Solving problems is something i love doing. My job presents me with problems to solve every day (e.g. how to test ‘invisible’ features or how to simulate a customer using our product), and outside of work I’m also faced with problems to solve on a regular basis (e.g. why the hot water tap doesn’t heat up or why the TV reception is bad for 1 channel only). Whatever the problem, there are ways of approaching it that will apply in most cases. Here’s the approach i use.
Step 1 – Gathering Intel
First up, we need to understand the problem we are trying to solve. So start by gathering all the information you can about the problem. Here are some starter questions worth asking:
- What steps are needed to see the problem occurs? (So you can determine if your fix worked)
- What is the expected outcome and what is the actual outcome? (So you can tell when it’s been fixed)
- Is it something that happens every time or just sometimes? And if so, how often? (This might tell you something about why it’s happening)
- Does anything you do seem to influence whether it happens or not or how severe it is? (Another potential cause to the problem)
- When did the problem start? Is it recent or always? Can you trace it back to start with a particular event? (Did that event cause the problem? Maybe it just made it visible)
Once you’ve got these questions (or similar ones) answered, you should have a good idea of what the problem is that you are trying to solve and a way to test out if your fix for the problem has worked.
Step 2 – Creating a hypothesis
Now that you know the problem, step 2 is to create a hypothesis on how you might be able to fix the problem. This will be based on your findings in step 1 on what factors seem to contribute to the problem with an idea on what change then needs to be made to fix the problem.
For example, a simple case might be that last month someone changed the type of paper the photocopier was using and over the past few weeks, there appears to have been an increased in paper jams. So your hypothesis is that by changing the paper back to the previous paper, the occurrence of paper jams should decrease. You’ve seen the change in behaviour, you notice an event which possibly contributed to the problem and you came up with a method to fix the problem, including a way to tell if it worked.
Perhaps the problem is a bit trickier. You’ve been using your iPhone for months as a calendar and it sends you email notifications when you have an appointment coming up. You recently upgraded to a newer model iPhone, at the same time as changing your email address that you receive notifications on. The iPhone calendar app has been updated and you decide that instead of checking those emails on your iPhone, you are going to start using your new iPad to read them. But the emails have the wrong times for your appointment.
Lots of variations and no obvious cause to the change. But the same rules still apply, find out all you can, and create your best guess hypothesis of what might help fix the problem.
Step 3 – Test out your hypothesis
Finally, you test out your hypothesis, make the change(s) required and see if the problem still happens. This might take a lot of testing depending on how often and predictable the problem was to reproduce. Eventually, you’ll figure out whether your hypothesis was correct or not.
Hypothesis didn’t solve the problem? Repeat Steps 2 and 3 again with the new information you’ve just learnt until you find your answer.
Applying it to my job as a tester
How does this fit in with being a tester? Well there’s two main types of problems i come across in my work.
- How to test tricky to see/verify areas of the project?
- How to create automated tests to cover those areas that are also useful, readable and maintainable (see my previous blog post for more on writing good automated tests)
In both these cases, the same rules apply as i’ve outlined above. It might take a while to gather the right data and come up with a solution, but I’ve found following this simple and general approach quite helpful in my work. In my role as a tester for the Online SaaS company, Campaign Monitor (which is awesome by the way!) I have lots of things to consider in how to test our product and where appropriate, automate tests to cover our various UI elements and so I constantly find myself with problems to solve. From how to run automated tests in parallel to figuring out how customer feedback should shape the way i test and the things i look for.
I hope this simple approach might help others solving there problems too!