Have you ever wondered if your website is truly helping your visitors accomplish their goals?
It is usually challenging to find out. Often, the opinions of your coworkers, managers, and complaining customers are the ones you hear the loudest. It is difficult to find out what the average site user thinks.
One way to find out is through a “Top Tasks” survey. This approach has been used by governments, health services, universities and software companies like Microsoft. At the core, this is a survey that reveals user’s basic needs by overwhelming their “thinking brain” with choice. The purpose of this document is to help you run your own Top Tasks survey. You will learn how to set up such a survey through a step-by-step case study of how I ran such a survey at LexisNexis Canada (a company that makes software for lawyers). You will discover surprising real-world insights that go beyond other published materials on Top Tasks.
Why is a Top Tasks survey special?
This type of survey is used to discover which end goals your website visitors value the most. It goes beyond “what information do visitors want?” towards answering “what action can I help them accomplish?”.
There is a twist: in this survey, the visitor is shown about 100 different tasks and urged to choose the top 5 quickly – in this way, we get a “gut reaction” choice rather than one that they had consciously thought about.
Here is an example of a Top Tasks survey question, with a few choices:
This methodology was created by user experience consultant Gerry McGovern, and explained in his 2010 book The Stranger’s Long Neck. It is essentially a sped-up version of a common technique called Card Sorting. The importance of focusing on Top Tasks was also covered in a popular A List Apart article in 2015.
The benefits of a Top Tasks survey are:
- It helps movee your organization’s thinking away from “creating content” towards “facilitating a user’s task”. After all, reading information on a website is not what your visitors are there to do – it is just a step towards their bigger goal.
- More objective than other ways of determining visitor priorities. These other ways include HiPPO (the Highest Paid Person’s Opinion), interpreting web analytics and running focus groups.
- Top Tasks Surveys are more affordable than focus groups when surveying large numbers of people. They can be run online and require less overhead.
Can’t web analytics software tell me what my visitors want?
No. They type of content we already have on the site biases our Analytics data. What if your visitors are looking for information that is missing from your site? There is no “top 10 missing pages” report that would show you that.
Web analytics data also can’t tell us if a person succeeded in performing a task. For example, let’s say that your company sells a variety of software products. When a visitor lands on your site and views 10 different product pages – is that a good thing? Did they come to your site to get an overview of what your range of offering is? (in that case, mission accomplished!), or did they come to read more about a specific product, couldn’t find it, and had to browse around while hoping to stumble on it? (in that case, it’s a bad thing that they had to view so many extra pages). Analytics can’t answer this type of question. You must ask the visitor directly
Who else uses Top Tasks surveys?
Here are organizations that use Top Tasks surveys. Click the links to learn more about their specific experience with the methodology.
- The Scottish Parliament
- Cisco Systems
- The European Commission
- University of Glasgow
- Government of Yukon (a Canadian province)
- Government of Canada
- US Federal Reserve Board
- Microsoft (learn how they used Top Tasks to improve Excel’s Help pages by reading Gerry’s book “The Stranger’s Long Neck”)
- Denver Public Library
- Ireland’s health service (HSE)
As you can see, these surveys are popular with big organizations. They are most useful to organizations that have many document, of different types, served to people with a variety of needs. (Contrast this with a company that makes 1 product which caters to 1 type of customer – they can learn the same lessons by just having 5 “Jobs To Be Done” phone calls with clients )
Creating your own survey: an overview
Do you think that your organization could benefit from a Top Tasks survey
Creating your own survey involves the following steps
Determine which team members to involve; how comprehensive to make the survey and your approach to recruiting participants.
a list of Top Tasks
Brainstorming a large list of tasks and reducing them to a “short list” of 50 to 100.
up the survey
Creating a plan for recruiting participants, setting up your survey software and deciding on any participant perks that are needed.
Observing survey responses as they come in, learning and adjusting for unexpected challenges.
the survey and applying the insights
Turning the raw data into insights. Turning the insights into action.
Let’s see how the team at LexisNexis Canada went through these steps, and what lessons we learned along the way.
The LexisNexis challenge
LexisNexis Canada is a company that makes software for lawyers. There are many software packages on offer, and several types of lawyers that we sell to.
To meet everyone’s needs, our site had to become quite complex.
It didn’t help that the site navigation was organized around products – so visitors needed to know what product they are looking for, in advance. It was difficult to discover new products if visitors hadn’t heard of them already.
