Deborah Ferreira
💻 Product manager at iFood
As one of the largest delivery platforms in the world, iFood faces a super service challenge, accounting for thousands of daily contacts from its various users. Between consumers, establishments, delivery people and companies that use our services, we need to resolve, quickly and safely, any impediment that occurs. To achieve this, we have a large service team, made up of attendants, supervisors, operations and service teams, as well as a robust technology team, which seeks to develop the best tools and provide a positive experience not only for those who are served, but also also for those who serve.
In this context, the automation front was born with the aim of increasing the delivery of the human operational team which, in addition to handling these thousands of contacts, still needed greater security to avoid manual errors and more time for humanized contact. By passing all calculations and information search to our automated solution, we increase the speed and simplicity of flows, reducing repetitive customer service work and resolving consumer problems more quickly.
The objective and challenge was scalability. We realized that, if iFood grows according to our “Big Dream”, we will increasingly need to expand our service operations to support the volume of calls, as the focus was on business growth. With this, the focus was on a “bot” that would not only have the function of providing service, but also of improving the service that my operation already provided. That's why we created Rosie — the internal name given to the chatbot persona. With the development of Rosie, the objectives changed, but the main focus always remained: improving service in general, whether through scalability (times) or through experience (humanization of processes).
The solution
To solve this, we invest in cutting-edge technology. We developed an artificial intelligence solution that understands the consumer and their report, applying the best possible solution taking into account a base of policies and procedures that guarantee the safety of all users involved in the order and employing all of this in the form of a chatbot, dialoguing with the consumer, offering all the necessary welcome but with a practicality and efficiency that delights.
Application details include:
- Automation of Repetitive Tasks: AI is used to automate tasks such as answering frequently asked questions, processing orders and solving simple problems.
- Personalization of Service: Technology is employed to personalize customer service, based on information such as customer purchasing history, preferences and demographic profile.
- Intermediary: Acts in the interpretation of interactions between users, aiming to guarantee the resolution of problems.
- Fraud Prevention: AI works to identify fraudulent activities, such as illicit orders or the use of stolen credit cards.
The solution was meticulously designed to integrate several services, ensuring ultra-fast and relevant information search for each specific customer request. More than a simple automation, it provided an organic and fluid integration with other products and service channels. This ensured that every step of the consumer journey was transparent and intuitive, regardless of the communication channel chosen.
It is worth highlighting the refinement of the chatbot which, in addition to being programmed to accurately understand the urgency and criticality of each case, was meticulously aligned with our established policies and processes. He proved to be able to conduct a friendly, resolute and efficient interaction with customers. Furthermore, we rely on continuous machine learning and the teams involved, aiming to evolve the solution and we always continue to monitor feedback from consumers who use the solution.
Team composition
Our multidisciplinary team was made up of three squads, each with a crucial specialty for the project:
- Engineering and Product — Aimed at the technical and conceptual development of the solution.
- Data Scientists — Focusing on analyzing and processing data relevant to the project.
Business Intelligence and Service / Continuous Improvement — Dedicated to business analysis and optimization of operational processes.
To approach the problem from multiple perspectives, different strategies were employed.
The first of these was related to the technical solution used, which will be detailed below. In the field of service, we have several products that make up the experience of both those who provide the service and those who are served. For our solution, we allocated a product team focused on developing tools that increased operational efficiency, integrating them with existing products and the global vision of the area.
Secondly, it was essential to merge the refined knowledge of the service operation with expertise in artificial intelligence, generating flows, decision trees and models that performed as well as, or better than, the existing processes until then.
With the team committed to creating robust solutions that integrate with our existing arsenal of tools, and aiming for a cycle of rapid experimentation and metrics evolution, we turned all our attention to an accelerated product cycle, immersed in rapid learning, and with an ambitious goal.
Technology and Implementation
Our multidisciplinary team was made up of three squads, each with a crucial specialty for the project:
The technological implementation covered several stages and even the transition from previously human validations to automated processes, also integrating new analyzes to ensure more assertive and informed decisions.
Workflow Development and Ticket Lifecycle:
- We've established a comprehensive flow that runs from the request entry tool, located in the app's help area, to our ticketing tools and internal analyst portal.
- We created an integration layer, which interacts with existing systems and with our request distribution solution, ensuring that automation receives, processes and returns demands and, when necessary, transfers tickets.
Conversational Solution and Natural Language Processing:
- We implemented a system where all service flows and content used in conversations are managed.
- This system evaluates and determines the requester's intention, defines the best subsequent approach and measures the degree of trust at each stage.
Incorporation of Artificial Intelligence:
- We introduce a core AI layer comprising multiple models for categorizing, interpreting and evaluating messages and images.
- This approach is based on evidence, considering the entire context and user reports.
Integration with Human Operations:
- The integration layer with human operators was structured into two parts: definition of flows and the service triage journey.
- Triage is performed automatically, but in specific scenarios, the dialogue from the chatbot is transferred to a human operator. All conversations and previously collected data are passed on to ensure continuity of care.
