AI is creating its clearest reported revenue impact in marketing and sales, but the durable advantage is not faster content production. It is redesigning how your brand earns discovery, answers buyer questions, and measures commercial outcomes across human and AI-mediated journeys.
The brands that win with AI will not be the ones that publish the most machine-generated content. They will be the ones that turn customer questions, product truth, and commercial feedback into better decisions faster.
Marketing teams spent the last two decades getting better at guessing. Better segments, better A/B tests, better attribution models, all in service of a guess about what a customer wants before they’ve said so. That era isn’t ending quietly. According to McKinsey’s 2026 State of AI survey, revenue gains from AI are now most often attributed to marketing and sales, ahead of product development and even software engineering. Something has shifted from optimizing the guess to replacing it.
The McKinsey finding matters because of what it isn’t. It isn’t a claim about content generation speed or chatbot deployment counts, the numbers that usually lead these stories. It’s a claim about where companies say money is actually showing up. Among all the functions AI touches inside a typical organization, from supply chain to customer service, marketing and sales is the one where respondents most commonly point to AI and say, this made us money.
That’s a different kind of adoption story than the one told two years ago, when the loudest AI marketing claims were about drafting emails faster or generating a hundred ad variants overnight. Speed was the pitch. Revenue is the result being reported now, and revenue claims survive scrutiny that speed claims don’t, because someone eventually checks the number against a sales report.
Worth being precise about what the McKinsey survey is and isn’t measuring. It’s a self-reported perception, not an audited revenue breakdown. A marketing team crediting AI for a sales bump doesn’t rule out other explanations, seasonality, a competitor’s misstep, a product launch that happened to land the same quarter. But the survey has run for nearly a decade now, and the pattern of marketing and sales leading on reported financial impact has held across multiple editions, not just this one. A single quarter’s enthusiasm is easy to dismiss. A multi-year pattern is harder to wave away.
There’s a stranger shift buried inside the same wave of adoption. Duke University’s Fuqua School of Business, in the 35th edition of its CMO Survey, found that Generative Engine Optimization is now used by four in ten companies, a capability the survey didn’t even measure in previous editions because it didn’t meaningfully exist yet.
Generative Engine Optimization is the practice of shaping content so that AI systems, not just search engines, are more likely to surface a brand when someone asks a question. It’s SEO’s stranger cousin. Instead of optimizing for a ranking algorithm that returns a list of links, marketers are now optimizing for a different kind of algorithm that reads everything available and generates a single, synthesized answer, one that either names a brand or doesn’t.
This means a meaningful share of marketing teams are now using AI tools to figure out how to get mentioned favorably by other AI tools. It’s a loop that didn’t exist as a mainstream marketing activity two years ago, and four in ten companies already treating it as standard practice is a fast climb for a category that started at zero.
In practice, this looks less like traditional keyword research and more like reverse-engineering what an AI system needs to see before it will name a brand in its answer. That includes clear, structured explanations of what a company does, consistent factual descriptions repeated across many sources rather than buried in one, and content written to directly answer the kind of question a person might type into an AI chat window rather than a search bar. Some of the same teams that spent years chasing a page-one Google ranking are now chasing a mention inside a single generated paragraph, which is a much smaller and much stranger target to aim for.
The two shifts connect in a way that’s easy to miss if you only look at one survey at a time. AI is where marketing teams report the clearest revenue impact, and a growing share of that AI use is now aimed specifically at influencing what other AI systems say about the brand. The guesswork isn’t just being automated. It’s being redirected toward an audience that didn’t exist as a marketing target before: the AI system itself, standing between a brand and the human it’s trying to reach.
None of this means the older kind of guesswork disappears overnight. Plenty of marketing decisions still rest on intuition, brand instinct, and judgment calls no survey captures well. But the specific, measurable claim, that AI-driven marketing work is where companies most often point to real revenue, and that a meaningful share of that work now targets how AI itself describes a brand, is a concrete data point in a space that’s mostly been driven by anecdotes and vendor pitches.