In 2017 we decided to discover the main things our website visitors wanted to accomplish. This was part of a big navigation and content redesign initiative. The redesign involved over 20 meetings with internal stakeholders, where we determined the top priorities for the website. The Top Tasks data was going to be the voice of the customer in these discussions – preventing us from focusing just on our own corporate needs.
Personally, I have wanted to run a Top Tasks survey for about 4 years. This redesign project was the first big opportunity I’ve seen to reap benefits from this kind of survey.
In the real world, you probably won’t run a Top Tasks exploration project on its own “just to improve the customer experience”.
It will be a tool you use to solve bigger problem that your organization is tackling. Here are a few types of projects that would benefit from this kind of exploration:
- Reducing the cost of customer support by making the website more “self-serve”
- Ensuring that time and money is being directed towards only the highest-value types of content
- An initiative to improve customer experience (usually will come from “customer experience” becoming a new top priority for management)
- A requirement to increase conversion rates on the site (for generating leads and sales)
For these kinds of projects, you will need to convince others that the Top Tasks methodology is powerful and appropriate. Establish the reliability of this method by referring to Gerry McGovern’s blog, book and examples of the EU Commission and other organizations’ adoption of it. Make the point that this approach isn’t risky – it’s been tried by many others.
Step 1: Planning
After getting approval to start, plan how you will tackle the project. Don’t rush into execution right away.
Whose Top Tasks do you care about?
Decide whose Top Tasks you’re going to focus on. For LexisNexis, we decided to focus on capturing the opinions of website visitors (because it was easy to do with the tools we had and represented a large portion of our actual stakeholders).
Another company could have chosen to focus on prospective customers. This might’ve required setting up pre-survey question to filter out existing customers, or to travel to an industry conference to survey a general group of potential customers in person.
For example, the University of Glasgow decided to discover 2 sets of Top Tasks, one for students and one for faculty.
How will you determine success?
Think about the value that your Top Tasks exploration will deliver… and it can’t be “a Top Tasks analysis report”!
For my project, good value-for-money looked like this:
- The production of a printed Top Tasks report. It would be distributed widely, so that these insights will not be lost on a shared network drive somewhere.
- The inclusion of the Top Tasks into every site-redesign discussion. This was going to be a solid document that I would use to support my arguments in favour of customer desires, whenever they clashed with the organization’s needs.
- The creation of the document you’re reading right now, to provide value to others outside of my organization. (This meant that I had to keep careful records and take plenty of screenshots as I ran the project).
Note something important that LexisNexis left out: we never did a “Before and after” assessment of how easy it is for people to perform key tasks on our website. This kind of benchmarking shows you whether the changes you made actually led to the desired improvement. Gerry McGovern uses a Task Performance Indicator score to do this assessment, but there are other established ways of doing this (Like this method from Jakob Nielsen). For this project, benchmarking usability improvements was too costly and time consuming.
How will you capture insights?
At LexisNexis, we decided to recruit website visitors to our survey by showing them a popup invitation (using the tool GetSiteControl). To incentivize visitors to take the survey, we offered a $5 credit to a popular coffee chain. The survey itself was going to be run through the Qualtrics tool.
To ensure a good cross section of visitors was represented, we had a pre-survey question about the type of organization that a visitor belongs to. The number of responses from each type of organization was limited. This helped us get a good cross-section of responders.
Because the Top Tasks survey was part of a bigger project, we decided to run it only for as long as it took to get about 50 responses in each of our major customer categories.
50 responses were deemed enough to give us a general idea of these visitors’ needs
Running a “pilot”
I recommend running a small-scale “pilot survey” whenever you’re giving away monetary incentives to survey participants. That’s because a pilot helps reveal errors in your setup that would cause a large loss of money.
We decided to run a 2 day “pilot” survey to discover any problems ahead of the full survey.
From just 17 pilot responses we discovered that law students were sharing the survey with their friends as an easy way to get a coffee card. This was a problem for us, because we were getting many similar responses. Read on to find out how we changed the survey to avoid this problem.
Who to involve in creating the Top Tasks list for your survey?
Your survey will ask people to pick their top 5 tasks from a long list of options. To think up good task ideas, you will have to recruit a team of people to help.
For your team, try to involve people who have contact with your clients at different stages of their relationship with your organization: sales, training, customer support, account service and billing. No need to involve more than 1 person from each organizational function.
