Closed Bug 758731 Opened 10 years ago Closed 8 years ago
[AAQ Flow] Reduce the steps to post a question
At the moment the steps to post a questions in the AAQ flow, since the user tell us what's their issue are as follow (1 page per step): - User adds it's question and searches. They read through the results and tell us that their questions is not covered. - If they are not logged, the usual usecase, they need to give us their details. - They fill the rest of the information and they send the question. After that they need to check their inbox (if they are no already registered as it's the main usecase) and validate it. This last steps is completed by 95% of the users. With the actual process, since a user clicks "None of this solve my problem" more than 50% of them don't end up completing the process. The 2 hypothesis are: reluctant to share email information/registering and the second one that we want to cover with this bug is that the process is too long (i.e. another step, yet another step...) Can we aggregate the "registration/login" page with the page where users introduce the details of their question? The goal of this bug is to validate the second hypothesis, so anything that works towards it is valid.
On the 2nd page of this proposed Ask A Question flow, the registration form (containing email, password and username) appear right after the question form. We think that this will make a lot less people bounce at the end of the form. We also explain why we need the user’s real email address, and the fact that their post will appear for the public to see. We think that this will also help users enter a real email, and increase their success in getting help.
I am assuming that stuff like "98% of questions receive answers. Average time for first response: 5 hours." isn't part of this bug. We are just combining the registration/signin page with the aaq form? Also, I don't see any reference to the case where the user already has an account. What would they do? This change is going to be complicated on our end, we need more information. Then we can do some research for estimating. I am bumping this out of the sprint until this is resolved.
Whiteboard: u=user c=questions p=
Target Milestone: 2012.11 → 2012Q2
(In reply to Ricky Rosario [:rrosario, :r1cky] from comment #2) You are correct. We don’t need the 98% or 5 hours figure. We are just combining the registration step with the AAQ form. In the case where the user already has an account, we simply eliminate the username/password. In a moment, I will upload an example based on our existing design.
AAQ with registration, applied to existing SUMO design.
Attachment #628417 - Attachment is obsolete: true
Bram, where do the users who have an account but are not registered confirm their identity? A possibility is to change the order to: - Username - Password - Email: Only if you are not registered. I can think of an example that uses this method so it may have a usability problem: i.e. users not being sure if they need to fill the email or not, or new users not adding an email, etc. Also, nitpicking, isn't better to use smaller boxes for Username and Password so they can identified quickly with short words vs longer text in the boxes above?
(In reply to :ibai from comment #5) > Bram, where do the users who have an account but are not registered confirm > their identity? I hadn’t thought of this possibility. I am hoping that users who ask a question would either be: 1) New: so we present them with the registration step 2) Registered and logged in: so we don’t present the registration step But you presented a new option, which is: 3) Registered, but not logged in Rather than making one field optional, I propose making a radio box selection that will trigger different set of fields to appear: 1) This is my first time asking a question on Mozilla Support. (The form would display username, password, email, like a registration form) 2) I have asked a question before. (The form would display username and password, like a login form)
This is a revision of the AAQ flow that resulted from our prototype test. It has been tweaked again from what you might have seen on the Phase 3 presentation yesterday. What’s changed in this version: * The steps are broken down into just four: write question, review question, confirm email, and done * Product and topic detection are done after the user asked his/her question, not beforehand. We think we can do some query source analysis and determine this. It provides a very smooth and expected user experience. But it might be very hard or developmentally expensive to do. Here’s a wild proposal: we eliminate the product or topic entirely, and just focus the process on just asking a question. This proposal only works if we assume that navigation and search are working so well, that user who wants to Ask A Question does not need to be bothered with product or topic selection. * Lots of just-in-time (JIT) help are provided to alleviate user’s fears and concerns: ** p1: “What does this mean?” will explain how the forum works ** p1: “See the forum to see how it works” will show how the forum works ** p1: The help on the right of the “Create an Account” form is not shown ** p3: “Why do I need to confirm my email?” ** p3: “Using a webmail? Log in below” will make it easy for the user to confirm ** p4: “See your question posted here” will alleviate the user’s fear that his question is gone somewhere he cannot see ** p4: “While you wait for your answer, see if you can help” is a nice advertisement to recruit contributor. I am hoping that, after we offer the user lots of help in asking his question, he’s willing to help us out a bit
Attachment #630182 - Attachment is obsolete: true
This is an interesting solution I found when asking for help on Distrify.com Rather than putting the possible answer before asking a question, they put it after you ask your question. It’s an interesting proposal. I’ll have to think about its merit some more before commenting on it. The headline reads: “We have not posted your email yet. Perhaps one of these articles will help you get your answer…”
This is something I've seen and tried before. Some considerations: - Search using all the verbose of the question doesn't work as good. While a short question title works good enough as a search query, a paragraph doesn't. Because of that we are better using the title than the whole question. - We are asking the user to write something that doesn't get used. - After having written the whole question, users have a higher tendency to skip the links. And answering the comment 7: Product and topic selection help users formulate better questions. In the process of selecting the topic we are introducing them the jargon that we and our community use. It's not only about tagging questions, is about showing them categories (topics) and links to articles that give context for a better answer. If we provide an open box without previous context we will end up getting lower quality questions that force us to go back and forth with the user instead of having all the context after the first post and having the ability to give the solution in the first answer.
I’ve added the topic selection, but kept the search field away. As Ibai had written on comment 9: (In reply to :ibai from comment #9) > - Search using all the verbose of the question doesn't work as good. While a > short question title works good enough as a search query, a paragraph > doesn't. Because of that we are better using the title than the whole > question. > > Product and topic selection help users formulate better questions. In the > process of selecting the topic we are introducing them the jargon that we > and our community use. It's not only about tagging questions, is about > showing them categories (topics) and links to articles that give context for > a better answer. I think that topic selection helps, but letting the user type his question into a search box would probably end in more failures than successes, so I have removed the search interface. I agree that we should implement some sort of a way to link to articles, and think that picking a topic is the perfect way to do this. Rather than being a simple listing of all articles under a topic, we will display the top 5 articles that are sorted by popularity. This top 5 list could be manually populated by us, if we think that most users coming to AAQ are having certain issues. The biggest difference you will notice here is the use of the word “Instant answer” rather than “Narrow down your question” or “Select a topic”. I think the wording switcharound from something that implies “I have to do more work to ask” to “getting a quick fix to my problem”, is going to be very encouraging to the user. It may cause people to complain less about having to pick a topic first before moving forward: which is something we observed in our prototype test.
Attachment #633638 - Attachment is obsolete: true
Moving 2012Q2 bugs to Future.
Target Milestone: 2012Q2 → Future
Revamping this bug with the new visual style. Taking Bram's proposal and adding comments in the attached file. Main takeaways: * Opportunity to reduce one step. * Change the position of the registration to the first: ** Pros: At the moment it shows up in the middle of the process, disrupting the flow completely ** Cons: Users don't have the opportunity to see information "for free". Although, they have the KB and forum for that. The main goal of the AAQ is to ask questions.
Ibai, thanks for the feedbacks! I was able to make all the UI changes you suggested, with the exception of the one about putting the registration step first. Bug 762255 covers only the reskinning of the AAQ interface, so I don’t plan to redesign the flow there.
Great! Let's keep that conversation here.
There was a recent attempt to revamp the AAQ and we reduced the number of steps needed to ask a question. Any new development deserves a new bug.
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.