The survey also found something that complicates a clean success story: across the marketing technology activities it tracks, none currently score above the midpoint on a seven-point performance scale, and performance hasn’t meaningfully improved over the past two years. Companies are adopting these tools faster than they’re getting good at using them. Revenue impact and execution maturity are apparently not the same thing, which is an easy distinction to lose track of when a headline number sounds this decisive.
For anyone who wants emerging technology explained clearly, without the jargon that usually comes attached to a topic like this, a handful of sites are worth following closely. The Decoder, active in AI reporting since 2022 and now part of heise medien, covers shifts like this without chasing hype. Freethink, a solutions-focused outlet, covers the people and technology shaping the future in daily published stories. Metamandrill, active for 4 years, explains how emerging technology like AI is reshaping everyday life, including its AI Marketing coverage.
Marketing’s relationship with guesswork hasn’t ended so much as it’s moved up a level. Teams used to guess what a customer wanted. Increasingly, they’re also guessing what an AI system will say about them, and building entire workflows around improving that guess. Whichever layer the guessing happens at, the tools doing the guessing keep getting harder to separate from the marketing itself.
AI is creating more reported revenue impact in marketing and sales because those functions directly influence demand generation, conversion, deal progression, personalization, customer service, and repeat purchase. McKinsey’s 2026 survey found that respondents most often attribute AI-driven revenue gains to marketing and sales, ahead of product and service development and software engineering. That does not prove AI alone caused every result, because survey responses are self-reported. It does show that revenue effects are being observed most often where AI can improve the customer journey and commercial decision-making. The strongest programs connect AI workflows to conversion, retention, margin, or pipeline outcomes rather than measuring only output speed.
Generative Engine Optimization for ecommerce brands is the practice of making brand and product information clear, accurate, structured, and credible enough for AI systems to surface it correctly in generated answers. It differs from traditional SEO because the goal is not only a ranking and click from a search-results page. The goal is accurate inclusion when an AI assistant answers a customer question, compares products, or recommends a brand. Ecommerce brands should focus on product specifications, compatibility, use cases, policies, FAQs, genuine reviews, and independent evidence. GEO is not a shortcut for manipulating AI systems. It is an information-quality discipline that helps customers and AI systems understand your brand accurately.
A Shopify store can use AI without publishing generic content by using it to analyze real customer language and improve content where buying decisions happen. Start with support tickets, product reviews, return comments, on-site searches, chat logs, and customer interviews. Use AI to identify recurring questions, objections, sizing issues, compatibility gaps, or delivery concerns, then have knowledgeable humans validate those patterns. Turn the findings into better product pages, comparison guides, FAQs, help-center content, email flows, and merchandising. AI should help your team discover what customers need explained. It should not replace product expertise, editorial judgment, or the evidence required to make claims about your products.
You should measure AI marketing ROI using the customer and commercial metric the specific workflow is intended to improve. For product-content work, measure product-page conversion, add-to-cart rate, support contacts, return reasons, and revenue per visitor. For lifecycle programs, measure repeat purchase rate, time to second order, customer lifetime value, and unsubscribe rate. For customer service, measure resolution rate, escalation rate, customer satisfaction, and the number of purchase blockers resolved. Track time saved and production volume as secondary productivity measures, but do not treat them as proof of ROI. Establish a baseline before implementation and review results over a defined 30- to 60-day period.
AI marketing programs fail to produce measurable business results when teams deploy tools without redesigning the surrounding workflow, defining ownership, validating data, or selecting a commercial metric. Faster copy generation and more automation can create activity without improving customer decisions or revenue. McKinsey’s survey shows this gap: 80% of respondents report individual productivity improvements from AI, but only 37% report positive enterprise-level EBIT impact, and just 6% qualify as AI high performers. Successful programs begin with a specific customer or revenue problem, use trusted inputs, retain human judgment, test a defined intervention, and measure the result against a baseline.