Don’t be tempted to simply come up with a long list of tasks on your own. It’s fast to do it alone, but you’ll also miss some great insights.
At LexisNexis, I asked the following people to help think up tasks:
- Scott, Team Lead for the customer support group
- Marla, an Account Manger on the Sales team
- Jeff, a member of our University relationship team. Former head of Customer Success & Training
Step 2: creating your long list of Top Tasks
The next step is the creation of a list of tasks that you will present to survey takers. You start by brainstorming a “long list” of hundreds of tasks and then reduce to a “short list” of 50 to 100.
Here is how to do this:
Create a first list of task ideas
Alone, start compiling a list of all sorts of tasks, questions and information that a person might want to accomplish on your site. Keep it openminded and sloppy – don’t try to write them out in a consistent format.
My task ideas came from the following sources:
- The top recurring Customer Support inquiries (chances are, your Support team already has a list of “common items”)
- Web navigation links from our e-Bookstore site
- Tasks that our current website supports
- My own brainstorming
- A preliminary email to meeting participants, to
gather some ideas
(Ask people to send you their ideas directly, without carbon-copying others. You want “fresh” ideas that aren’t influenced by what others have said.)
- Navigation items from the websites of our foreign sister-companies
- Tasks from the websites of different competitors
Gerry recommends that you record the source of inspiration for each potential task. I didn’t find value in doing this. I could see that this would be useful in a situation where you must prove to someone that you did a thorough job.
Your own list might also include task ideas from:
- Questions from users/prospects on social media
- Postings on user forums
- “Free text” responses to customer satisfaction / “Net Promoter Score” surveys
- Search terms from your own website/Knowledgebase (can find them through analytics software)
- Search terms from Google Webmaster Console
- Stakeholder interviews
Group session 1: adding more task ideas
The next step is to expand your list of tasks. Set up a meeting where members of your team brainstorm together. Don’t edit down the list just yet.
Ahead of the meeting, share your first list of tasks. Share
it in a way that lets everyone see the latest version and makes it easy to add
to it during the meeting.
In our case, I put the list of tasks on a shared Google Doc. Our meeting included a remote participant. Google Docs allowed him to see our task additions in real time, and to add his own as we went.
Begin the meeting with a review of tasks you have at this point. Then, have the group discuss the different types of stakeholders that they work with – what common questions do they ask? What common problems do they have? Have the group go on a meandering conversation and take note of any website tasks that are mentioned.
At the end of the meeting we had 123 proposed tasks. Including duplicates. You should aim for at least 100.
Group session 2: reducing the list of tasks
The next step is to reduce the long list of tasks you created.
On your own, start grouping related tasks together. When done, look for duplicates. For every group of duplicates, come up with a phrase that captures all the meaning of the duplicate phrases in a succinct way. Group duplicates together in a way that still lets you see all the original task phrases – you will review them with the team.
Schedule another meeting with your Top Tasks team. This time, the goals of the meeting will be a) to spot duplicate items together (ones that you missed) and b) to review whether the team agrees with your classification of certain phrases as duplicates.
Carefully evaluate whether certain tasks are indeed duplicates, or whether they represent several different tasks that people are trying to accomplish.
Here are some examples of actual duplicate tasks we encountered:
A place to buy a book online with a credit card
buy a print product/compare prices/look for sales
|Purchase a publication|
|live training (guided training)|
support & training
training on how to use a product
Refresher training for using software
Where can I find training videos?
Finding training documentation
Finding training videos
Contact information for training
Contact information for training (hopefully a live chat capability)
look for user guides and training videos
live training (guided training)
Resources for training
At the end of this 2nd session our list of tasks was down to 77.
Keep clarifying the tasks and reducing the list
After the team session, continue reducing duplicates on your own. Start rephrasing tasks so that they are written in a consistent short way.
You can find more tips on how to finesse your list of tasks at this post that contains notes from a Gerry McGovern seminar. Watch out for organization-centric language, product brand names, acronyms and tasks that are too lengthy.
Survey takers will have limited time to skim over the tasks, as they make their choices. Make the tasks “easy to skim”: start them off with the most “information rich”/distinctive word you can find. Instead of “find contact information” say “contact us”.
After going through this step, I shared my altered list with the workshop participants and got their OK. We asked for their input through 2 sessions, so it was only fair that they’ll have visibility into how this edited list looked.
