real thing. It also predicts that more than 40% of agentic AI projects will be canceled by the end of 2027 because of rising costs and unclear business value. The pattern behind most failures is the same one CXL saw when we spent a week building marketing automations: one “super-agent” asked to research, write, optimize and publish does all four badly. Specialized steps with clear jobs work better.

The five building blocks of AI workflows

Almost every useful AI marketing workflow is built from five patterns, described in Anthropic’s guide. Knowing them makes it easier to design your own:

8 AI marketing workflows worth building

These are workflows B2B teams run in production. Each has a clear trigger, a defined output and a human check. Use them as templates and adapt the inputs to your stack.

Step-by-step builds for several of these are on the CXL blog: ICP research and ad reporting with n8n and Claude Code, a competitor ad intelligence tool, an AI SEO agent for Search Console, an AI visibility tracker, a paid media co-pilot and an AI ABM fit score from CRM data.

How to build your first AI marketing workflow

1. Choose a job, not a tool. Pick something recurring that takes hours and has a clear output. Workflows 1, 3 and 4 above are good first builds because nothing goes to customers.
2. Write the standard operating procedure. List the inputs, every step, each decision and what a good output looks like. If you can’t write it down, the AI can’t follow it, and you will spend the build arguing with it.
3. Map where AI is needed. Most steps are plain automation: fetch data, filter rows, post to Slack. Use AI only for the steps that need reading, writing or judgment. Fewer AI steps means fewer places for errors.
4. Do it manually with AI once. Run each AI step by hand in ChatGPT or Claude with your real data. Fix the instructions and context until the output is something you would ship. Then automate.
5. Build the smallest version. Start with two data sources and one output. In n8n, Zapier or Make that is often under ten steps. Add sources and branches only once the core works.
6. Add an evaluator and an approval gate. Have a second AI step check the output against a short rubric, and route anything customer-facing or budget-moving to a person for approval.
7. Log every run and measure it. Store inputs and outputs so you can see what went wrong. Track time saved and the business metric the job serves for at least a month.

Teams that follow this order usually have a working first workflow within one or two weeks. Teams that start by building an agent usually have a demo.

Which tools do you need for AI marketing automation?

• n8n: a workflow automation tool you can self-host or use in the cloud, popular with technical marketers because AI steps, code and APIs sit in one visual canvas.
• Zapier: the largest library of app connections and the easiest start for non-technical teams, with AI steps and agent features built in.
• Make: a visual scenario builder that handles branching logic well.
• Claude Code: Anthropic’s agentic coding tool, which lets marketers build custom scripts and small apps by describing what they want. Useful when a workflow tool gets in the way.
• AI inside your platforms: CRMs and marketing automation platforms now include AI features and agents that work directly on your existing data. Check these before building from scratch.

You don’t need to code to start. You do need to think like someone designing a process: inputs, steps, checks and outputs. That skill set is what employers now hire for under the title marketing engineer.

What goes wrong with AI marketing automation

• One agent doing everything. A single agent that researches, writes, optimizes and publishes performs worse than separate steps with clear jobs.
• Trusting one data source. Ad platforms, analytics and your CRM rarely agree. A workflow that reads only one will report confident numbers that sales won’t recognize.
• Automating an undocumented process. The AI fills the gaps with guesses, and nobody can tell which output is wrong.
• No human gate on customer-facing output. One hallucinated claim in an email to 10,000 prospects costs more than the workflow saved all year.
• Nobody owns it. Workflows break when an API changes or a field is renamed. Someone has to be responsible for checking the logs.

For a candid account of what worked and what didn’t when CXL built five of these systems, read what we learned from n8n week.

When you shouldn’t use AI automation

• The task happens once a quarter. Building and maintaining the workflow will take longer than doing the job by hand.
• The output is a judgment call you’re accountable for. Budget cuts, pricing changes and messaging to angry customers deserve a person’s full attention, not an approval click.
• The process is broken. Automating a report nobody reads just produces more unread reports, faster.
• Your data isn’t trustworthy. An AI workflow on broken tracking scales the error. Fix the data first.

AI marketing automation FAQ

What is AI marketing automation?

AI marketing automation is the use of AI models inside automated marketing workflows to handle steps that need reading, writing or judgment, such as summarizing calls, researching accounts, writing first drafts or spotting anomalies in campaign data. It extends classic rule-based marketing automation.

What is the difference between marketing automation and AI marketing automation?

Classic marketing automation follows fixed rules on structured data, such as sending an email when a form is filled. AI marketing automation can work with unstructured input, such as text, web pages and transcripts, and make decisions inside the workflow, such as what an email should say.

What are AI agents in marketing?

AI agents are AI systems that receive a goal and tools and decide the steps to reach the goal themselves, for example researching an account before a sales call. They are more flexible than workflows but less predictable, so most marketing teams use them for research tasks and keep fixed workflows for anything customer-facing.

Can AI agents run marketing campaigns on their own?

Not reliably in 2026. Agents can research, draft, analyze and monitor well, but campaigns involve budget, brand and customer decisions that still need a person to approve them. Gartner expects agentic AI to make about 15% of day-to-day work decisions autonomously by 2028.

Do I need to know how to code to build AI marketing workflows?

No. Tools like Zapier, Make and n8n are visual, and Claude Code lets you build custom tools by describing them in plain language. What you do need is the ability to break a job into clear steps and define what good output looks like.

What is the best first AI workflow for a marketing team?

An internal one with a clear output, such as a weekly competitor digest or a daily paid media anomaly alert to Slack. Nothing reaches customers, so mistakes are cheap, and the time saved is easy to measure.

Learn to build AI marketing systems

CXL is a B2B marketing training company focused on AI and conversion optimization. In CXL’s Marketing Engineer cohort, you pick the pipeline job that eats your week and spend eight live weeks building the AI system that does it, with feedback from practitioners. You are certified on what you build, not on attendance. For background, start with our guide to AI marketing, our research on how B2B teams use AI and our list of the best AI marketing tools.

