The Yesser project is a national e-government program in Kingdom of Saudi Arabia. The KSA government wanted to improve the efficiency of public sector via digital transformation, and one of the objectives is to improve their open data portal in accordance with international standards to match up with the leading public data hubs.
Project Context
-
The site would be a one-stop shop for accessing all the available open data resources
-
The authorized publishers are enabled to create and upload their own datasets
-
The end users will be able to view data resources and use them for their own cases
Project Goals
Role
Date
Scope
Mission
UX Designer
UX Researcher
Apr - Sep, 2018
Desk Research
Design Review
User Interviews
User surveys
Wireframing
To assist KSA government's dev team to conduct UX research on open government data portals
Saudi Open Data Portal
Saudi open data portal is aiming to create a one-stop shop where the government agencies can upload their own datasets and where the end users can view and export these open data resources.
Phase 1: Defining The Users
Open government data portals have multiple audiences and each category has different goals to achieve when using the site. By clarifying the main user groups, we are able to collect more accurate user feedback for the portal development.
Overview
Activities
Goals
Literature review on user groups of ODPs
1. To clarify the major categories of the key users;
2. To understand the contexts when they are using open data portals;
3. To define the recruiting criteria for the next user research phase.
Deliverable
Research report
The ODP users have 4 main categories
Findings
-
Entrepreneurial Users
This group includes tech companies/tech startups, or development agents who rely on open data to build their products and services, and supply them to citizens or government sectors.
-
Government Agencies
Around one third of site visits came from within the government itself. ODPs increase efficiency in data flow between various departments, enhance data quality by reducing multiple copies, and reduce cost of data exchange & integration.
-
Social Users
Civic hackers, citizen scientists, advocacy groups, data journalists, data scientists fall within this group. This kind of users volunteer their time to their chosen public interest activities, or their works are supported by NGOs.
-
Academic Users
Researchers, data professionals, and students who study data management are classified into this group. They might be assigned to projects that need to work with open data, e.g. to carry out an analysis or make a visualization.
The 6 best features of the 3 reviewed data portals
Finding 02
Swipe Left To See More Features
Singapore’s public data portal categorizes all of their datasets into 9 different groups. They are presented as an array of icons on the homepage and served as a menu to help users navigate all the datasets through the site.
These iconic menu are straightforward, easy-to-follow, and highlighted in orange. Both first-time visitors and proficient users can see exactly where to go next and where they are now without getting lost.
Grouping data inventory also saves users from being overwhelmed by the content when first time landing the site. Users are visually guided to search within a certain topic.
1. Iconic Navigation
2. Interactive Charts/Maps
Both Open Nepal and Singapore’s public data portal display their feature datasets in the form of interactive charts/maps. These charts are colour-coded and responsive. When users hover over any points in the graphs, a related annotation shows up.
Site that complements the user journey with functional interaction drives more engagement and increases the time users spend on a website.
The charts on Open Nepal were created by a third-party web app called Datawrapper. It also exemplifies how to visualize data on ODPs and produce meaningful results in seconds.
3. Advanced Search
Canada’s open government portal offers advanced search options, which are easily accessible in the search results page. 8 different parameters, like organizations, subjects, formats etc., can be selected and applied to narrow down the scope.
An information-heavy site, whose success depends on content findability, should consider the search functionality of the website carefully, like giving users suggestions for limiting the scope if a search returns too many results.
4. Use Cases
Singapore’s public data portal has an official blog on Medium. The blog showcases inspiring examples of how to integrate open data into technology tools that make people’s life easier, and how to combine open data to create new knowledge and insights.
Similarly, Open Nepal and Canada’s open government portal have the Featured App/App Gallery pages, where they display a bunch of apps developed to address real-life problems using the datasets on the site.
Use cases gives us precise ideas about how users can make use of the product to achieve their goals successfully. They help communicate lingos in an easy-to-understand way, which in turn help promote and convert.
5. Users Toolkits
Singapore’s public data portal provides digital journalists with the embedding codes to use interactive charts in their chosen news media. As for developers, there is a page lists all the real-time API datasets and offer a “subscribe” option to receive updates on resources.
Open Nepal allows developers to explore datasets on GitHub, and Singapore’s public data portal offers them ready-to-use codes in the code editor.
