Listen to the audio version of this article:
AI Strategy Benefits
My research shows that AI can help you make better strategic decisions faster, at least for certain products.[1] Below are four examples of how this can be achieved. Note that I’ve decided not to state the names of the tools I found, partly as the AI landscape is changing rapidly and partly as you should research and select the tools that work best in your context rather than trusting my judgment.[2]
Market Research
AI-based tools can discover user and customer trends using predictive analytics. This can help you create a new strategy and evolve an existing one. It assumes, though, that enough good-quality data is available to make reasonably reliable predictions. This is unlikely to be the case for disruptive innovations, as I discuss below, as well as specialised products with a comparatively small user base, like tailored IT solutions.
Customer Insights and Idea Generation
AI tools can analyse market data, customer feedback, and emerging trends to suggest new products and features, assuming that enough relevant data is available. For example, you can use an AI tool to analyse support tickets to discover and address common issues. You might even ask a chatbot like ChatGPT to come up with ideas to understand if you have missed any opportunities.
Product Differentiation
AI tools can help you make your product stand out from the crowd and offer a clear reason for people to choose it over alternatives. This can be achieved in two ways: First, data mining can identify opportunities for differentiation, assuming that the relevant data exists. Second, offering AI-enabled product features, including a personalised user experience and user-specific recommendations, can give your product a unique advantage.[3]
Take Spotify’s DJ and TrainerRoad’s TrainNow features, for instance. The former is a virtual assistant who recommends favourite songs and allows listeners to discover new music. TrainerRoad is an indoor cycling app that offers personalised training plans. Its TrainNow feature suggests a workout for users who do not follow a structured plan but want to train whenever they feel like it and still get the right workout recommendation.[4]
Product Performance and KPIs
AI tools can continuously monitor how much value a product is creating and recommend improvements. This helps you understand if the current strategy is still valid or if it needs to be adapted. This, in turn, can facilitate continuous strategising and maximise the chances of proactively responding to opportunities and threats.[5]
What about Product Roadmap Generation?
You might have noticed that I didn’t list the creation of product roadmaps as an AI benefit, even though several tools offer it. There are two reasons for this: First, the AI-generated roadmaps I came across during my research were feature-based, which is a roadmapping approach I don’t recommend. Second, it wasn’t clear to me if and to what extent the roadmap elements were guided by an overall product strategy. This is necessary, though, to ensure that the roadmap is aligned with the strategy and that user needs and business goals are translated into more specific outcomes.[6]
AI Strategy Limitations
While it can be of great help, AI is no silver bullet. You should be aware of the following five limitations when using AI tools to discover and evolve a product strategy.
Data Dependency and Risk of Biases
As mentioned earlier, AI tools require enough good-quality data to generate helpful results. If you don’t have sufficient data available or if the data quality is not right, if, for example, it contains cognitive biases, you are unlikely to benefit from using AI tools. In the worst case, you get the wrong results and make the wrong decisions.[7]
Probability vs. Accuracy
Generative AI tools give the most likely answers, not necessarily the correct ones. Their predictions are based on degrees of confidence rather than absolute certainty. Consequently, you should consider if AI-generated results are likely to be good enough. As a rule of thumb, the higher the impact of a decision is, and the harder it is to reverse it, the more certainty you require.
Additionally, you should review AI-generated answers rather than blindly trusting a tool, even if the results sound very convincing. For example, while researching AI tools for this article, I used several chatbots, including Perplexity. Reading the references used by the latter, I found that most of the time, the bot did a great job at extracting and summarising information—but not always.[8]
Hard to Apply Disruptive Products
AI’s data dependency makes it hard, if not impossible, to use it for disruptive products. Such an innovation does not only create a new product but also a new market. Take the original iPhone as an example. Until its launch, smartphones primarily addressed the needs of business users. The iPhone, in contrast, was the first consumer smartphone. It made it attractive for non-business users to purchase smartphones. The challenge for AI is that “markets that don’t exist can’t be analysed,” as Clayton Christensen once said.[9] AI-based tools will therefore struggle to come up with helpful results for disruptive products.
Lack of Empathy
While you can use sentiment analysis to uncover user emotions, AI-based tools are no replacement for meeting users and customers. To make the right strategic decisions, you must have a sound understanding of the user and customer needs and be able to empathise with them. This is best achieved using methods like direct observations and interviews. You should therefore “get out of the building,” observe how people use the product in its target environment, and talk to them to find out what works well for them and what doesn’t.[10] Conversely, if you cannot meet actual users, you risk making suboptimal or wrong strategic decisions. It would be foolish to assume that AI tools can understand the needs and motivations of users better than a fellow human being.
Environmental Impact
Generative AI tools, like ChatGPT and Perplexity, increase the amount of electricity consumed, require additional hardware, and have led to the creation of new data centres.[11] As it’s difficult to meet the increased demand for electricity using only sustainable energy sources, AI tools may increase the environmental impact and carbon footprint of your company, and they may jeopardise your product’s ethicality.[12]
AI Strategy Foundations
I hope that the previous sections have given you an idea if AI-based tools are suitable to help you create a winning strategy for your product. To find out more, you should evaluate different tools, possibly together with your product management colleagues. While this requires time, it will help you choose the tools that are a good fit for your product and domain rather than picking them from a predefined list and hoping that someone else’s evaluation applies to your context.
