Companies that implement a dedicated eCommerce solution for the first time are often taken unawares in the last stage of work. In the following article, I wanted to share some practical experience of test and launch phases with you.
These are the topics most often overlooked by both retailers and decision-makers on the client-side, thus causing so much controversy.
Client: Since you’ve finished developing, let’s launch!
In today’s most popular software development process (SCRUM), tests are carried out in each sprint. Thus, it may seem that once we’ve completed the whole list of features, the software can be launched. Unfortunately, only finishing the work on the software enables you to start comprehensive testing. Features, integrations, business processes, performance, and security – all of these need to be tested.
NASA has estimated that will be 1 error in the average 500 lines of code. In large eCommerce systems, tests usually result in tens to hundreds of bugs and errors that must be taken into account. Comprehensive tests may last several weeks. Twice as much time is needed for fixes and re-testing. Only then, the system is transferred to the client for acceptance tests. Usually, there is one month between the completion of development work and transferring the system to the client’s testing.
Client: Since you’ve tested the software so much, why do we still find errors?
Customers often expect that after the IT company tests they will get ready and error-free software. Unfortunately, practice shows that it rarely happens, which often raises controversy and misunderstanding. The IT company tests software in terms of its features, but it’s the client who best understands their business processes – that is why clients always find a lot of errors during acceptance testing. The second reason – some processes can be tested in full only after transferring the software to the client (e.g. when fully connected to client’s systems), so some errors may result from this factor.
In large IT systems, clients typically detect from tens to hundreds of errors at this stage. To report them, it’s necessary to use a system that will provide control over notifications and their subsequent processing (e.g. Redmine, Jira). Clients point out that at the stage of these tests is not worth involving too many employees. Showing your staff an imperfect product can only discourage its use later.
Client: Why are there still errors if the system was already launched?
Typically after going live, the system needs to warm up. This means that due to the mass use of the platform by users with different software, a lot of unexpected bugs may appear. In my experience, this period usually lasts 3 months from the launch, and it’s the IT company’s job to make sure that the period is as smooth as possible. First of all, give your customers the possibility to report bugs and train store service so that they can receive notifications and report them precisely. Then, the Product Owner selects bugs and prioritizes them. The team of developers implements the fixes, keeping cool and maintaining deployment procedures. Otherwise, everything is done chaotically and hasty fixes start generating new errors.
Client: The system is slow
Sometimes, after the launch, it turns out that the project is much more popular than expected and because of this the website is slower. For this reason, it’s good to launch the new eCommerce first for a selected group of customers and make measurements to determine what loads you should ultimately be prepared for. Then you can adjust hardware and software configurations.
Read also: What your Software House will never tell you
Published July 11, 2018