Sources

• Anthropic. Building effective agents. December 2024.
• Gartner. Gartner predicts over 40% of agentic AI projects will be canceled by end of 2027. June 2025.
• Ju, H. and Aral, S. Collaborating with AI Agents: Field Experiments on Teamwork, Productivity, and Performance. 2025.
• Brynjolfsson, E., Li, D. and Raymond, L. Generative AI at Work. The Quarterly Journal of Economics, 2025.
• CXL. How teams use AI in B2B marketing in 2026 and What we learned from n8n week.

The best AI marketing tools for most B2B teams in 2026 are a general assistant (Claude, ChatGPT or Gemini), a research tool (Perplexity), an automation tool (n8n, Zapier or Make) and the AI already built into the platforms you pay for, such as HubSpot or Ahrefs. Everything else is a specialist tool that is only worth buying once you know which job it will do.

Most “best AI marketing tools” lists rank 30 apps with the same praise for each. This one is shorter and organized by job. Every tool below earns its place for B2B marketing work, each entry says who it is a bad fit for, and every price was checked on the vendor’s own pricing page on November 10, 2026.

Key takeaways

• Start with one assistant, not five. Claude, ChatGPT and Gemini cost about $20 a month each for individuals. Pick one and give it your positioning, ICP and examples.
• Employers now ask for named tools. In CXL’s analysis of US marketing job posts, mentions of specific AI tools tripled from 5% to 15% between January and May 2026. Claude mentions rose from 1.1% to 4.6%, ChatGPT from 3.4% to 5.5% and Perplexity from 0.7% to 2.2%.
• Automation is where the time savings are. n8n, Zapier and Make turn one-off prompts into workflows that run every week without you.
• Check what you already pay for. HubSpot, Ahrefs, Canva and most other marketing platforms now ship AI agents and features inside existing plans.
• A capable stack costs under $50 a month. Claude Pro, Perplexity Pro and Canva Pro, billed annually, come to about $46 a month per person.

AI marketing tools compared

Prices are for one user and were checked on each vendor’s pricing page on November 10, 2026. AI pricing changes often, so confirm before you buy.

How we chose these tools

CXL trains B2B marketers to use AI in real work, and we build our own marketing workflows with several of these tools. We included a tool only if it met three tests:

1. It does a recurring B2B marketing job well, such as research, content, reporting, enrichment or monitoring, not just a demo.
2. It has a usable entry point, either a free plan or a price a single marketer can justify.
3. It is not a thin wrapper. Dozens of “AI copywriting” apps resell the same models as the assistants below at a higher price with less control. We left them out.

Best AI assistants for marketers

A general assistant is the one AI tool every marketer needs. The three leaders are close in raw capability; the differences that matter are in how they fit your work.

1. Claude

Best for: long-form writing, editing to a style guide, research synthesis and marketers who want to build their own tools.

Claude, from Anthropic, is strong at following detailed instructions and keeping a consistent voice across long documents, which makes it a favorite for content and messaging work. Projects let you store your ICP, positioning and examples so every chat starts with context. Paid plans include Claude Code, Claude in Chrome and Microsoft 365 integration. It is also the assistant employers are adding fastest: Claude mentions in marketing job posts quadrupled in the first five months of 2026, in CXL’s job post analysis.

Trade-off: usage limits on the Pro plan can bite during heavy days; power users end up on Max at $100 a month.

Pricing: Free plan. Pro $20/month, or $17/month billed annually. Team from $20 per seat per month billed annually.

Learn more: how to make AI write in your brand voice.

2. ChatGPT

Best for: an all-round assistant with the broadest feature set: image creation, voice, deep research, custom GPTs and scheduled tasks.

OpenAI’s ChatGPT is the most widely used assistant and the one named most often in marketing job posts (5.5% in May 2026). Custom GPTs let you package a repeatable task, such as a brief writer or ad copy reviewer, and share it with your team. Scheduled tasks can run a prompt on a timetable, a light first step toward automation.

Trade-off: the wide feature set changes often, and plan limits are harder to predict than a fixed number of messages.

Pricing: Free plan. Go and Plus for individuals; Plus is $20/month. Business plans start at two users.

3. Gemini

Best for: teams that work in Gmail, Docs, Sheets and Slides all day.

Google’s Gemini is built into Workspace apps, so it can draft in Docs, summarize email threads and analyze data in Sheets without copy-pasting. For a team already on Google Workspace, that integration often matters more than small quality differences between models.

Trade-off: outside Google’s own apps, it is less flexible than Claude or ChatGPT for building custom workflows.

Pricing: Free plan. Google AI Plus $4.99/month and Google AI Pro $19.99/month for individuals; Workspace business plans bundle Gemini features.

Best AI tool for marketing research

4. Perplexity

Best for: fast, sourced answers on markets, competitors, people and trends.

Perplexity searches the web and answers with citations you can click and check, which makes it faster than a search engine for questions like “what has this competitor launched this year?” Its deep research mode produces multi-page reports. It is also a useful way to see what AI search engines say about your brand and category, which matters for answer engine optimization.

Trade-off: answers are only as good as the sources it finds. Click the citations before quoting a number.

Pricing: Free plan. Pro $17/month billed annually. Enterprise Pro $34 per seat per month billed annually.

Best AI marketing automation tools

Automation tools turn a prompt you run by hand into a workflow that runs on a schedule or a trigger. This is where AI saves hours instead of minutes. For how to design them, see what we learned building marketing automations in n8n.

5. n8n

Best for: technical marketers and marketing ops building multi-step AI workflows and agents.

n8n puts AI steps, code, APIs and hundreds of app integrations on one visual canvas, and you can self-host it. It is the tool behind several of CXL’s own builds, including an AI visibility tracker and an AI SEO agent for Search Console.

Trade-off: steeper learning curve than Zapier; you need to be comfortable with JSON and API concepts.

Pricing: free self-hosted Community edition. Cloud Starter €20/month billed annually for 2,500 workflow executions; Pro €50/month.