It shows these portals have deep understanding of their users’ work processes, and are able to support users with related toolkits to complete their tasks faster.
6. Online Community
Open Nepal and Canada’s open government portal are more of an online community when compared to Singapore’s public data portal. They convey this image successfully by:
A. Having a very clear proposition for responsibility to promote data revolution. They lists citizen initiatives that related to open data and government transparency submitted by users.
B. Inviting users to generate new contents on the site by allowing users to rate datasets, comment or provide feedback, suggest the datasets they want, upvote/downvote the suggestions, and submit apps they develop using the datasets.
C. Creating/listing both online and offline events to encourage users to get involved, collaborate, and network, such events includes app development competitions and open data conferences.
D. Supporting the community with a wide range of accessible online learning resources.
Online community provides a place for people who share the common passion and interest to come together. It helps form connections with users and keep them engaged, thus they are more likely to return and recommend the website.
Swipe Left To See More Features
Phase 2: Benchmarking
City of Melbourne. (2018). Melbourne Data: About Melbourne Data. Retrieve from https://data.melbourne.vic.gov.au/about
European Data Portal. (2015). Creating Value through Open Data: Study on the Impact of Re-use of Public Data Resources. Doi: 10.2759/328101
European Data Portal. (2018). Open Data in A Nutshell. Retrieve from https://www.europeandataportal.eu/en/providing-data/goldbook/open-data- nutshell
Middleton, T. (2011). Greater Manchester – Open Data City. Retrieve from https://blog.okfn.org/2011/08/25/greater-manchester-open-data-city
Nessi, G. (2016). Open Data for The People. AUS: Link Digital.
Rogers, K. (2015). Improving government access to government data. Open Data Handbook. Retrieve from http://opendatahandbook.org/value-stories/en/improving- gov-access
Yesser Government. (2018). The e-Government Program: Open Data Portal Project.
This phase is to evaluate world-leading the Open Data Hubs to understand their strengths & weaknesses, so as to determine opportunities to innovate with our client's ODP. I took a look at Singapore's public data portal, Open Nepal, and Canada's open government portal, and analyzed their sites with UX/UI design guidelines, such as 10 rules of thumb, etc..
Overview
Activities
Goals
Design review on competitors' ODPs
To give the client insight into:
1. how to design a successful open data portal and what are the dos and don'ts;
2. how to balance functionalities and ease of use of an open data portal;
3. how to design effective user flows for target users.
Deliverable
Research report
The fundamental structure of ODPs
Finding 01
-
Homepage with search box
-
Search results page with sort and filters
-
Dataset detail page with relevant specs
where users enter keywords in the input field to request for certain datasets
where users refine their results in the sidebar or launch a new search
where users view, download or interact with a specific dataset
Useful data insights
Users gain a sense of satisfaction when they are enabled to generate meaningful knowledge or discover useful patterns out of the rich data.
1
“One day, one person gave us his e-commerce datasets from his business, so we played with it to see if he has made any mistakes in his business. We discovered so many issues in his data. Like he had some issues with the delivery process. In recent months, it’s one of the most interesting experience for me.”
An engaging community
Users spend more time on a platform propels a two-way communication. They are more likely to return to where they can form connections with other users regarding common interests.
2
“We need a website so that people can work inside the work. There’s a website called Data World doing this now. It’s like, people can comment on the dataset, people can work together on the same datasets, or people can share their insights about the datasets. Not just download the datasets. If I have any questions, I can discuss about it with others.”
Data visualization tools
For users who don’t have any analytical software to process the datasets, built-in visualization tools can make the data more accessible for them and increase users’ interest of the ODP.
3
“Don’t share only the Excel and PDFs, try to give some basic understanding of the data in the form of visuals. Recently, I have come across a really good visualization tool that you can embed in your website or data portal. If the tools become part of the website, that would be good.”
Top view
datasets list
Browsing an ODP can be overwhelming. Too many contents can even deter visitors. A list of popular items on the homepage gives users some direction and provide them a quick access to feature datasets on the site.
4
“Finally, let’s add download count and the view count for the dataset. For example, I do remember on one of the portals, when I first time landed on the site, I chose the top-view datasets. It has a list of five to ten. That would be good to have, it would induce a lot of people open it, then draw more viewers.”
