AI is changing digital marketing by running four capabilities at production scale: content generation, audience analysis, decision automation and visibility inside AI answer engines. As far back as 2023, Salesforce found that 51 percent of more than 1,000 marketers surveyed were using or experimenting with generative AI, while only 17 percent had implemented it in their workflows, per Demand Gen Report. The gap between trying AI and operating it is where most teams still work.
By EPR Editorial Team. Originally published January 2026. Facts rechecked October 14, 2026. This page is the canonical framework in EPR's AI marketing cluster, and the case studies and personalization guardrails sit in AI Marketing Done Right.
What four AI capabilities do marketing teams run?
Marketing teams run four AI capabilities, and the difference between using them well and badly shows up in results, not only in speed. The brands that gain the most build AI into how the team works, while the brands that gain little attach a copywriting tool to an unchanged workflow.
Content generation
Content generation covers drafting, editing, localization and variant production with frontier models such as Claude, GPT and Gemini. The economics changed because one editor can now review a volume of drafts that once needed a team to write. Tools such as Jasper, Copy.ai and Writer add workflow features on top of the models.
Audience analysis
Audience analysis uses models to read CRM data, behavioral data and unstructured text such as support tickets, social posts and reviews, then produce segments and propensity scores. CRM and marketing platforms now ship these features, so a team can act on a working audience model without waiting for a separate data-science cycle.
Decision automation
Decision automation moves bid strategy, creative rotation, audience expansion and budget allocation into platform algorithms such as Meta Advantage+, Google Performance Max and TikTok Smart+. The marketing team's role shifts from execution to constraint-setting: define the objective, the creative set, the budget and the guardrails, then monitor the result.
AI engine visibility
AI engine visibility is the work of being named in the answers that ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews give when buyers ask category questions. This is Generative Engine Optimization, or GEO, and it is the newest of the four. The communications-side version is covered in where big PR actually uses AI, and the six-stage PR workflow is in The AI PR Stack.
Which ten use cases is AI doing in marketing?
The four capabilities appear in marketing teams as ten concrete use cases, each with a reference example and a caution.
1. Personalization
AI analyzes behavior and preferences to deliver tailored content, email and ads. Amazon's recommendation engine is the standard reference example. The case studies, including how to avoid personalization fatigue, are in AI Marketing Done Right.
2. Data analysis and insights
AI processes large datasets and surfaces patterns that people miss, so strategy can adjust closer to real time. Google Analytics' AI-generated insights on behavior and campaign effectiveness are a reference example.
3. Customer service and support
AI chatbots handle round-the-clock questions, guide users through processes and escalate complex cases to people. This works best for high-volume repetitive questions and fails when users cannot reach a person, a risk the Klarna case below shows.
4. Content creation and curation
AI generates product descriptions, summaries, social posts and copy variations at scale, using tools such as Copy.ai, Jasper, Claude, ChatGPT and Gemini. EPR's view is that generated content published without editorial judgment adds little that other pages lack, so it is less likely to be retrieved or cited. Article 50 of the EU AI Act sets disclosure duties for some AI-generated content, so EU-facing teams should confirm the current application date.
5. Advertising and targeting
Programmatic buying automates ad placement in real time, and Google Ads and Meta Ads run their targeting on AI systems. Privacy rules and platform changes limit which behavioral data these models can use, so teams should expect targeting accuracy to shift.
6. Search optimization and AI engine optimization
AI tools analyze search trends and competitor pages to improve keywords, headings and content structure, and Clearscope, Surfer SEO, Semrush and Ahrefs all include AI features. Classic SEO targets a position on a results page, while GEO targets a mention inside a generated answer.
7. Social media management
AI schedules posts around likely engagement windows, and listening tools monitor mentions of brands and topics. Hootsuite, Sprout Social and Brandwatch are reference tools, and the AI features are now standard, not a differentiator.
8. Email marketing
AI segments lists by behavior, personalizes content and predicts send times for each recipient. Mailchimp, HubSpot, Klaviyo and Iterable are reference platforms. Email is the use case where results are easiest to measure, because teams can test AI-optimized sends against a control group.
9. Customer journey mapping
AI tracks touchpoints across channels, and churn models flag at-risk customers so retention programs can act early. Salesforce Einstein, HubSpot's AI features and Adobe Sensei are reference tools. The most useful output is the pattern that separates customers who stay from those who leave.
10. Dynamic pricing
AI adjusts prices in real time based on demand, competition and customer behavior. Amazon, Uber, airlines and hotel revenue systems are the reference uses, and this is the hardest category to copy because the pricing models behind large revenue lines are closely guarded.