6. Zapier

Best for: non-technical teams that want AI steps across the widest range of apps.

Zapier connects more apps than any rival and adds AI steps, Copilot for building workflows in plain language, and separate AI agent and chatbot products. If a tool in your stack has an integration, Zapier probably has it.

Trade-off: per-task pricing gets expensive for high-volume workflows, and complex logic is harder to build than in n8n or Make.

Pricing: Free plan with 100 tasks a month. Professional from $19.99/month; Team from $69/month. Agents and Chatbots are priced separately.

7. Make

Best for: visual thinkers building workflows with lots of branching and data transformation.

Make’s visual scenario builder shows exactly how data moves between steps, which makes complex workflows easier to debug. It includes an AI assistant, Maia, and AI agent features.

Trade-off: Make bills in credits, and AI-heavy steps can consume more credits than simple actions, which makes costs harder to predict.

Pricing: Free plan with 1,000 credits a month. Paid plans (Core, Pro, Teams) are priced by credit volume.

Best AI tools for building marketing assets

8. Claude Code

Best for: marketers who want custom tools, scripts and automations without waiting for a developer.

Claude Code is Anthropic’s agentic coding tool. You describe what you want, such as “pull last week’s Google Ads data, compare it to the CRM and flag campaigns where cost per opportunity rose”, and it writes and runs the code. CXL marketers have used it to build a marketing second brain and automated ICP research and ad reporting.

Trade-off: you still need to understand what you are building well enough to test it. Code you can’t check is a liability.

Pricing: included in Claude Pro ($20/month) and higher plans.

9. Lovable

Best for: building landing pages, ROI calculators, quizzes and small web apps from a prompt.

Lovable turns a description into a working web app you can edit by chatting, which lets a marketer ship an interactive lead magnet without a front-end developer.

Trade-off: great for prototypes and standalone tools; less suited to pages that must live inside your main CMS and design system.

Pricing: Free plan. Pro $25/month with 100 monthly credits.

Best AI tools for B2B go-to-market data

10. Clay

Best for: enriching account and contact lists from many data providers and running AI research on each row.

Clay pulls data from dozens of providers into one table and lets you run AI on each row, for example “read this company’s careers page and tell me if they are hiring for RevOps.” It has become a core tool for ABM and outbound teams. See how to apply this to AI fit scoring from CRM data.

Trade-off: powerful but complex, and credit costs add up fast at volume. Overkill if you only need basic enrichment.

Pricing: Free plan. Launch from $167/month billed annually.

11. HubSpot

Best for: teams that want AI agents working directly on their CRM and marketing data.

HubSpot’s Marketing Hub now includes an agent hub and agent builder from the Starter plan, and Professional adds AEO tools and a nurture agent, both in beta. The advantage over a general assistant is context: the AI works on your contacts, deals and campaign history.

Trade-off: the most useful AI features sit in Professional, which is a big step up in price, and AI usage draws on HubSpot Credits.

Pricing: free tools. Marketing Hub Starter $20/month; Professional $890/month including three core seats.

12. Gong

Best for: mining recorded sales calls for objections, competitor mentions and the exact words buyers use.

Gong records and analyzes sales conversations. For marketers, its value is research: hundreds of hours of buyers describing their problems in their own words. That language is the raw material for positioning and copy, and it beats anything an AI writes from scratch. For the method, see how to write better than AI with voice-of-customer data.

Trade-off: bought by sales, not marketing, and priced for sales teams. Ask for access before you ask for budget.

Pricing: not published; quote only.

Best AI tool for SEO and AI search visibility

13. Ahrefs

Best for: SEO research plus tracking how often your brand appears in AI answers.

Ahrefs remains a core SEO tool, and its Brand Radar product tracks brand mentions and citations across AI assistants and AI search. That makes it one of the few tools that measures both traditional rankings and AI visibility in one place.

Trade-off: expensive for a solo marketer, and AI visibility data is still young. Treat trends as more reliable than single numbers.

Pricing: Lite $129/month; Standard $249/month. Brand Radar from $199/month.

Best AI tools for creative and video

14. Canva

Best for: producing on-brand social posts, ad variations and presentations quickly.

Canva’s AI features generate and edit images, resize designs for every channel and apply brand kits, so a marketer can produce a full set of ad variations in minutes. Controlled evidence supports AI creative: in a live Instagram campaign published in Marketing Science, AI-generated ad images averaged 0.98% click-through against 0.65% for a professional designer’s images.

Trade-off: template-driven designs can look like everyone else’s. Your brand kit and creative direction still do the heavy lifting.

Pricing: Free plan. Pro $144/year for one person; Business $250/year per person.

15. Descript

Best for: turning webinars, podcasts and interviews into clips and edited video.

Descript lets you edit video by editing its transcript, removes filler words automatically, and finds clip-worthy moments. For B2B teams that run webinars, it turns one recording into a month of social content.

Trade-off: built for talking-head and screen-recorded content, not polished brand video.

Pricing: Free plan. Hobbyist $16/month billed annually; Creator $24/month billed annually.

Which AI marketing tools should you buy? Three stacks

Before buying anything, list the three recurring jobs that take most of your week. Then buy the cheapest tool that does one of them. A tool without a defined job becomes another subscription nobody opens.

When you don’t need more AI tools

• You haven’t used your assistant’s project or custom GPT features yet. Most marketers get more from adding context to one tool than from adding a second tool.
• Your process isn’t written down. An automation tool can’t automate a job nobody can describe. Document it first.
• The problem is your message, not your output. AI writing tools produce more copy faster. If buyers don’t understand why to choose you, that just means more unclear copy.
• You are buying an “AI copywriter” app. Most resell the same models as Claude and ChatGPT at a markup, with less control over context.

AI marketing tools FAQ

What is the best AI tool for marketing?

There is no single best tool. For most B2B marketers, the best starting point is one general assistant, such as Claude or ChatGPT, loaded with your ICP, positioning and examples. Add Perplexity for research and an automation tool such as n8n or Zapier once you have a repeatable job to automate.