Chat bot
system
A Chat bot system is able to respond to users’ questions instantly 24/7, and help users find or sort things they demand immediately, so users don’t have to wait for help, which in turn optimizes their experience of the portal.
5
“I hope to see some smart tools like a chat bot agent in the portal. For example, if I want the datasets of the regional population in Saudi Arabia, or want to know about any insights about it, I can go to the portal and there is a chat window. I ask that question, and the portal will answer me, then I get the dataset I want directly without downloading a load of other datasets.”
Delighters
Delighters are the excitement features that creates dramatic customer delight and positive response to a product. Delighters for users of an ODP includes:
Phase 3: Discovering The Problems
I set up one-on-one remote interviews with 10 participants from Saudi Arabia in July, 2018. The participants were recruited by Master Works. After that, I conducted thematic analysis on a bunch of interview data to inform some meaningful insights. This phase is to learn about users' objectives and what isn't working for them.
Overview
Activities
Goals
User Interviews (semi-structured)
1. To identify the users' pain points in ODPs and their ideas for improvement;
2. To inform personas, user flows and feature ideas of the product;
3. To synthesize findings and make actionable recommendation for the next phase.
Deliverable
Questions script & Interviews report
What are the problems?
Finding 01
After analysing the interview transcripts , I recognized some reoccurring patterns. The data with similar labels were grouped together to form 4 major themes, which are Relationship With Data, Data Credibility, Frustrations, and Delighters. This section mainly talks about the Frustrations theme which related to the most common issues of initial Saudi ODP.
User journey mapping
Personas
The research findings were translated into two personas, Muhammad, the data analyst, and Akilah, the undergraduate / part-time software developer. I mapped out Muhammad and Akilah's scenarios with their thoughts and emotions as user journeys to identify their pains that need to be addressed.
Frustrations
01.
Data quality issue
Users regarded most files on the initial Saudi ODP were just some generic / aggregated statistics, and aggregated data is hard to analyze, thus they are of limited use to users. Plus the datasets were not provided in machine-friendly formats for processing.
-
Only aggregated statistics
“It’s still a little bit aggregated. There aren’t a lot of details to extract any interesting insights or knowledge from. I’d say that’s my major concern with the current data portal.”
“I’d like the dataset in lowest level, not to be aggregated, you know, what I mean just raw data. I just need the raw data without any aggregation or statistics.”
-
Limited downloadable formats
“Here we only have Excel, PDF or JPG, it’s readable for people, but it’s not good for me. I only care about the data my application can understand, give me XML or CSV, otherwise, I have to do the work and try to make my application understand it.”
“They don’t have any API. Usually. When there’s a huge dataset, you’re not actually screening through it, you’re not going to check it through on your own, you want to use it in an actual application.”
02.
Obscure dataset info
Most datasets on the initial Saudi ODP did not provide sufficient details on how such statistics were generated. Without full spec sheets or very detailed metadata about a dataset, users have to spend extra time to do investigation before processing it.
-
Lack of detailed metadata
"I only see datasets, but there is no description about it, and there is no any information about the fields, so I don’t know what this field means. Also, I need to know the background information. I want to know about what is the process and where it is come from.”
“Of course the date, like when was it published, is it getting updated recently? Yearly updated or just one time uploaded.”
-
No effective way to contact publishers
“If the actual datasets would have a contact with the owner, telling who posted the datasets.”
“I want to be able to ask the publisher direct questions. If there’s something not clear, I would like a channel to contact the publishers.”
03.
Poor dataset
discoverability
The search experience of the initial Saudi ODP is not intuitive, from putting in a search query to refining the search results with filters, from checking the preview to evaluating a dataset’s usefulness on an item page. Poor navigation leads to frustration and obstacles to use.
-
Poor information architecture
“It’s not very organized, but the US one, it’s really organized, and the search capability is really good, I can search anything.”
“The hierarchy of the data, for example, bigger category, then subcategory, then sub sub-category. This should make it easy for me to search it.”
-
Lack of desirable filters
“If I want some specific datasets during a certain period, they didn’t have a filter for that, It would be great to have a filter for the specific year, from 2012 to 2013, or to 2018.”