Next, I recommend sending an email to a group of website stakeholders. Show them the latest list of tasks and ask them to email you with any additions before a certain date. You are not waiting to hear from every one of them – just giving coworkers a chance to be heard.
When I did this, we only got 7 extra suggestions from 3 people. But the main result was that 12 additional people felt engaged in the project.
After a final round of rephrasing with our Lead of Customer Insights, the final list had 62 tasks.
Some examples of how you might rephrase tasks to be clearer:
|Before||After||What you’re accomplishing|
|status of my order||Check status of my order||More action oriented|
|Which tool will solve my problem?||Which products fit my needs?||Clarifying meaning|
|accounts and billing||Account information|
Pay a bill
|Breaking out 2 distinct tasks|
If you work at a software company, be mindful that task phrasing must be clear about whether the task is to be done on your marketing site or inside the product that you sell.
For example: one of the main use case for our website is for people to simply log-in to our “cloud” software products. Many people don’t think about our corporate site and the online products as separate entities – that’s why many of their task selections made more sense in terms of their desire to perform them inside the products we sell. Next time, I would phrase tasks more explicitly.
Step 3: Set up the survey software
Now that we had our list of 62 tasks, it was time to set up the actual mechanics of the survey.
We were targeting actual website visitors. Reaching this group was very straightforward: we did this through a popup which linked to the survey page.
We offered a $5 coffee card to participants as a “thank you”, to make it worth their time to take the survey.
If you are looking to give away a gift card as an incentive, then keep in mind that some companies have guidelines for such giveaways. They usually require you to use specific words to describe the offer (example Starbucks policy, example Amazon policy)
The popup tool I chose to use is called GetSiteControl (link), and it provides a variety of features for about $20 a month. The popups are good looking, and there is a high degree of flexibility around how the popups show up. Here is how the popup looked:
We set it up to show for English-language pages, once a person has spent 5 seconds on the page, and to show it a maximum of 1 time per day.
Under these conditions, 7.29% of people who saw the popup ended up clicking through to the survey. 86% of those people then went on to answer at least 1 survey pre-qualifier question.
You can figure out how long to run your survey by using our conversion rates as a rough guide. Let’s say that you are looking to get 100 responses and your website has 10,000 visitors a month. To get 100 responses, at an 86% conversion rate, you’ll have to get 116 clicks on your popup banner. At a 7.29% conversion rate, you’ll have to show your banner to 1,591 people. With 10,000 visitors a month, you’ll get to show that many popups in 0.1591 months – about 5 days.
The most important technical consideration is that your survey software can randomize the “top tasks” list for each participant. This prevents a situation where people short cut the survey by randomly choosing whatever tasks are at the start of the list – and those first tasks accumulating a high number of votes because of the ordering. Also, check how easy it is to set up 100 different tasks as answers to your main question (you should choose a tool that lets you upload a big list of answers all at once).
The LexisNexis survey was run through an excellent survey tool for large companies called Qualtrics
We aimed to get a certain number of responses from each of our key user groups. Getting more responses than necessary wouldn’t have added to our insights and would have cost us extra coffee vouchers. To cap responses, the survey started with 2 prequalification questions at the start – only accepting people who fell into groups for which we wanted additional responses.
So, how many responses do you need to get for your survey?
We aimed to capture 50 responses in key customer categories, and 25 in others. In the European Commission poll (107,000 respondents), the top 3 tasks emerged after the first 30 votes were cast (page 20). The University of Glasgow had the top 3 student tasks emerge after 470 votes (page 3). Consult a statistician for the number that is right for you.
The “demographics” question
Your survey should ask a question that segments people into the major groups your organization cares about. Different groups will have different types of tasks they want to perform, and you should be able to report on these differences.
As an example, if you are running a top tasks survey for a hospital, it would be great to ask participants for their age. At LexisNexis, we asked participants to indicate what kind of firm they work at (or, whether they are a student)
The “top tasks” question
The main question asked people to choose their top 5 tasks from an overwhelmingly long list, under a time limit. The purpose of this setup is to get a “gut reaction” feeling for a person’s desired tasks. Here is the phrasing we used:
Please select the top 5 key pieces of information you would like to find, or actions you would like to take, on the LexisNexis Canada company website.
You have maximum 5 minutes to make your selection. Trust your first instincts.