What are the best free AI tools for marketing?

Claude, ChatGPT, Gemini and Perplexity all have free plans good enough to evaluate them. Zapier’s free plan includes 100 tasks a month, Make’s includes 1,000 credits, n8n’s Community edition is free to self-host, and Canva and Descript have free plans for creative work.

Which AI tools do employers want marketers to know?

In CXL’s analysis of US marketing job posts from January to May 2026, the most-mentioned AI tools were ChatGPT (5.5% of posts), Claude (4.6%) and Perplexity (2.2%). Claude saw the fastest growth, quadrupling from 1.1%.

Is ChatGPT or Claude better for marketing?

Both are strong. Claude is often preferred for long-form writing, following detailed style guides and building tools with Claude Code. ChatGPT has a broader set of built-in features such as image creation, voice and custom GPTs. Try both on your own real tasks for a week and keep the one whose output needs less editing.

How much do AI marketing tools cost?

General assistants cost about $20 per person per month. A practical stack of an assistant, a research tool and a design tool costs about $46 a month billed annually. Specialist tools such as Clay and Ahrefs start at well over $100 a month.

Can AI tools replace a marketing team?

No. Controlled studies show AI makes marketers faster and lifts weaker performers most, but the gains depend on people choosing the work, giving context and checking the output. See our guide to AI marketing for the research.

Learn to use these tools on real work

CXL is a B2B marketing training company focused on AI and conversion optimization. Tools are the easy part; knowing which job to give them and how to check the output is the skill. To build that skill, start with our guide to AI marketing, compare the best AI marketing courses, or join CXL’s Marketing Engineer cohort to spend eight live weeks building an AI system on your own marketing work.

Sources

• Vendor pricing pages for Claude, ChatGPT, Google AI plans, Perplexity, n8n, Zapier, Make, Lovable, Clay, HubSpot, Ahrefs, Canva and Descript, checked November 10, 2026.
• CXL. AI in marketing job descriptions. 2026.
• Daviet, R. and Nishimura, Y. AI-generated ads in a live campaign. Marketing Science, 2026.

A CRO audit is a structured review of a website’s analytics, technical health, user experience and messaging to find what stops visitors from converting, and to rank those problems by how much fixing them is worth. It is the research step that comes before A/B testing. Done well, it turns “our conversion rate is low” into a prioritized list of specific problems, each with the evidence behind it.

This guide covers the six-part audit process CXL teaches, a checklist you can copy, how to prioritize what you find and the mistakes that make audits useless.

Key takeaways

• An audit is research, not opinions. Every finding should point to data: an analytics drop-off, a recording, a survey answer or a usability test.
• Check tracking first. If analytics is wrong, every conclusion built on it is wrong too.
• Combine quantitative and qualitative data. Analytics shows where people leave; recordings, surveys and user tests show why.
• Prioritize by impact and evidence. A short list of well-evidenced problems on high-traffic pages beats a 200-item spreadsheet.
• Fix obvious bugs immediately. Only test changes where the right answer is genuinely uncertain.

What does a CRO audit cover?

CXL’s ResearchXL model splits conversion research into six areas. Each answers a different question:

For the method behind step 2, see how to do a heuristic analysis. For step 3, start with an analytics audit.

How to do a CRO audit, step by step

1. Define the conversion and the funnel. Write down the primary conversion (demo request, trial, purchase), the micro-conversions leading to it and the pages in between. An audit without a defined goal turns into a list of design preferences.
2. Verify tracking. Confirm every step of the funnel fires correctly, that conversions match your CRM or order system within a reasonable margin, and that internal and bot traffic are filtered out.
3. Find the leaks with analytics. Build the funnel report and segment it by device, browser, traffic source, new versus returning, and country. Look for the steps and segments that convert far below the rest.
4. Run the technical check. Compare conversion rates by browser and device version. Test the funnel on the worst performers. Check page speed on mobile for the highest-traffic templates.
5. Review key pages heuristically. Assess each page in the funnel for clarity of the value proposition, relevance to the traffic source, friction, distraction and anxiety. Note every issue with a screenshot.
6. Collect qualitative evidence. Watch at least 50 to 100 session recordings of visitors who dropped off, run an exit poll on the leakiest page, survey recent customers about what nearly stopped them, and run five or more user tests with your target audience.
7. Synthesize and prioritize. Group findings into problems, attach the evidence to each, and score them (see below). Fix bugs right away, and turn the uncertain, high-impact problems into test hypotheses.

CRO audit checklist template

Copy this into a spreadsheet. Add a column for status, evidence and owner.

How to prioritize CRO audit findings

Use a scoring model that rewards evidence and traffic, not enthusiasm. CXL’s PXL framework asks yes-or-no questions such as: is the change above the fold, is it noticeable within five seconds, is it supported by user testing, qualitative feedback or analytics, and how easy is it to build. Binary questions make scores harder to inflate than 1 to 10 ratings. See how PXL works.

Then sort findings into three buckets:

• Fix now: bugs, broken tracking and clear usability failures. Testing them wastes traffic.
• Test: high-impact changes where the right answer is uncertain, on pages with enough traffic. Check with the A/B test calculator.
• Research further: problems you can see but not yet explain.

Common CRO audit mistakes

• Auditing opinions instead of evidence. “The button should be green” is not a finding. “62% of mobile visitors never scroll to the button” is.
• Skipping tracking validation. Teams regularly discover that a funnel step was never tracked, after weeks of analysis.
• Treating every page equally. Start with the pages that carry the most traffic and revenue.
• Delivering a list nobody acts on. An audit is finished when its top findings are fixed or in a test, not when the report is sent.

When a CRO audit is not the right next step

• The site gets very little traffic. With a few hundred visitors a month, talk to customers and fix positioning first. Funnel analytics will not have enough data to show patterns.
• The offer is the problem. If buyers don’t want what you sell at that price, page changes won’t fix it. Customer research comes first.
• You are about to rebuild the site. Run the research parts of the audit to inform the rebuild, but skip page-level fixes on pages that are going away.

