<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Silvia Spaggiari on Medium]]></title>
        <description><![CDATA[Stories by Silvia Spaggiari on Medium]]></description>
        <link>https://medium.com/@silvia.spaggiari_81539?source=rss-8427c771a1b0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*31Z4BZnjBJN9XeK1</url>
            <title>Stories by Silvia Spaggiari on Medium</title>
            <link>https://medium.com/@silvia.spaggiari_81539?source=rss-8427c771a1b0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 00:23:23 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@silvia.spaggiari_81539/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Applying Design Thinking to develop a successful Smart Assistant: a Subito case study]]></title>
            <link>https://medium.com/@silvia.spaggiari_81539/applying-design-thinking-to-develop-a-successful-smart-assistant-a-subito-case-study-3a5eb43e39d1?source=rss-8427c771a1b0------2</link>
            <guid isPermaLink="false">https://medium.com/p/3a5eb43e39d1</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[design-thinking]]></category>
            <dc:creator><![CDATA[Silvia Spaggiari]]></dc:creator>
            <pubDate>Mon, 14 Oct 2024 08:37:45 GMT</pubDate>
            <atom:updated>2024-10-14T08:37:45.707Z</atom:updated>
            <content:encoded><![CDATA[<h4>An overview of the process and methodology applied to achieve a successful, user-centric solution.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*t--EDsGb2KrRQx13LITvuw.png" /></figure><p>This case study starts with an overview of the business needs that motivate the project, to focus then on the design activities, with particular emphasis on <strong>the application of Design Thinking to reach the final solution</strong>. The approach described is not universal, but must be adapted based on specific project requirements, goals, and context.</p><p>The intent is to inspire other designers in creating user-centric conversational interfaces, showing how careful consideration of user needs and a structured methodology can lead to innovative and effective solutions.</p><h3>Project context</h3><p>In the dynamic world of digital products, innovation is essential to meet the ever-growing expectations of users and stand out from the competition.</p><p>At <a href="https://info.subito.it/about.htm">Subito</a>, we are constantly striving to fulfill our promise:</p><blockquote>To become the only go-to platform for anyone who wants to buy and sell second-hand items anywhere in Italy with total security and full respect for the circular economy</blockquote><p>by offering and guaranteeing the best possible service.</p><p>Recently, the significant increase in online purchases has put considerable pressure on the Customer Support team, leading to an increase in support requests and growing user dissatisfaction regarding response and resolution times. This created a dual problem: <strong>the need to improve internal efficiency and the urgency to enhance customer satisfaction</strong>.</p><p>Starting from this scenario, our multidisciplinary product team led the creation of a Smart Assistant to address the main user uncertainties, particularly concerning payments and shipments. <strong>Our goal was to reduce the volume of support requests to the support team and resolve issues promptly</strong>.</p><p>At a time when Generative AI was generating considerable buzz and various demos began to emerge, there was a noticeable lack of mature, production-scale implementations and no clear consensus on how to approach it from a user perspective.</p><p>We saw this as an opportunity to leverage our expertise and integrate GenAI at the core of the product. By adopting Large Language Models (<a href="https://medium.com/@thomasczerny/an-intro-to-large-language-models-llms-41f0c802b900#:~:text=LLMs%20are%20a%20type%20of,data%20they%20are%20trained%20on.">LLMs</a>), <strong>we introduced a new way to interact with our platform: the Subito Smart Assistant</strong>.</p><p>This groundbreaking approach not only met user needs effectively but also significantly <strong>enhanced the overall experience by providing instant, personalized support and conversational interactions</strong> that closely mimic human language.</p><h3>Problem definition and strategy</h3><p>To implement our action plan, we conducted an in-depth <strong>analysis of user support requests</strong>, identifying the three main categories of issues that accounted for approximately 25% of the overall volume.</p><p>Next, we set <strong>macro objectives</strong> for this project, defining our strategy based on three fundamental pillars:</p><ol><li><strong>Provide immediate, precise, and personalized responses, avoiding generic and dispersed documentation.</strong></li><li><strong>Reduce informational requests to the Customer Support team, streamlining and making their work more efficient.</strong></li><li><strong>Improve the quality in describing contact requests, providing relevant details about transactions.</strong></li></ol><p>Following this initial analysis and definition of macro objectives, <strong>the Design team was involved, working closely with all project stakeholders</strong>: Product Manager, Product Marketer, Data Scientist, BE Engineer, FE Engineer, and last but not least, the Customer Support team.