Content Llama
Content management
automatization app
Content Llama provides the power of AI to manage
and generate product image sequences for retailers.
Timeframe: Feb-Sept 2022.
The team involved: Product Owner, Stakeholders, Delivery Manager, Devs, 2 UX Designers, Engineers, Data scientist, Solution Architect.
My role: UX/UI Designer (as part of an agency).
Platforms: Desktop.
Constraints: The project was first oriented as client service, but later was upgraded in a new business pitch. Investors' pressure.
Category: eCommerce.
Tools: Figma, FigJam, Zoom, Photoshop.
When browsing on different eCommerce platforms, users often don't realise the amount of work, primarily manual, that is involved behind the scenes to keep all the catalogues updated and standardised with the proper information and photos.
Content Llama was created to assist these Content manager roles that performed tasks in the background.
The main issue these Content Managers dealt with daily was the massive manual work. Content Llama founders' goal was to simplify Content Managers' lives by automating their working routines. Before us, they worked with other agencies to dive into this automatisation process, but they couldn't tackle it with the right approach as it was a complex problem.
As a first approach to understanding the big picture of the problem, we ran user interviews among 16 companies (22 interviewees). We needed to know and comprehend the relationships and interactions between the brands and the retailers and figure out how they measured success. They were asked about their workload, KPIs, roles, tasks performed, and their relationships with other colleagues during their daily routines.
User interviews' sneak peek: The interviews focused on getting more understanding of the relations and interactions between brands and retailers, including workload, KPIs, roles, tasks, and relationships with colleagues (with particular attention to frustrations and pain points in these areas). Exchange, export and checking the info were some of the tasks manually done.
Sorting and analysing all the information from the interviews also helped me translate similar and outstanding insights into a content manager persona.
User interviews' sneak peek: The interviews focused on getting more understanding of the relations and interactions between brands and retailers, including workload, KPIs, roles, tasks, and relationships with colleagues (with particular attention to frustrations and pain points in these areas). Exchange, export and checking the info were some of the tasks manually done.
The creation of the Persona helped me to know a bit more about the Content Managers, and to understand the pains and frustrations they faced every day (e.g. exchanging, exporting and checking the info were some of the tasks manually done; dealing internally with getting the information from the retailers was a nightmare).
The UX research revealed numerous frustrations and pain points experienced by Content Managers and provided valuable insights, but it also highlighted the necessity for a more in-depth investigation and better organisation of the findings as the insights were too broad. To delve deeper into the user's journey and gain a comprehensive understanding of each pain and frustration, I conducted a second research using a User Experience Map exercise.
The User Experience Map allowed us to meticulously chart every stage of a Content Manager's working process. Unlike the first research, which gave us an overview of the issues faced by users, this second research provided specific details on when and why these pains were arising during the user journey. By identifying these exact moments, we were able to obtain crucial guidance on where an AI-powered automated process could be implemented to improve the overall user experience and minimise those pains.
The necessary steps for effective content management were established in this User Experience Map. The steps start with recognising the need to update content on the company's website and culminate in the thriving online publication of that content. However, manual checking for missing information and updates can be frustrating tasks.
These are the main pains, gains, opportunities and problems gathered from the User Experience Map at a glance:
The chart presents a summary of the current workflow's pains, gains, opportunities, and problems, aiming to optimise processes and strengthen retailer-brand relationships for improved productivity and efficiency.
These insights (pains, gains, opportunities and problems) not only validated the necessity of an AI-powered automated process to assist Content Managers but also provided the team with a comprehensive overview of the specific phases suitable for automation and suggested the exploration of other methods/alternatives to streamline information sourcing.
This overview highlights the phases within a Content Manager's workflow that can potentially be automated and proposes exploring other methods for automating information sourcing to benefit the end user.
Also, the chart showed us that the list of pains, gains, problems and opportunities, although organised, remained extensive. In an effort to streamline and focus our efforts, we organised a workshop with the aim of pinpointing the key issue for resolution. One of the workshop's objectives was to come out with a statement about the users' main problem, which we can set as a goal for our work. Like a North Star.
I am an eCommerce Content Manager for a multi-brand retailer
I am trying to source, organise and configure the product content I need from Brands and Suppliers so I can prepare it to the specification my eCommerce site/platform needs,
but it's a slow, manual and error-prone process that is difficult to keep track off
because there are so many different sources and the product files are all different
that makes me feel overwhelmed and stressed.
The statement summarises the main problem to solve in which all the efforts may be focused. Especially in the part "I am trying to…" as "source, organise and configure" are the main tasks to be automated.