CRO audit FAQ

What is a CRO audit?

A CRO audit is a structured review of a website’s tracking, technical health, user experience and messaging to find what stops visitors from converting, backed by analytics, behavior data and customer research, and ranked by potential impact.

How long does a CRO audit take?

For a typical B2B or ecommerce site, two to four weeks: about a week for tracking and analytics, a week for heuristic review and recordings, and time for surveys and user tests to collect responses.

What is the difference between a CRO audit and a UX audit?

A UX audit focuses on usability. A CRO audit includes usability but also covers tracking, traffic-source fit, messaging, offer and prioritization by business impact, and it feeds directly into an A/B testing roadmap.

Can I do a CRO audit myself?

Yes, with free tools: GA4 for analytics, Microsoft Clarity for recordings and heatmaps, a free survey tool and a handful of user tests. The hard part is not the tools but judging the evidence without bias, which is a learnable skill.

Learn to run CRO research properly

CXL is a B2B marketing training company focused on AI and conversion optimization. The audit process above is the foundation of CXL’s CRO course and certification. For the full picture of the discipline, read the conversion rate optimization guide, and for the tools, see the best CRO tools.

Generative engine optimization (GEO) is the practice of making your content more likely to be cited, quoted and recommended when AI systems such as ChatGPT, Perplexity, Gemini, Claude and Google AI Mode generate an answer. SEO earns a ranking and a click. GEO earns a place inside the answer itself, often without a click.

The term comes from a 2023 research paper, GEO: Generative Engine Optimization by Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, accepted at KDD 2024. Most GEO advice online paraphrases a headline from it. This guide goes through what the paper actually measured, what independent data has shown since, and the process we teach at CXL for getting cited.

Key takeaways

• Evidence beats keywords. In the original GEO study, adding quotations lifted a source’s visibility in AI answers by about 41%, adding statistics by about 31% and citing sources by about 28%. Keyword stuffing made visibility 8% worse.
• GEO helps smaller sites most. A source ranked fifth gained 115% visibility from citing sources, while the top-ranked source lost 30%.
• SEO is still the base. Ahrefs found 37.1% of pages cited in Google AI Overviews also rank in the organic top 10 for the query, and Google’s own guidance says AI search runs on its normal ranking systems.
• Placement matters. In CXL’s analysis of 100 AI Overview citations, 55% of cited passages came from the first 30% of the page.
• Being cited pays, but less than a classic ranking. Seer Interactive’s data across 5.5 million queries shows cited brands get roughly twice the clicks of uncited brands on AI Overview searches, but still about 38% fewer than on searches with no AI Overview.

What is generative engine optimization?

GEO is optimization for engines that write answers instead of listing links. A generative engine retrieves several sources, then uses a large language model to summarize them into one response with citations. Your goal is to be one of the sources it uses, and to have your words, numbers and brand appear prominently in what it writes.

The original paper measured this with two metrics. Position-adjusted word count counts how many words in the answer come from your source, weighted toward the top of the answer. Subjective impression uses a model to rate how relevant, influential and prominent your citation is. Both are better proxies for GEO success than rankings, because the answer is the new results page.

GEO vs SEO vs AEO

The three overlap heavily. The difference is where you want to show up and how you measure it.

In practice you do all three on the same pages. Our guide to answer engine optimization covers AEO in depth, and our look at whether AEO and GEO are just SEO hype covers how the three relate.

What did the original GEO research find?

The researchers built GEO-bench, a set of 10,000 queries across domains, split into 8,000 for training and 1,000 each for validation and testing. They rewrote source pages using nine different methods and measured how much each source then appeared in the generated answers. Every method used the same content, only rewritten, so the differences come from presentation, not from new information.

Three findings matter most for marketers:

1. Evidence is the lever. The top three methods all add verifiable substance: quotes, numbers and citations. The paper reports the best methods improved visibility by 30% to 40% on the word count metric and 15% to 30% on the impression metric.
2. Keyword stuffing backfires. The classic SEO tactic performed worst, cutting visibility by about 8%. Generative engines reward content that helps them write a credible answer, not content that repeats the query.
3. Lower-ranked sources gain the most. For a source in fifth position, citing sources raised visibility by 115.1%, adding quotations by 99.7% and adding statistics by 97.9%. For the source in first position, citing sources reduced visibility by 30.3%. The authors suggest this gives smaller publishers a way to compete with sites that win on backlinks and domain strength.

The team also tested on Perplexity, a live engine, with a 200-query subset. Quotation addition improved the word count metric by about 21%, statistics addition improved the impression metric by about 37%, and keyword stuffing again made results about 9% worse.

What the GEO study does not prove

Treat these numbers as direction, not a forecast. The experiments ran in 2023 on a research setup and a 200-query live sample, and AI search has changed a lot since then. The paper also measured visibility inside the answer, not clicks, leads or revenue. And the authors found that the best method varies by domain, so test on your own queries before rewriting your whole site.

How do AI search engines choose which sources to cite?

Most AI search products retrieve pages from a search index, then generate an answer from them. Google’s AI Overviews and AI Mode use Google’s existing index and ranking systems, and run several related searches around your question before writing the answer, a technique Google calls query fan-out. Google’s own guidance, which we review in what Google’s AI search optimization guide gets right, is that AI search is still SEO.

Ranking for the exact query is not the whole story, though. In Ahrefs’ analysis of 4 million AI Overview citation URLs, 37.1% also ranked in the organic top 10, 26.2% ranked between positions 11 and 100, and 36.7% did not rank in the top 100 at all. Fan-out explains much of the gap: your page can be cited for a sub-question the user never typed.

Position on the page matters too. CXL mapped where the cited passage sat inside the source page for 100 Google AI Overview citations: 55% came from the first 30% of the page, 24% from the middle and 21% from the bottom 40%. Kevin Indig’s study of 18,012 verified ChatGPT citations found the same pattern, with 44.2% coming from the first 30% of the document.

