Prompt Engineering for Product Managers: How to Get Things Right With Generative AI

I’ll get straight to the point – if used well, generative AI can transform the way you work, what you’re able to achieve, and the progress you make in your career. Right now, getting good at using these tools will set you apart, but it won’t be long before effective AI prompt engineering is considered a vital skill for a Product Manager. So don’t get left behind. Let me show you how to master prompt engineering for Product Managers. 

I don’t imagine any of you will be new to generative AI and language models. By now, everyone has given it a go, and most will be using it at least semi-frequently. But have you reached that transformative tipping point where it’s unlocked considerable performance benefits for you? 

The secret to really unleashing the transformative potential of generative AI lies in the prompts you’re feeding it. To truly power up what you’re able to achieve, you’re going to have to learn prompt engineering for Product Managers. That or use a tool already primed with Product Management context and instructions – like CoPilot. 

But, let’s dive deeper into how to use effective prompt engineering for Product Managers to get relevant outputs. 

We will cover: 

• The benefits of prompt engineering for Product Managers
• How Product Managers can use AI
• The principles of good prompt engineering for Product Managers
• How to structure prompts for Product Management

What are the benefits of prompt engineering for Product Managers? 

As you can probably tell, I’m a big believer in the benefits AI can offer Product Teams. But I also realize some may have a degree of concern – a fear that AI will come to replace Product Managers. 

Let’s nip this in the bud. No, AI won’t replace you. Don’t fear AI, rather embrace it as a superpower that will help you boost your performance and shore up your career prospects. 

Look, AI is a tool – a powerful tool in your toolbox. When Photoshop launched it didn’t spell the end for photographers, rather it gave them new capabilities and helped them do more. Yes, they had to learn how to use this new tool, but those who mastered this new way of photo editing, skyrocketed what they were able to achieve. AI tools can be your Photoshop. You just have to learn how to use them.

When you’ve mastered prompt engineering for Product Managers, you’ll have a game-changer on your hands. The benefits include: 

• Saving considerable time: When you feed generative AI a well-structured prompt, it can deliver highly relevant, and well-written outputs faster than you could draft yourself. Whether you’re summarizing a user research session or writing up your documentation, AI can save you time while keeping the quality high.
• Boosting your creative problem solving: Generative AI can become your creative sidekick, turning well-constructed prompts into a stream of fresh ideas. Need a new perspective on your product differentiation or possible solutions to a problem? A strong prompt can spark ideas you hadn’t considered.
• Speeding up your iteration cycles: Generative AI accelerates processes like concept testing, prototyping, and creating an MVP. By producing usable outputs faster than traditional methods, it allows you to iterate, refine, and adapt at a greater pace.
• Improving team productivity: Generative AI doesn’t just help you – it helps the whole  Product Team. Automating grunt work and speeding up tasks will mean everyone is more efficient.
• Enhancing your Product Management expertise: Generative AI isn’t just there to delegate tasks to. Yes, it can kick out great writing in seconds, but it can also answer your questions! Especially if you use an AI chat tool specifically trained on Product Management know-how, like CoPilot, you can lean on AI to help assess how you’re working, give advice on how to approach a piece of PM work, and generally help you understand best practice. 

How can Product Managers use AI? 

So, what can AI help with during your day-to-day as a Product Manager? Where is it best applied to unlock the time savings and performance boosts I’m promising? With effective prompt engineering, AI can help with virtually every area of product development. 

1. Product Strategy 🚀

AI tools can be a great help when it comes to your product strategy, for example:

Strategy communication ✍️

AI tools can help you articulate your product strategy. We all know how important it is to communicate your overarching ambitions and priorities in a way that everyone can understand. Without that, you stand no chance of getting alignment across teams.

You need to remove ambiguity and inspire your teammates to work towards the vision. The best place to start when it comes to getting AI help is your vision statement.

Either tell your AI tool what your product is and what you want to achieve with it and ask it to write a motivating, clear product vision statement, or, give your existing vision statement to your tool and ask it for constructive feedback and improvements. 