The Customer Carewords team phrases their question differently:
Please look at the following list and choose the FIVE most important things that help people find what they need using search.
Give a score of 5 to the MOST IMPORTANT to you, 4 to the next most important, then 3, 2, and 1.
Please give ONLY one score of 5, one 4, one 3, one 2, and one 1. Leave the rest blank.
Please trust your first instincts and spend no more than 5 minutes on this exercise.
The top tasks were set up with a 5-minute countdown.
A live countdown timer isn’t strictly necessary – Gerry’s surveys ask participants to limit their own time on the question to 5 minutes: (https://www.surveymonkey.com/r/FLM2CXS). You can also see it in a Cisco Systems survey:
Skip the countdown if the work of implementing it will slow down your project.
Ranking the tasks
Next, we asked participants to rank their selections, to show the relative importance of each task:
Open ended question
A good idea that was put in at the last minute came from Paul, the VP of Marketing & Strategy: adding open ended questions to the survey.
We asked people why they’re on our website right now (as opposed to the kind of theoretical tasks they’d like to perform on the site). This question gave us further relevant information to analyze.
Asking people about their theoretical most-important tasks and asking them about why they’re on the site right this moment are two different things. Unfortunately, the current task a person is thinking of influences their “theoretical” top tasks list. You should think about how to separate the two. In my survey, I got people to pick their top 5 “theoretical tasks” before asking them about their immediate task – I didn’t want to bias their top 5 task selection by reminding them of their immediate task. I’m sure there is room for improvement here.
Many firms use the top tasks survey to recruit participants for a follow-up round of user interface testing. The European Commission got 40% of survey takers to sign up for future testing.
At LexisNexis we added a “would you like to participate in a follow-up survey” question and had 68% of participants say “yes”. This gave us a pre-built list of participants for a follow-up round.
Why would you ask survey participants to take another survey?
As you build a new information architecture for your site, you will probably want to test your architecture with real users.
You can test your new navigation structure by creating a mock-up of it, and doing “first click testing”. This type of test reduces the risk of launching a navigation structure/ IA, and then having it fail under real-world conditions. Tools like Verify and UsabilityHub make it easy to check if people understand your navigation structure.
Remember that we promised participants a $5 coffee card as a “thank you”?
In the survey, we had to give people an option to decline the gift card. In Canada some government agencies forbid employees from getting any kind of gift from an organization that’s pursuing a government contract. Your jurisdiction might have similar rules, so be aware that you might need to add a “decline the gift” option to your survey.
Step 4: Running the survey
Now your survey is ready for launch. As it runs, continue monitoring and adjusting it. Your work doesn’t end when you hit the “launch” button.
At LexisNexis, we ran a 2 day pilot survey from May 9 to May 10. The aim was to reduce the possibility of visitors abusing the survey (because gift cards were on the line) and to detect any errors in the survey’s setup. We were going to do this by collecting about 10 responses and learning from the experience.
What we learned from the pilot:
Within those 2 days, we got 7 survey responses from the same organization (same IP address). All were articling students working in the same firm. I suspect they took the survey together because one person told their friends about the easy $5 card that they could get from filling the survey. This was a problem for us because we wanted to get responses from a wide range of people at different firms.
I aimed to make sharing the survey unreliable, and to have enough of a window during working-hours to block IP addresses that submitted multiple responses.
To prevent a “tell all your friends” scenario, I adjusted the survey invitation popup to:
- Run from 8am to 9pm on workdays
- Show randomly to only 75% of site visitors
- Appear only 1 time in a day for any given person
Insights from running the survey
Adding an open-ended question to the survey was very useful. Originally, we did not intend to add one, but at the last moment we included “What is the main reason you are on the LexisNexis website today? What are you looking to accomplish?”. This question helped us spot problems with the core Top Tasks question. When there was a major mismatch between a person’s Top Tasks and the goal of their current visit, it was a sign of a potential problem with our survey design.
Ambiguous task phrasing
As mentioned before, one issue we discovered was the ambiguous meaning of some top task terms. The same task could be interpreted as taking place inside our software tools, or as taking place on our corporate marketing site – we didn’t know how the respondent interpreted the task phrase. The solution is to be mindful of this potential problem and to have crystal-clear phrasing on top task options.
How appropriate is your incentive?