</p><h3>Methodology</h3><p>As experienced designers, but entirely new to the integration of artificial intelligence models, we strategically applied some of the best-known Design Thinking practices. This approach allowed us to understand the <a href="https://www.nngroup.com/videos/problem-solution-space/"><em>problem space</em></a> in more detail and, consequently, how to move in the <a href="https://www.nngroup.com/videos/problem-solution-space/"><em>solution space</em></a>.</p><h4>Analysis</h4><p>We started with an in-depth <strong>context analysis</strong> to understand the main problems and needs of users along their user journey. We mapped five values on a dedicated document that allowed us to create a clear and useful matrix for defining <strong>use cases</strong>:</p><ol><li><strong>User type</strong> (buyer / seller)</li><li><strong>Transaction status</strong> (physical-temporal tracking of the shipment)</li><li><strong>Context</strong> (description of the event / cause determining the transaction status)</li><li><strong>User needs</strong> (triggers regarding questions and doubts users have depending on the transaction status)</li><li><strong>Response buttons</strong> (contextual topics the Subito Smart Assistant can respond to depending on the transaction status, translated into interface buttons users can select)</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sg8hlAHeWUtomP-O6J3Ujg.png" /><figcaption>An overview of the key values mapped to identify Smart Assistant use cases, helping streamline the user experience during transactions.</figcaption></figure><p>This allowed us to <strong>outline in which transaction phases the Subito Smart Assistant could provide relevant responses</strong>, covering about 27 different use cases, excluding the multiple possible responses for each.</p><p>Finally, starting from this mapping and model, we created a <strong>user-bot interaction flow diagram</strong> to identify <strong>end-points</strong> and <strong>corner cases</strong>. We identified three <strong>main scenarios</strong>:</p><ul><li><strong>The user’s problem is resolved by the Subito Smart Assistant </strong>(Happy-Path).</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zIVfjKxF4aRt0rJjKTDB_A.png" /><figcaption>Representation of the ideal interaction flow where the Smart Assistant successfully resolves user requests.</figcaption></figure><ul><li><strong>The problem cannot be resolved by the Subito Smart Assistant, requiring a human operator intervention.</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xQAVjmSTWeF5SIIg_U1ntw.png" /><figcaption>The Smart Assistant identifies a complex issue and escalates it to the human support team for further resolution.</figcaption></figure><ul><li><strong>The problem persists, leading the user to a support page for further details.</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*92K4HMt5ICi0wLygHXWFSg.png" /><figcaption>When users have further questions about the response, the Smart Assistant provides a link to a detailed assistance page with more information.</figcaption></figure><h4>User Interface and Content Design</h4><p>With the preliminary design defined, we moved to the detailed phase, integrating Graphical User Interface (GUI) elements and content (UX Writing).</p><p>We conducted a <strong>market benchmark</strong> at the level of interactions and UX/UI design, extracting a synthesis of the most frequent patterns both in terms of <strong>content organization</strong> and <strong>conversational interface design</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nck-_LbeY2ENLV5q4gJiYA.png" /><figcaption>Summary of the most common UX and UI patterns for chatbots, used as a reference for the Smart Assistant’s design.</figcaption></figure><p>In parallel, we identified the <a href="https://medium.com/@sahu.sudarsan/conversational-design-f8efa18fd89c#:~:text=What%20is%20a,a%20human%20agent">bot persona</a>, establishing the <strong>tone of voice and personality of the Subito Smart Assistant</strong> to ensure a <strong>consistent interaction and language with the Subito brand</strong>, and defined the <strong>identity of the bot</strong>.</p><p>This process included defining a gender-neutral personality and using simple, conversational language. Specifically, the bot’s language was enriched with colloquial and empathic elements.</p><p>Finally, the bot’s language was adapted to the identified use cases, making each response useful and informative while ensuring a good user experience.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CgPZHyMdxHmSEEJXBI3F1A.png" /><figcaption>Template for creating the Smart Assistant’s persona, focusing on simple, conversational language that aligns with the brand <em>— Conversational Design Template FigJam board (</em><a href="https://www.figma.com/community/file/1126608139029167758/chatbot-or-voice-persona-conversational-design-template"><em>Source</em></a><em>).</em></figcaption></figure><p>Combining these elements shaped the user interface and overall experience, also defining a <strong>system of static messages depending on the scenario</strong>, identified during the analysis activity.