“A filter to filter the data types, a filter to see the time range of the data, or at least put it in the file name, so that I know if it’s the data I am looking for.”
-
Poor presentation of info
“If this is an outdated data that no one is checking it out, and it's not accurate anymore, why it still returns in the search results. Let’s archive it or whatever, don’t show it to me. They distracted me from doing my tasks.”
1. Participants
A much lager percentage of Developers (40%) and Data Scientists (40%) provided answers to this questionnaire in comparison with other occupations. This means the analysis from the reported data will be heavily inclined towards
these perspectives.
2. Homepage
The familiar landing page objects which the users most expected to see on a new open data portal are Search Box (31%), Data Category Icons (26%), and Navigation Menu (20%), followed by Brief Introduction (17%) and Latest Added Datasets (6%).
3. Search Filters
The 5 most favored dataset search filters, from highest ranked to lowest, are Year Range (35%), Publisher Department (25%), Category (20%), File Format (17%), and Licensing Type (3%).
4. Preview
Chart 4 shows what is the most useful preview information for the users to weigh the relevancy of search results. The top 5 wanted preview info items are Short Description (50%), Publisher Department (21%), File Format (13%), Release Date (8%), and Title (8%).
5. Detail Page
Chart 5 shows what is the most important detailed information for the users to evaluate datasets as fit to their purposes. The most wanted detailed info item is Background Information (14%), closely followed by Last Updated Date (12%), Timespan Covered (12%), Data Collection Method (10%), and Data License (10%). The other options were shared nearly equally among the participants.
6. Productivity
Chart 6 shows the elements that can improve the productivity of the users' work process. The features with highest priority are API (18%), Similar Datasets (16%), Saving Datasets (15%), and the ability to get codes to embed charts on other websites. The other options were shared nearly equally among the participants.
7. Dataset Interaction
Chart 7 shows in which way the users most want to interact with the datasets. The option to rank the datasets was considered the highest priority, followed by the ability to report mistakes in datasets and to add comments to datasets. One interesting idea in the additional comments was an option to submit their own use cases for the dataset.
8. Obtaining Datasets
Chart 8 shows the users' most preferred methods of requesting datasets which are not available within the portal. 50% of the users desired a method of requesting/suggesting the dataset from the publisher. The other two options were relatively equal but much less requested.
9. Support Service
Chart 9 shows general trends that related to the users' preferences for support services. Live Chat (30%) and Email (30%) were shared relatively equally among the users. Interesting, a place that allows the users to post their questions and to exchange ideas with experienced people were desired by 20% of the replies.
10. User Engagement
Chart 10 shows the potential portal features that can be valuable in achieving higher user retention. Use cases of the datasets or relevant data stories were of great interest to 75% of the survey population. The other features that promote repeat visit and exploration are Learning Resources (15%) and Events for Users (10%).
Swipe Left To See More Findings
Hover/Tap To See More Insights
Publishers
-
Lack of clear suggestion on acceptable dataset formats
“Is there anyone to check the data quality and suitability there? I want to know what are the criteria for uploading the datasets, how to upload it and what formats are acceptable.”
“I’m not sure whether they are checking the data that are actually being uploaded. It needs a criteria to show in which way the dataset should be formatted, and a statement to tell if this data ok to be upload for API or is it just for viewing.”
Frustrations
The initial publisher dashboard makes it difficult to either upload datasets or manage the uploaded files. Publishers often got lost in the uploading process, felt unconfident about the formats of uploaded datasets, and were unable to get any help from the portal regarding the issue at hand.
Delighters
Delighters are the excitement features that creates dramatic customer delight and positive response to a product. Delighters for the publishers of the ODP includes:
Single sign-on
Saudi government agencies manage their data across multiple platforms, the ability to log in with a single ID and password to any related systems will minimize the transferring time.
1
“Now I have different usernames and passwords. One account for the Yesser government website and another account for the open data portal. It’s good to be just a single sign-on, so I can use only one account to log into open data portal and Yesser government website.”
Datasets synchronization
Saudi government agencies manage their data across multiple platforms, the ability to sync up all the datasets between these sites will minimize the transferring time.
2
“We’re in partnership with the United Nation. Because of this, we have to upload the datasets twice, on both our website and the open data portal. It’ll be good to upload the data in one place and then the changes are synced up to our website.”