Generative AI is great at taking a lot of words or streams of notes and turning it into something concise. It’s also great if you tend to think in bullet points, and need that converted into more creative prose.

Goal setting 🎯 

Once you’re clear on the broad ambitions of your product, it’s time to get more specific and set some goals to work towards. 

Provided you give your tool the context of your product and the broad vision, you can ask AI to generate relevant objectives and goals. Be sure to specify your preferred framework – e.g. OKRs – and be clear on the format you want. It’s useful here to give one example and then let the AI generate others in line with that. 

Idea generation💡

Whether you’re thinking about a new product to solve a problem you’ve identified in the market, or looking for potential feature ideas as part of roadmap initiatives, AI can kick-start your thinking. 

Just outline the context for the AI, feed it your vision, objectives, and whatever else you have, and ask it to come up with some product ideas for achieving those goals and solving the problem.

2. Discovery 🔎

AI can be your friend when it comes to your discovery process, whether it’s your initial product discovery on a brand new product proposal, or your continuous discovery to validate each idea in your backlog. Here are a few areas where you can employ AI to speed things up. 

Market and competitor research (with caution) 📊

Now, I have to add a note of caution here. Yes you can use AI to help you with market or competitor research, but be conscious that most general AI models will have a knowledge cutoff. The knowledge cut off represents the point in time when the data feeding the AI model was last updated. For example, for GPT-4o models the cutoff is October, 2023 (at time of writing).

Therefore, in most cases, your AI tool is not going to have up-to-date intelligence on market trends or your competitors. So asking AI to do something like ‘create a feature comparison’ is unlikely to give very accurate results. 

However, you could ask AI to give you an assessment of a particular market to use as a base against which to manually fact-check and get updates. If you’re struggling to know how to structure a market analysis report, your AI tool could kick one off for you. At least it gets you off a blank page! Or ask your AI tool to take a long (and recent) industry report or a competitor annual report and summarize it.

User research 👥 (caution again)

OK, I’m going to add another note of caution here. You need to avoid overreliance on AI when it comes to user research. Nothing should replace your efforts to speak to real or potential customers. 

Thorough user research is crucial for the validation of ideas and ensuring what you build will drive the outcomes you want, so you have to be certain you’re using solid evidence to make informed decisions. 

Don’t think you can simply ask AI “Would a customer of a mobile banking app find a budgeting tool useful?” and make your decision based on the output. 

But does that mean AI can offer nothing useful when it comes to user research? Absolutely not. AI could help by: 

• Suggesting research methodologies
• Generating research questions for user interviews or focus groups
• Writing test scripts for user testing
• Helping to prepare research reports and presentations
• Analyzing data from your research efforts to help you draw conclusions

This brings us nicely onto….

Data analysis 📈 

AI is pretty darn good at analyzing large amounts of data and spotting themes, patterns, or irregularities. And that can be a huge time-save for Product Teams. No longer do you have to run your own affinity mapping workshops to find common themes in your feedback (for example), or spend hours wading through usage data to spot patterns. 

However, you should think carefully about what AI tools you use for your data analysis. If you use a general AI tool,  then you’re going to have to package up your raw data and upload it. Not only will that require exporting from wherever that data is, formatting it, and uploading, but you’ll also have to explain that formatting to the AI so they understand what they’re looking at. That’s a lot of hassle. 

The other option is to make sure you’re using a data capture tool that has robust built-in AI capabilities. This way the AI already has your data within its source content and you cut out all that exporting and importing. 

So look for product analytics tools that have AI capabilities and customer feedback tools that offer AI-powered automatic analysis.

AI prototyping 🛠️

When it comes to testing possible solutions and products with real users, AI can really accelerate what you’re able to achieve as a Product Team or lone Product Manager. You can use AI to get a prototype off the ground, without having to fight for development resources to help you do it. 

There are specialist AI tools for writing code, but equally, general generative AI models can write code and knock up a prototype for you. 