However, simply adopting the right AI tools may not be enough. To take full advantage of them, you have to have the right product strategy practices in place. Otherwise, you risk being tool-led. In the worst case, you will make the wrong decisions faster. It’s like buying a new pair of running shoes and hoping that this will make you a better runner. It won’t—unless you also improve your running technique, strength training, diet, and recovery process.
To understand if your strategy practice is in good shape, ask yourself the following five questions:
- First, is it clear who is involved in making strategic product decisions and who has the final say? Is it, for example, the head of product who determines the product strategy with input from the product managers and stakeholders? Or do you use extended, fully empowered product teams that own the strategies of their products
- Second, has your current strategy been captured and made visible using a tool like my Product Vision Board? Additionally, will it help you to achieve product success in the future? Is it a winning strategy?
- Third, are strategic decisions based on empirical evidence rather than opinions and beliefs? Do you systematically de-risk a new and significantly changed product strategy, as I explain in the article Product Strategy Discovery? Does this include meeting selected target users and developing a deep understanding of their needs?
- Fourth, do you systematically review the product strategy? Do you use the right KPIs to determine if the strategy is still effective? Do you regularly interact with (selected) customers and users to understand if their needs and expectations are changing? Do you keep an eye on the competition and relevant trends, including new technologies? Do you proactively adapt the strategy to exploit opportunities and mitigate threats?[13]
- Fifth, is the product strategy aligned with the business strategy and product portfolio strategy as well as the product roadmap? Do the business and portfolio strategies guide it? And does it effectively direct the product roadmap?
If your answers to the five questions above are positive, you’re in a good place, and using the right AI tools will most likely benefit you. But if that’s not the case, you should improve your strategy practice first before you heavily invest in AI. This, in turn, will avoid the risks of being overly reliant on the tools, making the wrong decisions, and taking your product down the wrong path. “A fool with a tool is still a fool,” as Grady Booch once said.
Conclusion
AI tools can support you in creating and evolving a winning product strategy: They can help you make better decisions faster. To leverage the benefits, you need to be aware of AI’s current limitations, select the tools that work best in your context, and have solid product strategy practices in place. To future-proof yourself, I recommend experimenting with AI tools as well as increasing your product strategy knowledge.
While the capabilities of AI tools will undoubtedly increase further, I don’t see them creating effective strategies without human involvement any time soon. After all, a key prerequisite for developing a winning product strategy is the ability to empathise with users and customers. Strategy development will therefore continue to be a deeply human activity. AI tools aid this process, but they don’t fundamentally change it.
Notes
[1] Special thanks to Janna Bastow for sharing her views on AI with me and helping me understand how ProdPad uses AI to assist its users. I did my best to take a balanced approach, neither glorifying nor vilifying AI but describing its benefits and limitations in developing a product strategy as objectively as possible.
[2] Note that I intentionally don’t list benefits, like time savings, offered by more generally applicable AI-powered tools such as notetaking and email management products.
[3] While I find that taking an interest in AI and having a fundamental understanding of AI technologies is important for product people, I view it as the job of a cross-functional development team to choose the right AI technologies to offer a great product. If you want to learn more about this topic, take a look, for example, at Marily Nika’s book Building AI-Powered Products and Bobcats Coding’s paper “AI Washing Explained.”
[4] DJ is a virtual assistant who recommends old favourites and allows you to discover new genres, playlists, and artists, according to Spotify. TrainNow gives cyclists the right workout every time, even if they’re not following a training plan, as TrainerRoad claims. My personal experience with both “intelligent” product features is mixed at the time of writing. They sometimes work well for me. Other times, they don’t.
[5] It assumes, though, that you have the right KPIs in place and that the related data can be automatically captured.
[6] I have found outcome-based roadmaps, like my GO Product Roadmap, to be more effective than traditional feature-based ones. To learn how you can systematically connect the product strategy and roadmap, please see the article Roman’s Product Strategy Model.
[7] A technique for recognising biases and reasoning issues is Chain-of-Thought (CoT) prompting. This results in a chatbot providing information about how the answer was achieved. See, for instance, https://www.promptingguide.ai/techniques/cot. Additionally, as humans, we also have biases. We may not always be aware of them, though.
[8] Gen AI tools can “hallucinate” and present false or misleading information as fact. For example, research published in May 2023 found high rates of fabricated and inaccurate references in ChatGPT-generated medical content. 47% of the references were fabricated and 46% were authentic but inaccurate. Only 7% were authentic and accurate.
[9] Clayton. M. Christensen. The Innovator’s Dilemma, p. 143.
[10] Note that it can be useful to employ synthetic users, which essentially are AI-generated fake users. They should, however, complement, not replace, real research, see Maria Rosala’s and Kate Moran’s article “Synthetic Users: If, When, and How to Use AI-Generated Research.”
[11] In North America, for example, the power requirements of data centres rose by roughly 50% from 2022 to 2023. This increase was at least partly driven by the demands of generative AI, see Adam Zewe, “Explained: Generative AI’s environmental impact.”
[12] I see ethicality as a product success factor that must be met in addition to desirability, feasibility, and viability, see the article Four Product Success Factors.
[13] I recommend meeting selected users and customers at least once every three months to learn to which extent the product addresses their needs and discover opportunities to innovate and offer more value.