- Additionally, we provide a solution for analysts that processes and suggests the best approach for each case, based on previously provided assistance.
Impact and Benefits of Automation
The adoption of our automated solution brought significant and strategic improvements to our operation.
Comprehensive Automation:
- Initial implementation targeted certain contact reasons.
- We achieved considerable automation, covering more than 40% of total services, demonstrating the transformative potential of technology in our operation.
Financial result:
- This extensive automation translated into significant savings, with a cost reduction of more than 40% in September/23.
Increased Efficiency:
- We observed an impressive reduction of 46% in the average service time for the contact reasons that were recently reformulated.
- This increased efficiency not only provides more agile and accurate service to our customers, but also frees up our human team to focus on more complex cases.
Focus on Specific Cases:
- By reducing the workload on routine services, our team can focus on more specific challenges and provide a high-quality service experience to our customers.
The human factor
In the constant search for technological excellence, iFood demonstrates a robust commitment both to the development and implementation of cutting-edge technologies and to maintaining the human character in interactions with its users and employees. Within this commitment, we support a practice based on responsibility and ethical guidelines, ensuring that artificial intelligence (AIs) are always instruments for improving people's quality of life in different aspects, and not a mechanism that can be used in a harmful way.
For consumers, our AIs have become essential enablers, making the food ordering process faster, easier and more convenient. The system provides a significantly more personalized experience, offering restaurant and dish recommendations that precisely align with customers' interests, and ensuring a problem-solving section that is both objective and personalized.
Technology also serves as an ally for our employees, automating tasks that were previously repetitive and freeing up time so they can dedicate themselves to more strategic activities. AIs, in this context, assist and suggest responses and approaches during care and offer insights that can be used to improve operational processes and strengthen worker safety.
Amid this scenario marked by technological advances, a pertinent question arises: “Will AI replace customer service jobs?” When we look at this question from a broader perspective, we realize that, although AI has the potential to automate and improve various processes, it does not have the capacity to completely replace customer service positions.
While some experts maintain that AI could replace a considerable number of jobs, the consolidated perspective and expectation among us is that customer service jobs will be enriched and automated, but not completely eliminated. Therefore, as already observed in our operations at iFood, even with the substantial advancement of AI, the need for human interaction continues to be an essential component in customer service. We envision a future where the efficiency provided by AI will be harmoniously combined with human empathy, consolidating the key to a promising future in this sector.
What to expect in the future
With innovation and technology pulsating in its essence, iFood not only aims to follow the path of technological advancement but also maintains an unwavering commitment to focusing on people, always aiming to improve the lives of all individuals who permeate our business.
The previous year, marked by intense automation, was also a journey of deep learning. We have acquired expertise in aligning expectations with consumers and mediating negotiations between all parties, consolidating our position as efficient mediators. The direction is now directed towards the incessant search for excellence, bringing more innovation while promoting an even more humanized attention to service. We value our mission to be an essential tool in supporting the after-sales relationship with our customers, and we are resolutely committed to ensuring that each interaction is remarkably meaningful and valuable.
Navigating the Seas of Machine Learning
The journey does not end here. The future holds a range of innovations and refinements in the solutions we provide. We aim to intensify the use of Machine Learning (ML) as an intrinsic component of the artificial intelligence already embedded in our automated service system. The goal, with this progress, is to further improve the user experience, anticipating needs and personalizing interactions in order to make each contact unique and extraordinarily satisfying. This enhanced use of AM also aims to expand our ability to understand and anticipate trends, behaviors and potential obstacles, always with the aim of facilitating the customer journey and providing interactions that are primarily valuable and meaningful to them.
Also projecting its gaze towards the horizon, iFood has used AI to improve various aspects, such as optimizing the delivery process, using artificial intelligence to anticipate delivery times and prevent delays; identify consumption patterns, employing AI to decipher what consumers are looking for; and enhance the performance of images and descriptions, thus facilitating the creation of content by establishments.
Let's continue the Innovation Odyssey
The wide application of AM and AI is an intrinsic part of a panorama of continuous innovation, where we intend not only to maintain the efficiency and personalization already achieved, but also to enter new fields that these technologies can reveal. At the same time, our devotion to the team and the human element will remain steadfast and unchanging, as we discern that authentic innovation emerges from the synergy between advanced technology and human acumen. The path ahead is exciting and, armed with a talented team and solid technological solutions, we are ready to explore all the facets and opportunities that the future has to offer us.
The realization of this solution and the creation of this article are the fruits of the diligent work of a wide group of people, whose efforts turn each challenge faced into a valuable achievement. The team, together with the board, who trust and validate our work and vision, deserve a sincere thank you. Gratitude extends to everyone who has contributed and continues to offer their support, as we continue together on this path of innovation and overcoming, anticipating more learning, growth and achievements in the future that lies ahead. Thank you for being an integral part of this extraordinary journey.
