Automation test is becoming more and popular these days, but the fact is that not all automation projects is a success. From my personal experience, I've experienced several failed projects.
My question is from your experience what are things to consider before starting an automation project to reduce the risk of failing the project?
imo:
1 – Project Scope
2 – Budget
3 – Human Resources
4 – AUT characteristics
5 – Your working process
Those metrics will likely affect:
1. If the project should have automation or not.
2. If yes, which tool to support.
3. When have tool, which resource to prepare.
4. When have tool and team, how to adjust the working process to best optimize the automation
@Thong,
Great comments and I think those have covered most (if not all) aspects of automation. Btw, can you elaborate more on the following points because I think it’s quite important:
1. If the project should have automation or not.
> In your opinion, what project should have automation and which should not?
2. If yes, which tool to support.
> Do you have any criteria to choose a tool?
Yeah this is a big question to dig into it afaik, anh. Some of my highlights are below:
1. If the project should have automation or not.
> In your opinion, what project should have automation and which should not?
– If your project life is quick and one time delivery >> No should not auto it. The automation test often needs time and effort to build up and scale the working scope to make it best effective.
– If your UAT was not built on a standard languages/technology/platform >> No should not auto it unless you decide to develop an automation tool by yourself to support the test (not an easy job). Each UAT itself is only suitable by one or a few automation tool using the same system platform and languages. More common structure of UAT, more easy to find the tool to support it and lesser effort to customize the tool
– If your QC team does not have knowledge on automation already >> No should not auto unless you are willing to pay for more automation resources or training fee to entire team
– If your budget is low for testing >> No should not auto unless you can pick a free tool with free support
– If your working process needs too many modifications to apply automation test >>> No should not auto unless you are willing to deal with risks to go a big modification to the entire process.
2. If yes, which tool to support.
> Do you have any criteria to choose a tool?
– The tool picking often depends on:
The platform of the AUT (mobile, OS, Web/Offline App, …)
The languages and technologies to build the UAT (.Net / Java / Python …)
The price of the tool (OpenSources / LicencesSources)
The support line for the tool customization (open sources communities, web, support team from the tool provider, …)
The working process which apply the tool (Scrum, CI, waterfall, …)
The kind of test deliverable by the tool (performance test, functionality test, regression test, ….)
The ability of the tool to integrate and adapt the test report demand (execute & get report manually, execute and report automatically through web such as Jenkins or batch file…)
There are something that I want to confirm before starting a automation project:
The Operation System that AUT support
Type of AUT
Automation tool: Programming language; Free or License; Record/Playback; Report/Log
Team member: Skillful or need to be trained
And, very detail questions for each above category.
@Thong, Lan,
Thanks for the detailed and informative clarification. That would be helpful for those who planning to start an automation project.
I am agree with the Thong Khuat and lanhuynh but i want to add few more things
you need to select the framework based on the need of your project , framework for automation testing .
@Amir,
Ya, building automation framework plays an important role in automation project.
How about ROI, does anyone here calculate ROI before starting an automation project? I’m curious to know.