According to Klaviyo’s 2025 State of B2C Marketing Report, 47% of marketers invested in AI as a priority last year.
But just 31% invested in customer data platforms (CDPs), which suggests those same marketers aren’t getting the most out of their AI efforts. This is because AI is only as effective as the customer data it uses as inputs.
Without a unified data infrastructure, AI falls short of reaching its full potential. It can’t personalize its outputs, which means those outputs can’t resonate with audiences.
As many brands experiment with ChatGPT for marketing, they may find they’re getting competent drafts that remain basically generic because they don’t reflect who their customers are or how they behave.
In this guide, we describe how to best use ChatGPT for marketing and where it can save time. Then, we expand on how to personalize ChatGPT outputs with real customer data, so you can get the most out of your efficiency gains with AI.
What is ChatGPT for marketing?
ChatGPT for marketing is the process of using OpenAI’s LLM as an assistant during a marketing workflow. Marketers use ChatGPT for a variety of marketing tasks:
- Performance analysis: ChatGPT can serve as a friendly interface for marketing analytics. When marketers have easy access to performance data through conversational AI, they can “talk through” their data and get strategy ideas much faster.
- Strategy: ChatGPT can suggest strategy pivots and new approaches to acquire and retain customers. When those strategies are based on real performance data, suggestions become relevant and immediately actionable.
- Ideation: ChatGPT can help marketers connect strategy to content through brainstorming and suggesting relevant offers. When the AI ideation process is based on real data, suggested offers and tactics are based in reality rather than theoretical.
- Drafting: ChatGPT can come up with a first draft that marketers can use as a basis for personalization, especially when it has access to real-time customer data.
Marketers can use ChatGPT as a standalone tool for marketing tasks. But they can also use it with an MCP server, which empowers brands to connect AI interfaces like ChatGPT and Claude to the systems they already use in their tech stacks.
MCP servers are particularly useful for getting performance data through ChatGPT conversations, which can guide how you use AI to assist you with other parts of the marketing process.
How ChatGPT helps marketers save time while preserving quality
According to Bain's 2025 Generative AI in Marketing report, leading marketers have reduced time to market by 50% and reduced content creation time by 30–50% with AI.
But marketers don’t save time by completely removing themselves from the strategy and creation process. Most marketers use ChatGPT on a “human in the loop” basis, meaning humans are involved in reviewing and editing marketing content before it goes live.
Here are 4 of the best ways to save time with ChatGPT while preserving quality:
- Generate more subject line variations than you normally would to test. You can quickly scan the results to remove any outliers that don’t fit your brand voice, but AI can test for more volume in less time than you’d be able to without it.
- Summarize longer market research reports, survey responses, or reviews into themes you can scan in minutes instead of hours. AI can help you digest more content about your industry and audience in less time, so you can make decisions faster.
- Get plain-language insights and recommended next steps on strategy. When your customer data infrastructure is connected to your ChatGPT interface, you won’t need to manually check strategy recommendations against your actual reporting. Instead, everything happens in the same place and you can act right away.
- Repurpose one long piece of content into many shorter pieces, so you can promote special offers across more marketing channels faster. This is one of the main ways marketers can reduce time to market.
How to write ChatGPT marketing prompts that produce results you can use
ChatGPT can produce somewhat decent results with short prompts, but it performs a lot better when you can anchor the prompt with a few examples.
This is the difference between zero-shot vs. few-shot learning: asking ChatGPT to complete a task without examples vs. giving it a few examples to lean on when generating outputs. The latter is more likely to lead to a result that fits your brand, audience, and offer.
Here are a few components you can add to your ChatGPT prompts to improve results:
- Hook: Specify how you want a piece of content to pique a shopper’s interest.
- Goal: This will depend on the offer and the product, but you may want to describe either the value to the customer or the feeling you want them to walk away with.
- Lifecycle stage: Describe where the customer is in their lifecycle and the brand interactions they’ve likely already had. If you’re using AI that’s already pulling from customer data, you may just need to describe the audience segment.