You can prompt AI models with some well-crafted prompts, feed them a design or even a PRD from which they can formulate the necessary code to bring the prototype to life. 

3. Feedback 🗣️

Managing customer feedback is a huge part of the Product Manager role, and is often where a lot of time is lost. So how can AI help you move through user feedback faster, so you can get to the insights and start working on solutions? 

Capturing feedback 📥

One way AI can help with capturing feedback is through turning customer interactions into usable content. For example, taking advantage of generative AI capabilities offered by many video conferencing tools can turn a video call into a written transcript in moments.

There are also AI note-taking tools that you can add to any call and get instant write-ups that you can add to your feedback inbox. 

Summarizing ✏️

And if those long transcripts are too much to easily digest and make sense of, AI can give you a succinct summary and save you from reading through reams of text.

Analysis 🧐

We’ve already touched on this when we covered data analysis, but it’s worth saying again! AI can save you a bunch of time and surface the themes across your entire body of feedback in moments. To get the idea, take a look at how ProdPad’s Signals tool works. 

4. Prioritization ⭐️

Prioritization is both an art and a science. It’s where Product Managers shine, but figuring out what to tackle first and balancing stakeholder demands, customer needs, and strategic goals is no small feat. Luckily, AI can help simplify the process. 

AI can analyze inputs like customer feedback, user behavior, and business objectives to provide priority scores for your product ideas. For example, with CoPilot, you can ask it to analyze Ideas on specific roadmaps and review them with whatever prioritization framework you like.

If you’re using prioritization frameworks like RICE, AI can crunch the numbers for you. Input your data—such as the effort estimates or customer reach of a feature—and let AI calculate scores or assign categories. This saves time and ensures consistent, unbiased assessments.

5. Backlog Management 🗂️

Depending on what tool you use to manage your backlog of product features ideas, AI can help you save time when it comes to grunt work. 

Let’s face it, you didn’t become a product professional to push tickets around a board – you’re here to make strategic decisions and drive outcomes. So the more you can rely on AI to handle the admin stuff, the better. 

ProdPad customers enjoy AI assistance when it comes to managing their backlogs with duplicate ideas being automatically flagged and removed, and feedback being linked to related ideas (and vice versa). That, amongst other things, saves a bunch of time and frees them up to concentrate on discovery and decision-making. 

6. Product Documentation 📄 

OK, here is another place where there are rivers of time that can be saved with the help of AI. Generative AI has been a game changer when it comes to writing copy and producing documentation. Just give your chosen tool the context of your product, the particular feature idea and/or the intended user and ask it to create whatever documentation you need. 

In some cases you might have to be explicit about the structure and format you want to see, at other times you might be happy to see what the AI generates. 

Some of the documentation you could delegate to your AI assistant might include:

• Idea descriptions
• Product requirement docs
• Specifications
• User stories
• Acceptance criteria
• Release notes
• Customer emails
• Internal updates 

7. In-Product Copy Creation ✍️ 

Internal documentation isn’t the only writing you have to do as a Product Manager. You need to write convincing and helpful in-product copy that helps to drive users towards certain actions. 

Whether you’re encouraging users to make a purchase, take an onboarding step, or try a new feature, you need to craft conversion-focused words – and that’s not easy. 

If you get your prompt right (keep reading to find out how!) you can get very convincing copy out of your chosen AI tool. Then you just need to copy and paste it where it’s needed and sit back and watch the results. 

8. Stakeholder Management & Communication 🤝 

This is another area of Product Manager responsibilities where time sinks are all too common. Here at ProdPad, we’ve always focused on how we can make this easier for Product Teams and reduce the manual work. From customizable roadmap views, to easy external roadmap publishing, automatic update notifications, to tight integration with tools like Slack and Teams. 

ProdPad has a whole host of capabilities that take the stress out of stakeholder comms. But how can AI help even further? 