Another factor that made it difficult to get a broad number of participants is the $5 “thank you” coffee card. This was a survey of Lawyers, legal professionals and law students. For a student, getting a $5 card would be a great reason to take our survey. For a lawyer earning $500/hour, that’s not such an appealing perk. Also – some people don’t like coffee, or that specific coffee chain – so they are less likely to take our survey. I suspect that this giveaway resulted in more student respondents and fewer established-professional respondents.
To be clear: I don’t think that the responses were biased, simply that we filled our quota of student respondents quickly and would’ve had to wait longer to get the same number of responses from seasoned lawyers.
If you have difficulties with getting the right number/type of respondents, I recommend reading the University of Glasgow’s writeup of their recruiting experience (page 3). Here is a graph showing how different activities impacted the number of respondents. Ultimately, the University got 1,074 responses in 5 months.
Some ideas for how you can recruit survey participants:
- Popups on your site (with or without an incentive)
- Sweepstakes (beware special legal conditions)
- Posting to your social media followers
- Pay for online ads (great for recruiting people who haven’t heard of your organization)
- Announce the survey in a direct email / include in an e-newsletter
- Coverage in your physical newsletter
- Printed insert in physical packages that you mail out
- Announcements on your intranet
- Physical kiosks on your properties, where people can fill out surveys on a tablet
- Physical survey at an industry conference
Ultimately, the LexisNexis survey got 261 full responses. It ran from May 9 to June 30, 2017.
Step 5: Get insights from your data. And use them.
Now that you collected your Top Task survey responses, it is time to get some insights from the data.
Tasks that are important to many people
The first place where every Top Tasks analysis starts is adding up the votes that your survey participants gave to each task. This gives each task a score that lets you rank them by order of importance.
At LexisNexis we did this scoring in a simple way: each time that a person picked a task as part of their “top 5”, we gave it a score of 1. This means that the 1st most important task to a person, and the 5th most important to that person were scored the same. There are different ways to do the scoring – Gerry McGovern recommends giving 5 “points” to a person’s most important task, and only 1 “point” to their least important one. You can learn more about the consequences of the scoring system you choose at https://measuringu.com/top-tasks/ in the section “Different ways of calculating top tasks”.
You will likely find out that the few most popular tasks are important to a very large portion of your users. This is what Gerry calls the “Long Neck”. Discovering those few “tasks that you have to do right” is the main appeal of the Top Tasks approach. It helps you focus on ones that play a much bigger role than the others.
This is a graph of how the votes for tasks on our site were distributed:
Compare this distribution with what the European Commission found:
Out of 62 possible tasks there were 6 tasks that were voted in the top 24% of all priorities. Overall, the 15 most popular tasks got 50% of all the votes for Top Tasks.
The most important tasks had to do with exploring features, understanding pricing and getting training/support. This kicked off an internal discussion around transparent pricing. The data was also a personal wake up call. Before the survey, I felt that my energy should be spent on parts of the site that drew in new customers. The survey showed that serving existing customers through training and support was more important to focus on.
Least important tasks
Aside from highlighting the most important tasks, the survey shows you the least important ones (sometimes called “tiny tasks”). These are often tasks that are important to the organization, but not to the users: things like the company’s history, press releases, leadership profiles etc.
Note that just because a task is unpopular, doesn’t mean that it is unimportant or that it should be hidden. For example, at LexisNexis, the tiny tasks were important to professors who are including LexisNexis software in their courses – a group that’s very important to accommodate.
You should aim to create a navigation structure that makes it very easy to perform popular tasks, while still making it possible to perform tiny tasks.
Tasks that appear together
I recommend exploring which tasks get frequently chosen together in the “top 5”, for a given person.
You can do this using a “network map”, like this one:
Each task is a “point”, and the lines between tasks represent the number of times that the 2 tasks were chosen together. (For example: a pair of tasks that was chosen together 10 times will have a 10-pixel thick line between them. A pair that was only chosen together once will have just a 1-pixel line between them).
You can do this analysis with free network-visualization software called Gephi. It takes quite a bit of Excel wizardry to import your survey results into Gephi. Here is a quick guide to the terminology you’ll encounter:
Node: the task’s name
Edge: the connection between two tasks. An Edge connects 2 Nodes – and is the entity that represents the fact that 2 tasks were chosen in the “top 5” together. The type of Edge you will need to set up is called “Undirected”.
Weight: the number of people who chose a given pair of tasks together in their “top 5”.
