You’re not the user: 5 things I learned from using Hotjar for a year
How feedback and analytics guided my design decisions.
Hotjar is an all-in-one Analytics and Feedback tool that makes it easy to truly understand your web and mobile site visitors. As the UX designer in an online travel agency, I relied majorly on Hotjar for studying user interactions, observing consumer behaviors and gathering user feedback. All these constituted to guiding my design decisions.
This article assesses how a tool like Hotjar can help in making user-centered design decisions but the intents should not be seen along the lines of a tool recommendation, as a matter of fact, any tool of similar kind should do just fine.
Back to business, these were the 5 key things I learnt from consistently doing analytics and observing users for a year:
1. Provide users with actionable error messages
Out of 500 users that got an error message, only about 30% refined their search queries further in hopes for a positive outcome.
Hotjar has a recordings feature where you can set up visitor playback to replay sessions where users exit barrier pages.
According to Jakob Nielsen’s usability heuristics for user interface design, designers should help users recognize, diagnose, and recover from errors. Error messages should be expressed in plain language (no codes), precisely indicate the problem, and constructively suggest a solution.
Our error messages were visible, but neither did it reduce the work required to fix the problem nor educate users along the way.
An example of a major error message that fell into this category was : “Sorry, no flight results was found that matches your criteria”.
I had a chat with our developers about this and they confirmed it was possible to include possible suggestions (change of return or travel date).
In the e-Commerce space, a practical example could be: instead of saying “out of stock,” your error message should either tell users when the product will be available or provide a way for users to ask to be notified when the product is restocked.
2. Eliminate optional form fields, reduce required fields
The messages field on our visa interest form had the longest time on field (> 41seconds), 42.9% blank rate and 13.3% drop off rate.
The visa interest form is for people to fill in their contact, destination country and dates they would love to travel, but we had an issue, the messages field was causing a lot of drop offs, especially in comparison with other form fields.
Here is the design problem: People spent a lot of time on the field because the placeholder reads: “Please enter your message here”.
Removing the cognitive load from the users as much possible should be a major priority of user-centered designers.
To increase form completion rates, it was necessary to collect data from the completed forms. What I found out shocked me:
Majority of people that completed the form entered vague terms into the messages field, Responses like “later”, “okay”, “thanks” flooded that field. This could be because the field was mandatory yet not well explanatory.
It was ideal for us to either make the placeholder more explanatory or resort to a drop down that will contain things that we have the business capability for (vacation, education, e.t.c). While I preferred option 2, the business decided that option 1 will create more room for them to see what people are demanding for and how possible it will be for them to include such offerings. I agreed.
Removing unnecessary fields from form fields require approval from stakeholders and insights from the customer service, but the reduced user effort and increased completion rates (time + accuracy) on the forms made it worthwhile.
Also, after watching 200 visitors recordings, of the small percentage that used the Close Form button to clear form fields, not up to 5% refilled the form. I already addressed this in my Blurring The Lines Between Usability and Security article.
3. Good search skill is not natural
Visitors had a hard time making purchase decision when flight search results were not displayed in ascending order of price.
Collectively, users spend more time on other websites than they do on our website, this becomes even more serious when we consider search.
It was discovered via visitors playback that users became stuck when they found some lower prices underneath higher prices. The big fail here was that the search results were neither linear nor prioritized. It became imperative for users to look for ways to sort the results before they found a pattern that could inform their purchase decisions.
Then the big question came:
What is the reliable threshold we need to look for to inform a search redesign decision?
The threshold for an effective search system differs across product, users, geography e.t.c but so far it has been debated to be findability. This is a metric that is quite difficult to derive from a qualitative setting. In general, I assess that big search engines like Google has logs of your redirects, how long you took on the redirected site, e.t.c to measure user satisfaction in terms of findability.
I had user interviews with about 5 customers that came to the office and the key things that informed their purchase decisions were in the order of Airline >= Price >= Stops , Duration.
The solution that we resorted to was to show 2 airlines each with the lowest prices and a show more button to view more flight results from that particular airline. Currently, I have no details about the implementation of this.
4. Reducing bounce rates is not always the answer
For several months we ran an un-intrusive Hotjar polls to identify reasons why we had users bounce off certain pages. The Poll read “If you did not make a purchase today, what was it that stopped you?”
While we got feedbacks that made sense and were actionable, some responses were plain useless. While reducing bounce rates could be a good attempt at increasing conversion, focusing solely on lowering bounce rate might end up affecting other ROI generating actions that could happen across the entire journey of a customer.
It proved better to us to determine why users were visiting in the first place than knowing why they leave on visit. We happen to sell something that is not a cheap commodity, and except we have a base figure of what the bounce rate for our industry and target audience is, optimizing for return visits could just be a better approach, and it was.
Digging deep into Google Analytics, it was discovered that the bouncing users are not one-time visitors who bounce and never return but are visitors who return frequently.
We refined our polls, and found out that bulk of these people were Affiliates that wanted to book flights for clients. So their main aim is to check for flights and give feedbacks to their respective clients. Since they have different clients and were called at different times, it became immediately apparent why the bounce rate for those individual pages were high.
5. You’re not the user and you might never be
For 3 months straight, some feedbacks we got from the polls had “Pay Later” feature as the third most requested on the list.
Since we already had on ongoing “instalmental payment” process that was utterly done offline, I worked with the travel consultants to draft wireframes on a new instalmental payment feature. Time did not permit us to test these user flows with the users, so we did what we hoped would work, which was to create a page that allowed people use the feature.
What we faced?
The onboarding process was slow and conversions resulted in dead ends. This continued until we got considerable amount of valuable feedbacks using the Hotjar Incoming Feedback feature.
Why?
The system did not prevent the users from entering inputs that did not comply with the business regulations on this feature. E.g travel dates cannot be 3 weeks further from date of booking.
Another of Jacob’s heurestic of user interface design is: error prevention (a careful design which prevents a problem from occurring in the first place is better than good error messages).
To address this scenario, we limited users in the dates they can choose. This constraint became really helpful as it forced users to pick a date range that fits the instalmental payment feature requirements.
Conclusion
As UX Designers, much of our decision-making has been based on empirical data derived via qualitative research. This data helps us make informed decisions which are free from bias. The way users interact with our products can increase or reduce our ROI- the business bottom line, so instead of focusing our time and resources on what we think will work, we can remove these guessworks by getting to know exactly how users interact with our website.
Data points are just events, they do not have a clear view of the processes behind them, a combination of feedbacks and analytics can provide the motivations behind these processes and ultimately guide our design and re-design decisions.