CoPilot, as an AI assistant that sits deep within your Product Management system, has access to your roadmap, your backlog, all your customer feedback, your strategy, OKRs, and more. This unique knowledge means that CoPilot can answer almost any question about your product work. This is a complete game changer when it comes to fielding those day-to-day, impromptu questions from stakeholders across your organization. 

For example, let’s say your boss wants to know everything on the roadmap that relates to a certain strategic objective. Sure they could look at your roadmap (and even group it by Objective in ProdPad), but the chances are they’re just going to fire the question over to you. 

With CoPilot you can give them an alternative outlet for those ad-hoc questions – CoPilot can tell them exactly which Initiatives and Ideas answer their chosen objective and even provide links to each. 

With CoPilot fielding all the questions from your stakeholders, you’re no longer going to get pulled away from your deep-focus work and get to crack on with more of what matters most.  

9. Coaching and best practice advice 🎓 

Where you might have gone digging around in forums, asking in online communities, or searching online, now you can add AI to your sources of best practice advice and guidance. 

Sense-checking the way you approach a certain Product Management job, or asking for advice on how best to do something, is a great idea if you want to be the best Product Manager you can. So I always advocate the use of AI tools as sounding boards or fast-access coaches to help you understand best practice ways of working. 

But, of course, the advice AI will give is only ever going to be as good as the advice the model has been fed and trained with. 

Take CoPilot for example, CoPilot is an AI sidekick built specifically for Product Management and has been carefully fed with certain, curated sources of best practice information to ensure it always delivers the best coaching and advice.

The secrets of prompt engineering for Product Managers

OK, now you know the potential – all the different ways generative AI can help you do more and move faster across the whole Product Management lifecycle. But, as I’ve said, you won’t necessarily get results you’re happy with right off the bat – certainly not with the most common generalist AI tools. So let me show you how to master the science (or is it art?) of good prompt engineering for Product Managers.

Here are the general principles you need to remember when engineering your prompts:

• Provide context and information: AI can’t read your mind – it needs relevant details to work effectively. Always include background information in your prompt. If you’re asking for suggestions about your product, it needs to know what your product is! Feed it clear context like user personas, product goals, or the value proposition.
• Keep it simple and structured: Overloading the AI with too much detail can confuse it, just like handing someone a 50-step IKEA manual. Instead, focus on concise, goal-oriented prompts. For complex tasks, break them into smaller, manageable parts to ensure clarity and accuracy in responses.
• Use natural language: AI responds best to prompts written in everyday language, just like talking to a teammate. Avoid robotic phrasing or overly formal tone, and stick to clear, conversational language to get the best results.

Of course, things get wayyyy deeper than this. To help you master this valuable skill, there’s a useful framework I want you to meet:

The W-I-S-E-R Framework. 

This framework was created by Allie K. Miller, one of the most influential voices in AI in business. It’s designed to help you give generative AI all the context and information it needs for to deliver a cracking result.

Here’s what the W-I-S-E-R structure gets you to do. 

W – Who is it? 🗣️ Assign the AI a role. For example, “You are a Product Manager creating a go-to-market strategy for a SaaS platform.”

I – Instructions ✏️. Be specific about the task. Say something like, “Draft a high-level GTM plan with key action points.”

S – Subtasks ✂️. Break the request into smaller pieces. For example, “Start by outlining the target audience, then list three marketing channels, and finally suggest KPIs to track success.”

E – Examples 🖼️. Provide a reference or template. Say something like, “Here’s an example of a roadmap format we’ve used before—align your response with this structure.”

R – Review 📖. Refine the output. Ask for adjustments like, “Add more detail to the target audience section,” or “Reformat this as a presentation outline.” Iterate as needed.

Nice. But we can go EVEN DEEPER! Let’s look at each of those step by step and discuss some advanced prompt engineering techniques to help you build a better structure for your prompt engineering.

How to structure AI prompts for Product Managers

W – Who

I’d like to expand on the first step in the WISER framework, because yes you need to tell the AI tool from what perspective they should be generating their output, but there’s more to giving relevant context setting that just ‘who’. 