- Brand voice: Provide example phrases that align with your preferred brand voice. If you’re using AI that’s trained on your brand materials, you may be able to skip this step.
- Do-not list: Tell ChatGPT what you don’t want it to do, including banned words and phrases the model isn't allowed to use.
- Real customer data: The best AI for marketing pulls from real customer data, so it can personalize outputs without the need for drastic human editing. This is the most optimal few-shot learning approach to getting AI results you can use right away.
- Channel: Specify which channels you’d like content for, and give ChatGPT several examples of messages across each channel that have worked in the past.
How to personalize ChatGPT marketing content
When marketers connect ChatGPT to their customer data, it becomes instantly more useful.
Rather than relying on elaborate prompts assembled after checking against real data, marketers can get a sense of performance and draft content with more relevant prompts without constantly switching between systems.
Connecting ChatGPT to your existing marketing stack
If you’re using ChatGPT for marketing but want to connect it to your customer data, here’s a practical step-by-step process for doing so with Klaviyo:
- Open a ChatGPT interface that supports Apps.
- In the sidebar, browse available apps and select Klaviyo.
- When prompted, sign in to your Klaviyo account and approve the requested scopes.
- After authentication, the Klaviyo app appears in your ChatGPT sidebar.
Once connected, you can call up the Klaviyo app by typing "Klaviyo" or simply start asking questions such as:
- “In my Klaviyo account, how did my campaigns perform this week?”
- “Show me the top-performing flows from the last 7 days.”
- “Analyze the performance of my welcome flow.”
Standalone ChatGPT vs. ChatGPT integration vs. CRM-native AI marketing agent
Marketing activity | Standalone ChatGPT | ChatGPT integration | CRM-native AI marketing agent |
|---|---|---|---|
Customer data | No access | Limited | Full access to real-time data |
Personalization depth | Limited to prompting | Grounded in real account performance data; can draft campaigns/templates from live data, but cannot build or edit segments in-account | Trained on real segments, product catalog, and performance history |
Sending | No sending capabilities; must send in a separate platform | Human review and sending, using deep links from the app | Drafts launch-ready email and text messages in-platform |
Compliance guardrails | Opaque ChatGPT guardrails, no custom brand guardrails | ChatGPT can’t send or change flows/segments in your account, and compliance rules still live in your CRM; review required before anything goes live | Custom guardrails based on product exclusions, legal disclaimers, banned phrases, and regional compliance |
Omnichannel messages | Manual prompting, manual sending | Email-focused drafts and templates via MCP; cannot create or send SMS, build/edit flows, or auto-route messages by preferred channel | Builds email and text message campaigns and sends automatically to the channel each subscriber is most likely to engage with next |
Measurement | No measurement, drafting only | Live campaign and flow performance with analysis cards and deep links; no full CRM analytics suite or charting beyond the app’s tables | Performance measurement with built-in analytics |
Speed up time to market with Klaviyo AI
Klaviyo is the autonomous B2C CRM that brings together real-time customer data, built-in agentic AI, marketing automation, and customer service so brands can power more valuable customer experiences across the customer journey.
If you already work with ChatGPT, the Klaviyo App for ChatGPT can connect your Klaviyo data to your ChatGPT interface. That way, when you’re drafting marketing strategy and content in ChatGPT, you can do it with real performance data at your fingertips.
Klaviyo Composer, meanwhile, bridges the gap between data and content creation. Composer is an AI marketing agent that can identify your biggest growth opportunities and turn them into ready-to-launch campaigns in minutes. With Composer, you can:
- Draft more than just copy. Composer builds the audience, the copy, channel coordination, and timing, all in one conversational session.
- Send omnichannel campaigns and flows. Build multi-step cross-channel message sequences with channel-appropriate content, timing, and sequencing in a single generation session.
- Measure what each send actually earned. Klaviyo reports across channels, so you know what needs to improve for your next campaign.
Personalize your AI-generated messages with Klaviyo.
Get a demo