</p><h4>User Testing and Feedback</h4><p>Once the design was consolidated, we validated our <strong>Minimum Viable Experience</strong> (<a href="https://medium.com/@userpilot/minimum-viable-experience-mve-what-product-managers-should-know-88688bf6ec49">MVE</a>) through <strong>user tests with real users</strong> before moving to the development phase. The feedback collected provided valuable insights to refine the design and further improve the user experience. We incorporated this feedback into the first version of the Subito Smart Assistant, collecting additional suggestions in a repository for future qualitative and quantitative studies aimed at product evolution.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TW8BMo4vEp72t1N8cbNboQ.png" /><figcaption>Key insights gathered from user tests, used to refine the design and enhance the overall experience.</figcaption></figure><h4>Ideation and Product Envisioning</h4><p>As the final step in designing the experience, we conducted a <strong>Design Envisioning</strong> session to explore <strong>potential bot evolutions</strong>. This collaborative activity involved the entire UX team with a generative workshop, ideating new functionalities and usage contexts. The ideas collected were then documented in an <a href="https://medium.com/@paulmyersmba/every-company-needs-an-ideas-bank-497919942c53#:~:text=The%20mechanics%20of,or%20prototype%20development.">Ideas Bank</a>, ready to be analyzed and integrated into the Product Backlog.</p><p>Imagining a future scenario allows us to establish a <strong>product and business direction</strong> to strive for, enabling teams to explore more complex scenarios and defining a <strong>long-term strategy</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FoQ9VkmBBQKugJJmC2A4dg.png" /><figcaption>Exploratory sketch of potential Smart Assistant evolutions, created during a Design Envisioning session.</figcaption></figure><h3>Results and Impacts</h3><p>After the first release, we already observed significant results: we were able to handle over 2500 responses daily and <strong>reduced transaction-related requests by 40%</strong>. The use of the Smart Assistant in various possible scenarios <strong>improved the overall efficiency of the Customer Support team</strong>, reducing response and problem resolution times. One of the main challenges was ensuring that the GenAI model clearly understood the information provided by users, offering accurate and prompt responses. This required a content review to make the bot’s message precise, understandable, and in natural language, creating an <strong>emotional and empathetic connection with the user</strong>.</p><h3>Iterations and continuous improvements</h3><p>Currently, the Subito Smart Assistant can respond to questions about transactions and shipments through a guided user interface. The next step will be to allow users to describe their problems in words, or even by voice, so that the model can provide the most appropriate response.</p><p>With a user-centric approach, the Subito Smart Assistant can be extended to various other usage contexts, not just related to transactions, becoming an <strong>intelligent agent</strong> capable of offering <strong>personalized and advanced suggestions</strong>, improving the entire user journey.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JSUl2VdYIoZfn6PgUZ6esg.png" /><figcaption>Concept for the future evolution of the Smart Assistant as an intelligent agent, capable of providing personalized and advanced assistance.</figcaption></figure><h3>Conclusions and reflections</h3><p>Our exceptional collaboration among various professional figures, combined with dedication, creativity, and commitment to innovation, demonstrated the importance of synergy between skills to ensure the success of a project.</p><p>Applying a <strong>user-centric approach</strong>, integrating <strong>Design Thinking methodologies</strong>, and <strong>AI design</strong> proved crucial in developing an effective and innovative Smart Assistant.</p><p>The main lesson we learned is that a collaborative and user-oriented approach can lead to highly effective and impactful solutions, continuously <strong>generating value for our end users</strong>.</p><h4>Acknowledgements</h4><p>I would like to express my gratitude to the following colleagues for their invaluable support during the design phase and for contributing to the writing of this article:</p><ul><li><strong>Donatella Pirata</strong> (Content Designer &amp; User Researcher)</li><li><strong>Erica Grandi</strong> (Sr. Product Manager)</li><li><strong>Sonia Vivace</strong> (Data Scientist)</li><li><strong>Raphael Assal </strong>(Data Scientist)</li></ul><p>Their dedication and passion made the success of this project possible.</p><p>If you are a designer passionate about <strong>AI topics applied to the world of UX</strong>, I invite you to subscribe to the “<strong>AI Goodies</strong>” newsletter by <a href="https://medium.com/u/e2e1cbef4b57">Ioana Adriana Teleanu</a>. You will receive the latest updates on resources, design trends, AI news, and the coolest design tools.</p><p>💌 Subscribe to the newsletter to stay updated and continue exploring the future of design and technology.</p><p><a href="https://aigoodies.beehiiv.com/subscribe">UX Goodies ✨</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3a5eb43e39d1" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>