Off-site reputation is the third input. In Ahrefs’ brand study, branded web mentions correlated at 0.664 with visibility in AI Overviews, one of the strongest signals they measured. That is a correlation, not proof of cause, but it matches what practitioners see: AI engines repeat what the wider web says about you.

Does GEO actually drive traffic?

Yes, but less than a classic ranking did. AI answers reduce clicks overall, and being cited claws back part of that loss rather than all of it.

Seer Interactive’s 2026 analysis covers 53 brands, 5.47 million queries and 2.43 billion organic impressions from January 2025 to February 2026. In February 2026, organic click-through rate was 2.36% on searches with an AI Overview and 3.82% on searches without one. Being cited made a large difference:

So on a query with an AI Overview, being cited roughly doubles your clicks compared with not being cited. Pew Research Center’s study of real browsing behavior points the same way: users clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when it did not.

The business case for GEO is therefore not traffic alone. It is presence in the place where buyers now form their shortlist. If a buyer asks ChatGPT for the best tools in your category and you are not named, you do not get the chance to win the click at all.

How to do generative engine optimization: 7 steps

1. Build a fixed prompt set. Write down 30 to 50 questions your buyers ask on the way to a purchase: category questions (“best B2B analytics tools”), problem questions (“how to reduce churn in SaaS”) and comparison questions (“X vs Y”). Use sales call notes and support tickets, not only keyword tools. Many of these questions show zero search volume and are still asked in chat every day.
2. Record a baseline. Run every prompt in ChatGPT, Perplexity, Gemini and Google AI Mode. Log whether you are mentioned, whether you are cited, which competitors appear and which third-party sites get cited. That last list is your map of where AI engines get their evidence.
3. Put the answer first. On each target page, answer the main question in the first two or three sentences, then support it. Make every H2 a real question with a direct answer underneath. Given that most citations come from the top third of a page, this is usually the fastest win.
4. Add evidence, not adjectives. This is where the GEO research is clearest. Replace vague claims with specific numbers, attribute them to named sources, and add short quotes from credible experts or your own customers. Original data you own, such as a survey, benchmark or analysis of your product data, is the strongest version, because nobody else can cite it first.
5. Earn mentions where AI engines look. Use the third-party sites from your baseline: review sites, Reddit and community threads, industry publications, comparison posts and podcasts. Real participation and customer reviews build the kind of consensus AI answers repeat. Manufactured mentions on low-quality sites do not, and Google explicitly warns against them.
6. Keep the technical door open. Make sure AI search crawlers can reach your pages. OpenAI uses OAI-SearchBot for ChatGPT search and GPTBot for model training, and you can allow one while blocking the other; see OpenAI’s crawler documentation. Google-Extended controls whether your content is used for Gemini training and grounding, and it does not affect whether you appear in Google Search or AI Overviews. Content hidden behind scripts, logins or tabs is easy for crawlers to miss.
7. Re-run the prompt set every week. Change one group of pages and leave a similar group alone. After four to six weeks, compare citation rates between the two groups. Treat each change as a hypothesis, the same way you would treat an A/B test.

Which GEO tactics don’t work?

A lot of GEO advice is untested. Three popular tactics have little evidence behind them:

• llms.txt files. SE Ranking analyzed nearly 300,000 domains and found no meaningful relationship between having an llms.txt file and how often ChatGPT cites a site. In OtterlyAI’s 90-day log study, only 84 of more than 62,100 AI crawler visits requested the file, about 0.1%. Google says it does not use llms.txt for Search.
• AI-specific schema. Ahrefs tracked 1,885 pages that added JSON-LD markup against a matched control group and found no citation lift in AI Overviews, AI Mode or ChatGPT. Structured data still helps rich results. It is not a citation switch.
• Chunking content for AI. Google says its systems handle long pages and you do not need to split content into artificial chunks. Clear sections with one job each help readers and retrieval alike, which is different from breaking pages apart.

We cover the evidence on each of these in detail in our review of Google’s AI search guidance.

How do you measure GEO?

Measure GEO at four levels, from closest to the answer to closest to revenue:

A spreadsheet of 30 prompts re-run weekly is enough to start. If you want to automate it, see how to build an AI visibility tracker with n8n and how to track AI visibility with Bing.

When GEO is not worth your time

• Your pages don’t rank for anything yet. AI search engines retrieve from search indexes. If your site has no organic visibility, fix your SEO basics first. GEO will not rescue a site search engines barely know.
• Your buyers don’t research with AI. Some local, impulse and low-consideration purchases rarely start in a chatbot. Check your own buyers before you invest, for example by asking new customers how they found you.
• You have nothing original to say. GEO rewards evidence. If your content repeats what the top ten results already say, AI engines have no reason to cite you over the sources you copied.
• You expect a tool to do it for you. Many GEO tools sell dashboards for a problem a spreadsheet and an hour a week can measure. Buy one when you know which prompts matter and need to track them at scale.

Generative engine optimization FAQ

What is GEO in marketing?

GEO, or generative engine optimization, is the work of getting your brand mentioned and your content cited in answers from AI tools like ChatGPT, Perplexity, Gemini and Google AI Mode. It focuses on the evidence on your pages and your reputation on other sites, because that is what AI engines draw on when they write an answer.

Is GEO the same as SEO?

No, but they overlap. SEO aims for rankings and clicks in search results. GEO aims for citations and mentions inside AI-generated answers. Most AI search engines retrieve from search indexes, so good SEO is the foundation GEO builds on.

Does GEO replace SEO?

No. Google’s AI features run on its normal ranking systems, and 37.1% of AI Overview citations come from pages that already rank in the top 10. GEO adds a layer on top of SEO: more evidence on the page, answers placed earlier and stronger presence on third-party sites.

What is the difference between GEO and AEO?

AEO, answer engine optimization, focuses on being the direct answer to a specific question, including featured snippets and voice assistants. GEO focuses on generative AI systems and covers your brand’s presence across the whole answer, not only one snippet. The tactics overlap heavily.

How long does GEO take to work?

