You might have heard that we’ve recently launched a little something called AI Sprint. Think of it as consulting on steroids where our team joins forces with a client during a discovery session. As a result, together we come up with the most suitable software solution for the client.
AI sprint is especially helpful when the client only has a general direction or idea. At this point, clients usually don't know how to execute them, or they might be simply looking for guidance with the first AI implementation in their business. The session aims to find the right shape and direction for the AI project.
Our AI sprint is divided into several steps, and you can read about them in our post How to start with AI in your company: a checklist. These steps include defining strategy, market research, and goal definition, among others. This is the “what.” On the other hand, this post will focus on the “how” to help you learn even more about the perfect workshop. Let’s dive into some useful tips that can transform the outcome of the AI Sprint process.
The golden trio of team assembly
The first step that can make or break your project is assembling the team. Once you reach this stage, make sure to carefully consider who should participate to ensure success. In our experience, you need three types of specialists to build your dream team: one who’s business-oriented, one technology-focused, and one who knows exactly what the end user is looking for. This will diversify your thinking and cover all crucial aspects of the project. Remember to also include individuals with domain knowledge if your product is directed towards a specific audience or industry.
This approach was especially helpful during one of our projects that involved redesigning a retail eCommerce platform. While the business strategist highlighted the importance of personalized marketing, the tech lead focused on implementing new functionalities, and our UX specialist handled the user experience transformation. Combining these perspectives helped us pinpoint user engagement at the intersection of the three sets of priorities. By focusing on this aspect, we developed a platform with a high user engagement rate as the primary means of boosting sales. This resulted in a 70% YoY revenue growth.
How to define your project’s goal
It’s crucial to know what questions to ask and how to formulate them for the most effective outcome. Try taking a step back and looking at your project from a wider perspective. Carefully refine your definition of success and the path to achieving it to avoid possible misunderstandings. If done correctly, you should have complete alignment within your team and can move on to the next phase.
It’s important to remember that during the research phase, the initial project goal may change. This was the case for us when the goal of one of our projects changed from ‘increase sales’ to ‘create a seamless experience’. It wasn’t about abandoning the goal of increasing sales numbers altogether. We concluded that increased sales would be the natural result of improving the user experience. This pivot, backed up by insights from our analyst, ensured the project stayed on the right track and was aligned with the true pain points of the client.
Why do the real answers come in so late?
Generating ideas too early instead of diving deep into the research phase can cause your project to fail because you hinder your search for the real solution. Simply put, if you get attached to a particular idea too early, your view might become narrow, and you’ll only focus on confirming that this is the right way to go.
To prevent this, make sure to immerse yourself in knowledge. Search broadly, creatively, and loosely. Conduct a data lineage workshop to discover where your data is created, how it flows through the company, and which processes are done manually and which automatically. This will help you determine bottlenecks, potential savings, and possible accelerations.
When our team was working on an AI-powered marketing tool, early brainstorming sessions led us to insufficient solutions. It wasn’t until we conducted a data lineage workshop that we identified data flow issues and manual work that could be automated. This extended research and deeper understanding allowed us to visualize and later implement an automatic, easy-to-scale product for analyzing collective marketing data that delivers valuable insights and supports decision-making.
How to prepare for interviews
User interviews are always an essential part of the process. They’re your best shot at getting a complete picture of pain points and needs of everyday users of your future solution. When planning for this stage, it’s easy to generate numerous questions, but your preparations should go a step further.
The secret ingredient is collaborating with your team to simulate potential answers and determine their value and practicality. Are the answers truly helpful in advancing the project? Evaluate your script for its usefulness and be prepared to delve deeper during the interview, leveraging the diverse competencies of the golden trio.
During an AI integration workshop for one of our eCommerce sites, we simulated the customer journey from beginning to end, which helped us discover the potential touchpoints. Since our trio closely collaborated to evaluate the feasibility and impact of each idea, the workshop was highly productive and gave us actionable insights: personalized product recommendations and improved customer support were clearly the key areas to work on.
How to come up with the right questions
Once you reach the ideation phase, where you start to generate actual solutions, it might be tempting to stick to an “anything goes” approach. After all, you’re initially looking for quantity over quality. There’s a better way to approach this.
Use the “how might we” questions to ideate on problems. Start with your end-users' pain points or insights gathered in the previous phases. Keep your questions broad, but always keep the desired outcome in sight. Avoid negative questions that use verbs like “reduce,” “remove,” or “prevent,” such as “How might we reduce the frustration of long wait times for customer service?” Instead, replace them with “create,” “promote,” or “enhance,” for instance, “How might we create a more engaging customer experience?”.
For example, if you’re developing an AI chatbot to support your customer service and are thinking “How could we reduce the customer service costs?”, try framing it as “How can we make customer service more efficient through AI?” This more positive approach will quickly show you that a chatbot will resolve your issues and provide a personalized experience, enhancing overall satisfaction.
Tech isn’t always the answer
Remember that the solution will not always be the technology you choose. It may as well be a business process, a roadmap, architecture, or, of course, an implementation of a specific minimum viable product (MVP) that will transition into a fully-fledged product. Don’t be afraid to talk about the potential risks behind your solution. Take responsibility and become the co-author, even if the process might get ugly.
An inventory management system that we developed for our client is yet another example of this approach in real life. After the initial workshops, we came up with new business processes and a detailed roadmap, which led us to the development of a working MVP. Later on, we transformed it into a complex product that optimized stock levels and reduced overhead costs. That’s how a seemingly focused project naturally transitioned into something very different for a greater benefit for the client.
Final thoughts
Embracing the AI Sprint approach can significantly enhance the discovery and development of AI solutions tailored to your business needs. The collaborative, flexible, and thorough methodology allows for a deep understanding of the project’s true goals and user needs.
By focusing on the right questions, leveraging diverse expertise, and staying open to non-technological solutions, you can drive innovation and achieve impactful results. Remember, the AI Sprint is not just about finding a quick fix but about laying a strong foundation for sustainable success and growth in your AI initiatives.
Published July 30, 2024