You need to outline ‘who’, ‘what’ and ‘why’. 

Since we’re here to talk about prompt engineering for Product Managers, let me illustrate this with a Product Manager example. 

Who = a Product Manager 
What = managing a mobile banking app 
Why = designed to help young people better manage their finances 

There are a couple of advanced prompting techniques that I’d like to introduce here, each of which can prove useful when setting this context within your AI prompts for Product Managers. 

Domain priming

Domain priming involves instructing the AI to adopt a specific role or perspective when responding. This technique is how you make the AI answer from a ‘Product’ perspective. 

Role-playing

This is kind of  like domain priming, but slightly more creative. It’s a good technique if you want to explore different perspectives on something. This could be useful if you wanted to kick off some customer research and generate a list of possible pain points for different user types. You can get the AI to pretend to be a user, getting some creative outputs as a result.

So, the opening of your prompt might look something like this:

You are a Product Manager for a mobile banking app. The app is designed specifically for young people (aged 16 – 25) to help them learn financial acumen and better manage their finances.

I – Instructions

The next stage is to set your instructions. This is where you’re prompting the AI with exactly what you want it to deliver. Want a table of results? Tell it that. What a mindmap? Demand it. Keen for a bullet point summary? You better mention that.

If your instructions aren’t clear, the AI is going to do what it thinks best – which might miss the mark if you have a set idea of what you need. 

Now there are a few advanced prompting techniques that could help you here. One option is:

Chain of Thought (CoT). 

This technique involves asking the AI to reason step-by-step. It’s particularly useful for complex prompts, as it ensures that the AI breaks down the process logically. For example:

“List three common objections personal banking customers might have to using a budgeting feature. Then, for each objection, suggest a solution or feature improvement to address it.”

This clear structure encourages better-organized responses and helps you get actionable insights faster.

Remember: the more precise your instructions, the better the output. Vague instructions will yield vague results, but thoughtful direction will maximize the AI’s potential to deliver exactly what you’re looking for.

So, if we continued our prompt, the ‘I’ section may look something like:

Adoption and usage rates are low for our budgeting feature. We need to come up with ideas to solve this problem. List three common objections our banking customers might have to using the budgeting feature. Then, for each objection, suggest a solution or feature improvement to address it. 

Present your ideas in a concise bullet-point format, including how each solves the problem.  

Of course, if you’ve got a more complex ask, that has a few steps, you’re going to want to break things down so that everything remains simple. This leads us to…

S – Subtasks

You can break your prompt into different sections if what you need is a bit more complicated. 

For example, say you want to map out a Product Manager’s approach to increasing feature adoption. This task involves many steps: understanding user pain points, brainstorming potential solutions, evaluating their feasibility, and creating a communication strategy. 

To get meaningful responses, you’ll want to break these steps down into smaller, more manageable subtasks. You do not want to ask for all of this at once otherwise the machine might get its wires crossed. 

Prompt-chaining

One useful advanced AI prompting technique here is prompt-chaining. This is where you connect multiple prompts together to build on the results of the previous responses. Instead of asking for everything at once, you guide the AI through a logical sequence of tasks, step by step. For instance:

1. Start by asking the AI to list common reasons why users don’t adopt new budgeting tools.
2. Once you have this list, ask the AI to generate possible solutions for each identified reason.
3. Finally, prompt it to suggest the best way to communicate these solutions to users, keeping their needs and preferences in mind.

By chaining prompts like this, you can get detailed and well-structured outputs that align with the complexity of your task. It also helps maintain focus, ensuring the AI doesn’t get overwhelmed by too many simultaneous instructions. This is one reason why many prompts fail – you’re asking too much.

Continuing on our example prompt, the subtasks section will look like: 

After generating three solutions, rank these solutions, using two criteria: 
1. Technical feasibility: How easy is it to implement each solution from a technical standpoint?
2. Impact versus effort: How effective will the solution be in increasing user adoption, versus the resources (time, cost, etc.) needed to implement it?