The layer above the ten
Every use case above improves an existing digital channel, and none addresses where discovery is moving: buyers increasingly ask AI engines instead of searching. EPR measures this with Citation Share, the percentage of category prompts in which a brand appears across five engines. Teams that run both workstreams, AI tooling across the ten use cases and Citation Share building through original research and earned coverage, protect their discovery position while they cut costs.
Most marketing teams run a small set of vendors for each workflow, and the table below groups common tools by the job they do. It is EPR's grouping of common choices, not a ranking, and EPR has not verified each tool's features.
| Workflow | Common tools |
|---|
| Segmentation and lead scoring | HubSpot, Salesforce Marketing Cloud, Adobe Experience Cloud, Marketo |
| Content generation | Jasper, Copy.ai, Writer, plus the frontier models |
| Email and SMS | Klaviyo, Attentive, Mailchimp, Iterable |
| Support chat and virtual assistants | Intercom, Zendesk |
| Sentiment and social listening | Brandwatch, Sprout Social, Talkwalker |
| Reporting | Looker, Power BI, Tableau |
Two constraints apply to every tool. Data privacy rules such as GDPR and CCPA limit which behavioral data a team can feed into these systems, and implementation cost falls hardest on smaller teams. The practical sequence is to automate repetitive tasks first, such as data entry, lead scoring and report generation, and keep people on creative direction and positioning.
What did Klarna's AI rollout show marketing teams?
Klarna's rollout showed that AI can replace large amounts of routine work and that cutting too far damages quality. In February 2024 Klarna said its OpenAI-powered assistant was doing the work of 700 full-time agents and handling two-thirds of customer chats, and it projected a $40 million profit improvement for the year, according to Fortune.
CEO Sebastian Siemiatkowski told CBS News that Klarna does not employ customer service agents itself but uses outside providers, so the 700 figure describes avoided capacity. The company also reported a 25 percent drop in repeat inquiries. On marketing, CB Insights reported that Klarna cut marketing agency spend by 25 percent and image production costs by $6 million using generative tools.
In May 2025 Siemiatkowski told Bloomberg that cost had been too dominant a factor and that the result was lower quality, and Klarna began rebuilding a human support option, per Fortune. EPR's read is that AI substitution works fastest on internal workflows and slowest on surfaces where customers decide whether to trust the brand. The matching guardrail in PR, a named person who owns each decision, is laid out in The AI PR Stack.
How do Anthropic, OpenAI and Perplexity differ for marketers?
The three companies sell to marketers in different ways, and the descriptions below are EPR's characterization, not vendor claims. Anthropic describes its approach to model behavior as Constitutional AI and emphasizes safety, which matters to brands with regulatory or reputational exposure. OpenAI emphasizes breadth of capability and ecosystem reach, which makes its models a common default for general-purpose work.
Perplexity is positioned as an answer engine more than a model provider, so for marketers its main value is being cited in answers when buyers run category queries. That citation view is now standard inside GEO programs.
How should marketing teams reorganize around AI?
Teams that absorb AI well share four operating choices, and each one prevents a specific failure.
- A named owner. One senior person owns the integration of the four capabilities into team workflow. The communications-side role is described in The AI Communications Lead.
- An evaluation cadence. A quarterly review shows which workflows AI improves, which it degrades and where the team over-relies on outputs.
- A content policy. Written rules cover disclosure, fact-checking and human review for customer-facing assets.
- A measurement framework. The team tracks productivity gains, quality outcomes and brand outcomes together.
Skipping these steps has cost publishers visibly. In 2023, Sports Illustrated removed articles after Futurism reported bylines tied to AI-generated author profiles, and Sports Illustrated said the articles came from a third-party vendor, per an Associated Press report. The same report notes that CNET posted 77 machine-generated stories, several of which needed corrections. AI is leverage, and leverage without operating discipline magnifies whatever the team was already doing, including the mistakes.
Where is AI marketing heading in 2026 and 2027?
Three developments look likely, and all three are EPR's judgment, not forecasts from a named source. First, agentic marketing, where AI systems plan, run and measure campaigns inside human-set limits, is moving from pilots to production at large enterprises, with products such as Salesforce Agentforce and Microsoft Copilot as early examples. Second, multimodal generation of video, audio and interactive media narrows the cost gap between large creative teams and small ones. Third, brands that built a record of third-party coverage early keep a retrieval advantage, because AI engines can only cite what already exists in published sources.