On-page changes can show up in AI answers within weeks once pages are recrawled, especially for engines that search the web live, such as Perplexity and ChatGPT search. Changes that depend on off-site reputation, like reviews and mentions, take months. Re-run the same prompts weekly so you can see which changes moved what.

Does llms.txt help GEO?

There is no good evidence that it does. Studies of hundreds of thousands of domains found no link between having an llms.txt file and being cited, and Google says it does not use the file for Search.

Learn GEO by doing it

CXL is a B2B marketing training company focused on AI and conversion optimization. If you want to apply this to your own site, CXL’s Optimize Pages for AI Search with GEO/AEO course, taught by Steve Toth, walks through auditing your AI visibility and building the comparison content buyers use in AI research. Compare it with other options in our list of the best AI SEO and GEO courses.

Sources

• Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K. and Deshpande, A. GEO: Generative Engine Optimization. KDD 2024. arXiv:2311.09735.
• Seer Interactive. AIO Impact on Google CTR: 2026 Update. April 2026.
• Ahrefs. AI Overview citations and the organic top 10. 2026.
• Pew Research Center. Google users are less likely to click on links when an AI summary appears in the results. July 2025.
• CXL. Where Google AI Overviews take their citations from. 2026.
• CXL. What Google’s AI search optimization guide gets right and what’s missing. 2026.

AI marketing is the use of artificial intelligence, today mostly large language models and AI agents, to research, create, personalize, analyze and automate marketing work. In 2026 nearly every marketing team uses it in some form. Very few use it well: in CXL’s 2026 benchmark of marketers’ AI skills, only 9% worked at an AI-native level, where AI runs whole workflows instead of helping with single tasks.

This guide is for B2B marketers who want more than a list of tools. It covers what controlled studies show AI actually improves, where it makes work worse, how B2B teams use it today, and a practical way to move from prompting a chatbot to running AI systems that do real marketing work.

Key takeaways

• Adoption is near universal, skill is not. 75% of marketers have adopted AI (Salesforce, 4,450 respondents), but 57% of marketers CXL assessed were still only AI-assisted and 9% were AI-native.
• The productivity gains are real and measured. In a study published in Science, professionals including marketers finished writing tasks 40% faster with 18% higher quality using ChatGPT.
• AI also makes some work worse. In a Harvard and BCG experiment with 758 consultants, AI users were 19 percentage points less likely to get the right answer on a task outside what AI handles well.
• Budgets are ahead of readiness. CMOs put 15.3% of marketing budgets into AI in 2026, yet only 30% say their organization is ready to scale it (Gartner, 401 CMOs).
• The job market has moved. The share of US marketing job posts that require AI skills rose from 15% to 40% between Q1 and Q3 2026, in CXL’s analysis of 878 postings.

What is AI marketing?

AI marketing covers three generations of technology that now sit side by side in most marketing stacks:

When people say “AI marketing” in 2026 they usually mean the last two. Predictive AI has been inside ad platforms and CRMs for a decade. What changed is that generative and agentic AI put the capability in the hands of individual marketers, without a data science team.

How are B2B marketers using AI in 2026?

Mostly for content, and mostly one task at a time. In CXL’s 2026 survey of B2B marketers:

The gap between those two middle rows is the story of AI marketing right now. Marketers are confident writing with AI and inexperienced at building with it. CXL’s AI maturity benchmark shows the same split by skill area: 12% of marketers reached AI-native level in production work, but only 2% in marketing operations and about 2% in analytics.

Larger surveys agree on the adoption side. Salesforce’s tenth State of Marketing report, based on 4,450 responses collected in late 2025, found 75% of marketers have adopted AI, yet 84% admit they still run generic campaigns, and 98% hit barriers to personalization, most often siloed or poor-quality data. AI has not fixed the data problem underneath it.

Does AI actually make marketing work better?

For a well-defined set of tasks, yes, and the effect is large. For tasks outside that set, it can make work worse while feeling more productive. The best evidence comes from randomized experiments, not vendor surveys:

Three patterns repeat across these studies:

1. AI raises the floor more than the ceiling. The weakest performers gained most: 43% in the BCG study and 34% for novice support agents. For experienced marketers the gain comes less from better first drafts and more from doing work they never had time for.
2. The frontier is jagged. The BCG researchers’ term describes how AI is excellent at some tasks and confidently wrong at others that look similar. Knowing where that line sits for your own work is the core AI marketing skill.
3. Quality control decides the outcome. In the ad studies, the AI output that won was vetted, selected and tested by people. Unreviewed output at volume is how teams end up with the generic campaigns Salesforce’s respondents admit to.

AI marketing use cases by function

These are the B2B use cases where AI does reliable, measurable work today. The right-hand column links to a CXL guide that shows how to build each one.

From AI-assisted to AI-native: the three levels

CXL’s benchmark sorts marketers into three levels. Most teams sit at the first and think they are at the second.

Moving up a level is less about tools and more about process. You cannot automate a workflow nobody has written down. That is why the marketers who get furthest usually start by documenting how a job is done today.

How to build your first AI marketing workflow

1. Pick one recurring job that eats your week. Good candidates take hours, happen often and have a clear output: a weekly competitor digest, a monthly paid media report, research briefs before sales calls.
2. Write down how a good version is done today. Inputs, steps, decisions and what “good” looks like. This document becomes the instructions for the AI.
3. Give the AI context, not just a prompt. Add your ICP, positioning, brand voice guide and two or three examples of excellent past output. Most poor AI output comes from missing context, not bad models.
4. Build the smallest version that works. A saved prompt with context files often beats an elaborate agent. Add automation with tools like n8n or Claude Code only once the manual version produces output you would ship.
5. Define a quality check. Write a short rubric and review every output against it until you trust the system. Keep a person responsible for anything customer-facing.
6. Measure time saved and outcome. Track hours before and after, plus the business metric the job serves. If the output is faster but worse, you have not improved anything.

For worked examples, see how CXL built a marketing second brain with Obsidian and Claude Code and what we learned from a week of building marketing automations in n8n.