-
Confusing uploading process
“I have to go through a lot of things to upload one single file. The steps should be short and clear... and I got lost trying to see if it was uploaded correctly.”
“Once I uploaded the file twice, so I’m trying to delete it, I could not. I have to send an email to the technical support, so they could delete one of my files.”
-
Lack of direct technical support
“No one is supporting the portal. They’re running, but no one can be found on that. We get help in our own way. We knew someone in the other agency, so we called them and he gave us all the help, but we cannot get help from that portal. There’s no live chat or something like that.”
Stakeholders perspective
Finding 02
Although the primary focus of the interviews is the end users, it was also important to get some insights from the data owners and the portal admins. Some key point of view from publishers and administrators are listed below.
Admins
-
Lack of visual indicators in the report page
“In the status report page, it’s just the table with errors and blogs. We need a visual indicator, so I could know how many errors and warnings at a glance when I open the site, I don’t have to speculate.”
“When a user tries to log in thousands of times, there should be a way to visually show the failed login attempts. Now I’m not even sure whether the users or we did it. We need a visual interface for that.”
Frustrations
The initial admin dashboard has neither an intuitive process for portal curators to create/verify publisher accounts nor the visual indicators regarding the health of portal itself, which caused huge inconvenience for admins to manage the portal.
Delighters
Delighters are the excitement features that creates dramatic customer delight and positive response to a product. Delighters for the admins of the ODP includes:
Advanced log filters
An administrator dashboard system with advanced log filters to better inspect the activities from different user groups.
1
“We need a better way than the current way to filter the logs. I want to be able to review the logs regarding people adding data/removing data, and be able to track the activity logs of publisher roles instead of just admins roles.”
Ability to manage SEO & security
An administrator dashboard system with a visual interface to better manage the SEO and web security.
2
“Just so you know in Drupal, the final upload setting is attached to the PHP file setting. It would be great to have an interface to alter the file size and robots.txt, and a way to control users’ session duration for security purposes. I think these’re basically just missing from the site.”
-
Unintuitive request-approval process
“I couldn’t assign a user to be a certain publisher in one step. I have to create the publisher name, then create a group for that publisher, then create an account for the portal, then go notify the users. The process takes time just to add a user as a publisher. I want to do this in one step or multiple stages in a flow.”
“Currently, the publishers can’t create their own accounts. They have to request an account from us through e-email. It would be easier if there’s a place on the portal where the publishers can submit their requests. So as an admin, I just go to the portal, check out the requests, verify them, and notify the users right away from there.”
Phase 4: Testing The Feature Ideas
After the qualitative research, I designed an online survey to test and validate my findings and assumptions. The survey consists of 11 questions to assess users' preferences and expectations regarding open data portals. The questionnaire was given to respondents from July 2018 to August 2018. Participants were recruited in Saudi Arabia by Master Works.
Overview
Activities
Goals
Quantitative online surveys
1. To collect quantitative data on the most important features based on users' preferences;
2. To understand how we can help users to accomplish their tasks on the site more effectively;
3. To be better able to design an ODP that is accessible, ease of use, and serves users' needs & goals.
Deliverable
Questionnaire & Surveys report
Users' Preferences For The ODP Functionalities
Findings
Phase 5: Ideation
After being inspired by the users' problems through gathering insights and data, now I've got two personas in place to design for. This phase is to spark of ideas and develop solutions to problems that affect the users' experience of Saudi open data portal. I kept the key requirements in mind while designing, which are:
-
Good search capability
-
Detailed metadata & work-compatible file formats
-
Ability to evaluate datasets faster
-
A direct way to contact data owners
-
Compelling data insights & visualizations
Overview
Activities
Goals
Idea generation & Concept development
1. To generate ideas as potential solutions for the defined and articulated problems;
2. To select the best and most practical ones and to develop them into concept;
3. To determine the ODP's key features and MVP objectives.
Deliverable
User flows, Information architecture & Wireframe sketches
User flow diagram
Ideation
With an understanding of the user needs, I started by developing a series of user flows to map out how ideally the prototypical users could navigate through the Saudi open data portal to complete their tasks. The diagram below is generalized to illustrate the basic paths taken on the ODP to achieve the user goals.