E – Examples 

If you really want to steer your AI prompt in the right direction, give an example of what you’re looking for. The example acts as a clear target that guides the AI’s reasoning and structure.

Plus giving an existing example also ensures that the AI doesn’t come up with something you’ve already considered.

Few-shot prompting

One advanced technique related to this is called few-shot prompting. This method involves providing the AI with a few examples of the type of response you’re expecting, instead of just a single example or no example at all. 

So when giving examples for our prompt, you can add something like:

A couple of pre-existing ideas we had include: 
1. Implementing an in-app tutorial that explains how to use the budgeting feature. This addresses the pain point of finding the feature too confusing but is a large development time sink. 
2. Gamifying the budgeting feature by offering personalized incentives for users who complete goals when using the feature. This encourages continuous adoption but may not get approval from other stakeholders.

R – Review 

Now, after following the first few steps of W-I-S-E-R, you’re going to get a far better response compared to basic prompts. But still, this first response isn’t going to be the best it can be. Just like any writer revising their first draft, the AI’s initial output can often benefit from some refinement. This is where the review stage comes in.

By reviewing the response and using the reflection technique, you can further improve the quality and relevance of the output.

Reflection

The reflection advanced prompting technique allows you to engage in a second round of thinking with the AI. In essence, you ask the AI to reflect on its own work, evaluate its decisions, and identify areas that can be improved. This iterative process helps with refining prompts by encouraging the AI to be more accurate, focused, and creative.

To nail this, specify what you want it to look at during the reflection, such as if it addressed all the pain points you provided, or aligns with the context. What you ask is specific to your goals, but some general things you want to check with a reflection include: 

• Clarity 
• Creativity 
• Feasibility 
• Gaps

So once you’ve gotten your first response from our example prompt, you can follow up with:

Review the proposed solutions, paying close attention to clarity, creativity, and feasibility.

Next, identify any gaps in your response, or if anything has been overlooked. Could certain aspects of the solutions be more aligned with the target audience’s pain points?

Finally, reflect on the effectiveness of the solutions you proposed. Could they be made more actionable or user-friendly?

So with that, we’ve got a complete prompt, and follow-up, following the W-I-S-E-R framework, alongside some advanced AI prompting techniques to generate accurate responses.

Can’t be bothered with all that?

Now many of you might be thinking – ouch, this is a lot of work for something that’s meant to be making my life easier. If I need to put so much effort into creating an AI prompt just to make the response okay, why don’t I just go and do the thing myself? 

Fair comment, and a fair complaint. Luckily, there’s an AI tool for Product Managers that doesn’t need this level of extra context and detail. An AI tool where you don’t need to add context every time you write a prompt because the model already has an understanding of your product, roadmap, backlog, and more.

I think you already know where this is going…

This is exactly what CoPilot does! 

When using our AI tool, you don’t need the preamble. Want it to refine your roadmap? Just go ahead and ask it.

So before you worry too much more about prompt engineering for product managers, go give CoPilot and try and see how much faster you can get to the results you want. 

To learn even more, check out our webinar on writing great AI prompts for Product Management, hosted by yours truly. 

The prompt engineering for Product Managers playbook 

AI is a game-changer for Product Managers, but the real magic lies in knowing how to use it effectively. Think of it like a violin: in the right hands, it can produce breathtaking music, but without the skill, it’s just ear-piercing noise.

Mastering the art of prompt engineering for Product Managers is a valuable skill that unlocks AI’s potential. It transforms AI from a mere tool into a powerful ally in your Product Management toolbox.

Whether you’re skeptical, excited, cautious, or curious about AI, the reality is clear: there are countless AI tools out there that can make your work as a Product Manager more efficient and impactful. Among them, CoPilot stands out as the ultimate choice for Product Managers.

Ready to see CoPilot in action? Start a free trial and try for yourself. We’re confident you’ll be impressed.