With the workflow phases in mind, and focusing on the problem highlighted in this statement, I established the most common task flow, enabling a comprehensive analysis of the problem and explaining in a more detailed way which tasks could be automated, and how this automatisation will look like.
By identifying the most common task flow, we conducted a comprehensive analysis of the problem and presented a graphical representation of how the automation of specific tasks will function. The organisation and configuration of the pictures' tasks are represented here.
The graphical representation reflects the step-by-step progression of activities that Content Llama will cover:
At this stage, Content Managers will continue manually handling the 'Collect' step, while we concurrently explore new methods for sourcing information.
The 'Sort & Map' process will occur in the background, thanks to the AI-powered automated process that matches the images with the specific requirements of each retailer, including sequence, size, format, and style. Previously, this step demanded extensive manual work, leading to inefficiencies, errors, and wasted time. Content Llama cuts this wasted time by over 50%.
The 'Approve &/or Edit' step is the one that the Content Manager should perform
The ‘Integration’ of image sequences on the retailer's website will be automated. Before Content Llama, this was done by uploading the pictures one by one manually.
This graphical representation illustrates the step-by-step progression of the different activities within the Content Llama app's workflow.
After identifying the key automatable tasks, it was crucial to create a graphical solution. The following flowchart illustrates the primary steps a Content Manager must take to onboard and approve/edit picture sequences within the Content Llama app. This flowchart offers a clear visual representation of the proposed automated process.
The flowchart played a crucial role in validating the flow and the automated actions with the Product Owner.
At this stage of the project, we assessed the available resources and time, both for the current date and the future. Employing a collaborative approach, I iteratively improved the flowcharts based on the assessment results and consequent feedback from peers and stakeholders, which informed the creation of the wireframes.
The evolution of the 'Product's List' designs throughout the project was influenced by the assessment results regarding resources and time, as well as the subsequent feedback from peers and stakeholders.
The evolution of the 'Edition’ process' designs throughout the project was influenced by the assessment results regarding resources and time, as well as the subsequent feedback from peers and stakeholders.
Also, to validate the new flow and user experience further, I ran prototypes for each main task that was meant to be performed, including 'Approve', 'Edit', and 'Reject' sequences. The prototypes bring to life our user-centred vision. Thanks to the feedback resulting from the prototype, a few more updates were included:
The introduction of an alternative to streamline information sourcing for the 'Collect' step. Going forward, Content Llama Content Managers will procure brand images instead of the end user. While this may not be considered automation in terms of the task itself, it represents automation from the end user's perspective, delivering a significant benefit through the app.
The manual upload in the Content Llama app of lists of products in Excel format was removed.
The 'Approve &/or Edit' step will no longer be done by the Content Managers but by the Content Llama team.
Interactive prototypes for the 'Approve,' 'Edit,' and 'Reject' sequences: exploring the user journey and enhancing the user experience.
With all these changes implemented, the retailers’ Content Managers will no longer be the end users of the Content Llama app, now the end users of the app will be Content Llama Content Managers.
The new graphical representation reflects the updated step-by-step progression of activities that the Content Llama app will cover, minimising the manual tasks from the point of view of the retailers’ Content Manager who from now on is the full beneficiary of the app's implementation without having to interact with it. They will only need to get the already approved image sequences from an API installed on the eCommerce platforms they work with (eg. Shopify).
This new graphical representation illustrates the updated step-by-step progression of the various activities within the Content Llama app's workflow. Going forward, the app will be used by Content Llama Content Managers, not by the retailers' Content Managers, who will only need to get the already approved image sequences from an API installed on the eCommerce platforms they work with (eg. Shopify).
In order to also achieve an aesthetically pleasing result in these wireframes, I did a digital Style Guide according to their existing branding. By adhering to it all the designers involved and the engineers could create a cohesive branding and user experience across all screens, meeting the needs and expectations of the audience.
Some excerpts from the Style Guide, that helped to maintain the brand identity and consistency throughout the designs.
See the transformation of the main wireframe screens as the application of the Style Guide on them brings the design to life.
The result of the application of the Style Guide in some of the wireframes.
Not so happy ending
While working on the different iterations and updates, the company ran out of money. It would have been ideal to have the time to continue with the following phases, such as creating a second prototype and running a usability test. After implementing the resulting improvements, the next steps will be focused on finishing the product by creating the rest of the screens for the app and all the devices and finishing the implementation in Shopify, designing the onboarding in that platform. Another round of usability tests should be run before the product release.
Cytidel
Vulnerability and risk management platform
AI-driven security platform that reduces breaches, saves time and ensures maximum protection.
Carbery
Insights AI based assistant
Exploration of a natural language interface integration into the Carbery CRM platform.