So, if 20 people chose “features” and “pricing” in their top 5 tasks list. Then there are 2 Nodes (“features” and “pricing”), interconnected by an edge, which has the Weight of 20.
This analysis showed us that there were 3 major groups of tasks that commonly occur together. Some tasks were part of several groups. Those tasks – ones that interconnect major groups – can serve as a jumping off point for introducing people to a variety of related content.
For example, your “product features” page could serve as a branching-off point for content that serves your existing customers (support for a specific feature, training on it) and for other content that’s aimed at potential buyers (benefits + screenshots).
Relative tasks importance
Up to this point, we looked at analyzing the data from the perspective that every “Top 5” task selected by a visitor is equally important. However, we did ask people to rank the tasks in order of relative importance. This lets you learn about the relative importance of tasks to the people who chose them.
Here are a few ways in which to think about relative importance of tasks:
Important to a small group
These are tasks that were ranked “most important” by a small group. They usually belong to niche groups that use your site. At a software company, these visitors could be the IT People maintaining the software for the main customer. At a university, these could be the students’ parents, who interact with the site in a very different way than students.
Low-importance, to a large group
You might find that people included a set of tasks in their list but chose them as the “least important of the top 5”. At a business that makes clothing, these might be a task like “how environmentally friendly is the production process?”. You should address this kind of task on your site, but it is more of a “box to check” on a visitor’s checklist, than a major item in their journey.
One way to address these tasks is to mention them on key pages, but to hold back from creating a lot of dedicated content around them.
Important, to a large group
This is not necessarily the same as your overall “top tasks”. In our case, they tended to be administrative tasks like “log in to the software”. Make sure that these tasks are easy to perform on the site.
Step 6: use the report
Once you finish making conclusions from your data, it is time to drive some action.
Your Top Tasks analysis is a tool that helps your organization accomplish a goal. You can use it to champion the interests of end-users during a large project (such as changing an interface or improving user experience).
If you’ve ever tried to advocate for customer interests in the past, you might have seen how difficult it is to get people to focus on the customer. Your personal expertise will only get the team to a certain point – beyond that, they need to be convinced by someone more authoritative. Your Top Tasks analysis could be exactly the kind of authoritative data that would show what customers are thinking.
When your analysis is complete, decide how you will use the data. Simply preparing a report is not enough – to drive action, your data needs to be actively “sold” to people in your organization.
Answer these “5 W” questions:
- Why were these Top Tasks insights needed?
- When and where would you like your analysis to be seen? (This determines the format of your final outputs.)
- What is the minimum information someone needs to see so they understand your point?
- Who needs to see this data?
Tip: Always note the number of people who responded to your survey. It lends credibility to the data and shows why people should take your analysis seriously.
Here is how I used the Top Tasks data in our information architecture project at LexisNexis:
Why was the analysis commissioned?
This analysis’ intention was to represent the “voice of the user” during website planning discussions.
When and Where will it be used?
The analysis was not the focus of our short discussion sessions. This meant that there would be limited time to present the findings.
As a result, I prepared a short presentation with quick insights:
- The first 6 top tasks (representing 25% of votes)
- The large number of survey participants (this established how solid the data is)
- What interests are unique to each group of visitors.
I also gave out a printed copy of the full report to each participant – in case they needed deeper insights during the meeting (and for easy reference during disagreements on a point of contention).
Who needs to see your data?
Besides project participants, I also gave printed copies to key stakeholders outside the project. The reason form making so many printed copies was to ensure this information won’t be lost on a shared drive. It would survive longer as a physical object.
I shared a digital version with interested parties at our similar “sister companies” outside of Canada. This was a way of getting more benefits from our investment in the survey.
What is the minimum analysis you need to make your point?
Although the full report contained several separate analyses, I presented the key insights in an executive summary that was broken down by “groups of interest”. This let busy Vice Presidents quickly see the conclusions relevant to them, if they just wanted an overview.
The absolute smallest insight from the survey was a table of the top 6 overall tasks. This was short enough to put in the body text of an email or meeting invitation. This was a way of showing the Top Tasks to people I communicated with – even if they weren’t going to open the attached report.
At LexisNexis, the Top Tasks survey led to 3 main benefits:
- It woke me up to the importance of serving existing customers by emphasizing training and support content. As a result, these 2 types of content were made easier to find and use.
- It started a discussion around a key part of the sales process that was not addressed on the website at all. This activity was task #2 in the survey results.