Which AI marketing tools do teams use?

Most B2B teams combine a general-purpose assistant with a few specialized tools. For prices, trade-offs and three ready-made stacks, see our list of the best AI marketing tools.

• General assistants: ChatGPT, Claude and Gemini for writing, analysis and research. Claude mentions in marketing job posts rose from 1.1% to 4.6% between January and May 2026 in CXL’s job post analysis, faster than any other assistant.
• Research and search: Perplexity for sourced web research.
• Automation and agents: n8n, Zapier and Make for workflows; Claude Code for building custom tools without a developer.
• AI inside your existing platforms: the AI features in your CRM, marketing automation and ad platforms, which work on data you already have.

Pick tools after you have picked the workflow. A tool without a defined job becomes another subscription nobody uses.

What are the risks and limits of AI marketing?

• Confident errors. Language models invent facts, sources and numbers. Anything published needs a source check. The BCG study shows experienced professionals trusting wrong AI answers on tasks that looked routine.
• Sameness. When every competitor prompts the same models with the same inputs, the output converges. Your own data, customer language and point of view are what keep AI-assisted work distinct.
• Pilots that never pay off. MIT Project NANDA’s 2025 report, based on 300+ public AI initiatives, interviews with 52 organizations and a survey of 153 leaders, found that 95% of organizations saw no measurable profit and loss impact from generative AI pilots. The sample is small and the finding is debated, but it matches other data. McKinsey’s 2026 State of AI survey of 1,719 leaders found nearly nine in ten organizations use AI regularly, yet only 37% attribute any EBIT impact to it, and Gartner found 70% of CMOs say their processes are not mature enough to scale AI.
• Data and privacy. Customer data pasted into consumer AI tools can breach your contracts or privacy rules. Use business plans with data controls and agree a policy on what can be shared.
• Skill erosion. If juniors only ever edit AI drafts, they may never learn to write a strong brief or spot a weak argument. Build in work that keeps those skills sharp.

When AI won’t fix your marketing

• Your positioning is unclear. AI will produce more content faster, all of it saying the same unclear thing. Fix the message first.
• Your data is a mess. AI analysis on broken tracking produces confident, wrong conclusions faster. Salesforce’s respondents named poor data as the main barrier to personalization for a reason.
• The work needs a relationship. Partnerships, enterprise deals and community building run on trust between people. AI can prepare you. It can’t show up for you.
• You are automating a bad process. If a report nobody reads takes six hours, the fix is to stop producing it, not to have an agent produce it in six minutes.

How AI is changing marketing jobs and skills

Employers now expect AI skills, and increasingly they expect proof. In CXL’s analysis of 878 US marketing job descriptions, the share explicitly requiring AI skills rose from 15% in Q1 2026 to 40% in Q3, and 54% of growth marketing roles required them. A new title has appeared for marketers who build AI systems: see what a marketing engineer is and why companies are hiring them. For the skills involved, read the 5 skills marketers need to become AI-native.

AI marketing FAQ

What is AI marketing?

AI marketing is the use of artificial intelligence to research, create, personalize, analyze and automate marketing work. It includes predictive models built into ad platforms and CRMs, generative AI tools like ChatGPT and Claude, and AI agents that run multi-step marketing workflows.

How is AI used in marketing?

The most common uses are content production, research synthesis, ad creative, personalization, reporting and competitive monitoring. In CXL’s 2026 survey, 75% of B2B marketers used AI for content production and 42% for research synthesis.

Will AI replace marketers?

The evidence points to AI changing marketing jobs rather than removing them. Controlled studies show AI makes marketers faster and lifts weaker performers most, while still needing people to set strategy, judge quality and catch errors. Employers are adding AI requirements to roles, not cutting the roles: 40% of US marketing job posts required AI skills in Q3 2026.

What is the best AI tool for marketing?

There is no single best tool. Most B2B teams use a general assistant such as ChatGPT or Claude for writing and analysis, plus an automation tool such as n8n or Zapier for workflows. Choose based on the job you want done, and check what your existing platforms already include. We compare 15 options in our AI marketing tools guide.

How do I start using AI in marketing?

Pick one recurring task that takes hours each week, document how it is done well, give an AI assistant that context plus examples, and review the output against a simple quality rubric. Automate it only once the manual version works.

Is AI marketing worth the investment?

For defined tasks the productivity gains are well documented: 40% faster writing with higher quality in the Science study. Organization-wide returns are harder; most companies have not yet seen profit impact. The difference usually comes from redesigning workflows rather than handing people a chatbot.

Learn AI marketing by building it

CXL is a B2B marketing training company focused on AI and conversion optimization. The fastest way to move from AI-assisted to AI-native is to build a working system on your own marketing work, with feedback. CXL’s Marketing Engineer cohort is eight live weeks of doing exactly that, and you are certified on what you build. To compare options, see our list of the best AI marketing courses.

Sources

• Noy, S. and Zhang, W. Experimental evidence on the productivity effects of generative artificial intelligence. Science, July 2023.
• Dell’Acqua, F., McFowland III, E., Mollick, E. et al. Navigating the Jagged Technological Frontier. Harvard Business School Working Paper 24-013, 2023.
• Brynjolfsson, E., Li, D. and Raymond, L. Generative AI at Work. The Quarterly Journal of Economics 140(2), 2025.
• Ju, H. and Aral, S. Collaborating with AI Agents: Field Experiments on Teamwork, Productivity, and Performance. 2025.
• Daviet, R. and Nishimura, Y. AI-generated ads in a live campaign. Marketing Science, 2026.
• Salesforce. State of Marketing, Tenth Edition. February 2026.
• Gartner. 2026 CMO Spend Survey. May 2026.
• McKinsey. The state of AI. 2026.
• MIT Project NANDA. The GenAI Divide: State of AI in Business 2025. July 2025.
• CXL. How teams use AI in B2B marketing in 2026, Marketer AI maturity benchmark 2026 and AI in marketing job descriptions.