Wireframing
Ideation
I used the user flows as a skeleton and sketch up the wireframes for each needed screen. Note that only the low fidelity clickable wireframes were hand off to the client Master Works. The high fidelity wireframes shown below are a refined version based on the initial low-fi screens for better presentation.
1. Improved Information Architecture
Easier navigation with more than one destination guides on the homepage
For users who have clear keywords in mind, a prominent search bar on the home page with the help text prompts them to search through the data portal. An array of icons that visually categorizes datasets into different groups will guide users to navigate across the topic-specific datasets. The most popular items with view and download count, as well as the latest added items are listed on the homepage, purposefully intriguing users to check out the feature datasets on the portal.
Enhanced search functionality with the most useful filters and preview
The 4 most required filter parameters, Year range, File format, Category, and Publisher Department are made available on the left sidebar of the search results page, making the search the most relevant to users’ intention and to minimize pagination. The preview info matters the most, including short description, publisher, release date and available file formats are presented in the results list, accelerating users’ comparing process.
2. Smoother Interaction With Datasets
Better presentation of essential dataset information
The layout of the page are optimized for users’ browsing habit and evaluating process. Users get a brief introduction of the dataset at the top, and a range of file formats that are available for download in the middle. As they scroll down, a detailed metadata sheet is provided at the lower part for further inspection.
Assist users in interacting with datasets effectively
Prior to downloading, users are able to preview the data in an online spreadsheet, and export the metadata with just one click. Subscribe or share the dataset package as they like. If users want to view more similar data resources, a list of similar datasets on the right hand side is within easy reach.
Enable users to assess and report on data quality
To provide feedback to the data custodians, users can either rate the datasets at the top right corner regarding the quality, or share their opinions about a specific issue in the comments section at the bottom.
3. Simplified Dataset Requesting Process
When it comes time to request an unavailable dataset, instead of emailing, users are able to fill out a web form to suggest a specific government agency to open up the wanted data. After the request is submitted, it will be listed on the Request A Dataset page, where users can check their application status and upvote other peoples’ suggestions to show what are in demand.
4. More Inspiring and Engaging to Stick Around
Exhibit compelling cases of using data
The Use Cases page is a place to invite users share their data stories with uncovered insights and visually-appealing charts, a place for users to demonstrate the apps they build to tackle real-life challenges using the datasets on the portal, and a place where users get feedback from and interact with other users. An online community is naturally formed by simply enabling Comments in detail pages. The Use Cases are also featured on the homepage, aiming to drive more engagement.
Encourage users to get involved regularly
The Get Involved page lists both online and offline events for users to get together on a common topic of interest. It assists users in making new relations and connections with people who share the common passion, which in turn promotes regular returns.
Wrap Up
All the works were handed off to Master Works for usability testing by September, 2018. Even though I no longer work on this project, my ideas for moving the product forward would be:
-
Additional tests with users after the initial version is released
-
Exploring the research insights that didn't make the initial MVP
The next steps for possible future reiteration includes:
-
Building a chatbot agent that provides support to users instantly
-
Enabling built-in data visualization tools that responsive to users' chosen data
-
Refining the request approval process for "Become A Publisher"
-
Refining the review and approval process for "Publish A Dataset" that incorporates the data quality statements and metrics
-
Reiterating the design of dashboards for the end user and publisher
I’ve noticed that the MVP (data.gov.sa) was first launched in April, 2021, and Yesser government has adopted most of the solutions recommended in the initial user research reports.
Next Steps
Saudi open data portal was my first project at Link Digital after graduation. Just like everyone else would ask the same question “What is open data?”, the realm was totally new to me when I started my first job as a UX designer.
More challengingly, neither the client nor my old agency could provide a clear demographics of the target audiences of ODPs. So I got myself familiar with the domain by doing tons of desk research and then located the key user groups to start off the project.
As a one-person UX team, who worked with the client in a different time zone, it was important to make Master Works be more realistic about the project scope and timeline, and had them recruit the participants from Saudi Arabia for the research.
Way back in 2018, when working from home hasn’t been as prevalent as it is today, I was already doing the user interviews remotely via Zoom. And thank Greg & Kass, who were the open data specialists at Link, for giving some sound technical suggestions in the reports.