- We learned that we could do better on the task that was #3 in importance. The network analysis of inter-related tasks showed that this content was important to several distinct groups of visitors.
How long will it take you to run a Top Tasks Survey?
I can’t predict how long it will take your specific organization to complete this type of survey. But, I can give you a general idea.
At our organization, it took almost 6 months to run the survey (261 responses, in a business-to-business setting). Planning started on February 9, 2017 and the final analysis report was finished on July 27, 2017.
Here is a timeline of the specific activities that took place and how long they generally took.
The larger-scale University of Glasgow survey, with 1074 respondents, took about 5 months just to collect survey responses (12 June to 6 November 2017). The analysis portion appears to have taken at least 2 months.
One Customer Carewords presentation indicates that it takes their team from 9 weeks to over 3 months to perform an entire Top Tasks test. This is a much shorter time than what it would take an in-house team. I believe that they’re able to perform the setup and analysis quickly because of their previous experience with this type of survey. Also, their team can focus 100% on running your survey (where an in-house team has other duties that demand their time).
Helping your organization see beyond its own interests
Do you feel pressured to highlight someone’s “pet content” on a website, or to add “pet features” to a product?
If so, then a Top Tasks survey can be a tool that brings the focus back to end-user priorities. Consider running a separate Top Tasks survey just for employees of your organization, asking them to choose the tasks they think that customers will choose as their “top 5”.
When you compare the internal results with the public-facing ones, you will see which tasks your team thinks are valuable but are actually unimportant to customers. For example, such a comparison on the OECD website showed that the task “Overview of what the OECD does” was 4 times less important to customers than their employees thought.
A note of caution when highlighting top tasks on your site/product
Once you know the most important tasks for your website, you will be tempted to just list all of them in a “popular tasks” area on your site.
Don’t do it!
Gerry McGovern has a great writeup on how a UK organization called Citizen’s Advice put all their top tasks into a “quick links” box on their homepage. As a result, every visitor expected their immediate task to be present as a “quick link” and it became a general catch-all category for any task whatsoever.
When you restructure your site and navigation to account for Top Tasks, remember that the navigation needs to flow logically for all tasks. Those less popular tasks (called “tiny tasks”) are still a top priority for someone. They need to be discoverable with a reasonable amount of effort.
For example, the University of Glasgow team acted on their Top Tasks findings changing the kind of content they highlight on the homepage. These adjusted areas make it easy for visitors to see important tasks. (https://medium.com/uofg-ux/top-tasks-management-part-1-64bf071fc83f)
Beyond Top Tasks
What do you do after you restructure your site to fit users’ Top Tasks?
There are several other techniques that you can read about. Here are my recommendations
- Learn how the Government of Canada improve their pages to help people perform Top Tasks. They use a continuous optimization process. It involves creating prototypes of improvements and measuring the impact they had on task performance speed
- Once you work on improving the performance of a certain top task, how do you know if you succeeded?
Gerry McGovern has a couple of articles on a “Task Performance Indicator” score that helps you see how much faster it is becoming to perform a task on your site:
- Once you’ve sped up your most important tasks, how do you determine which “medium priority tasks” you need to tackle?
The UK organization Citizen’s Advice has a great explanation of how they used data to prioritize over 3,000 pages of content for improvement:
Go and do it!
Now that you know how LexisNexis and other organizations have run Top Tasks surveys, you are ready to try it yourself.
If your organization considers “speed of access to information + task performance” a core activity (due to, for example, great pressure to make services “self serve”), then I encourage you to undertake a Top Tasks analysis on your own. Building out your team’s ability to run these types of surveys will be an investment of time that will bring benefits over and over in the future. Typical organizations that should consider doing this type of project with their in-house staff:
- Municipal, provincial and federal government agencies
- Educational institutions
- Tourism organizations
- Software companies with a large base of “legacy” users
- B2C companies that operate at scale (Walmart, social networks)
If speedy access to information + tasks is a “nice to have” in your organization, then I recommend that you engage external consultants on a one-off basis to perform a Top Tasks analysis. There will be limited benefit to trying to build out an internal team that’s capable of running these kinds of surveys.
Examples of such organizations:
- Monopolies and duopolies, where ease-of-use is not a “killer feature”
- B2B companies that make most of their money from just a few large customer companies
Thank you for setting aside time to read this document. If you have questions or comments, email me at j @ <this site>.