Featured image of post How to Complete Testing Quickly

How to Complete Testing Quickly

A reader asked how to complete testing quickly when the team has no dedicated testers.

A reader asked how to complete testing quickly when the team has no dedicated testers.

Let me start with the conclusion: there is no standard answer here!

Having no testing role only means there is no one specifically responsible for this function; but the work of testing itself has not disappeared.

“Completing testing quickly” is really better put as “quickly building quality confidence.”

Because testing is endless, there is no such thing as absolutely “finishing testing.” Based on past experience or a general consensus, once everyone basically agrees the product is probably fine, testing counts as done.

So if what we want is to quickly build this quality confidence — or to quickly reach a consensus with stakeholders that quality is OK — what factors should we consider along these lines?

Scope to Be Tested

One way to finish testing quickly is to have a small test scope — for example, a tiny change with no related affected modules, so a quick check is enough. The logic is the same as committing development code promptly and integrating frequently: don’t accumulate changes, so you save the time of tracing back through a pile of changes when something goes wrong.

Do thorough self-testing during development and integrate and test in time; don’t accumulate changes.

Existing Quality

Another aspect is whether you have confidence in the product’s existing quality. Development is incremental, so was the quality of the existing baseline up to standard? Are the relationships and impacts between the new and existing parts all known? Product quality is holistic; historical features and their effects also bear on current quality confidence. So whether the relationship between new and existing parts is clear, and whether the existing baseline itself is sound, are also critical.

Potential Problems

Then again, testing is black-box, so you can’t clearly know how much danger lies beneath the iceberg. But quality confidence actually has no direct relationship with potential problems; it mainly relates to known problems and tested scope — the fewer known problems and the larger the tested scope, the stronger the confidence. The logic behind it: more known problems often mean more potential ones; spotting one cockroach may mean a whole nest of them. So the emphasis is still on the quality of what’s handed over for testing! If self-testing isn’t thorough, don’t move on to integration testing yet — every developer should first sweep their own little patch clean.

Risk of Problems Surfacing

Finally, there is one more firewall: the risk of problems surfacing. A problem surfaces — boom! But the impact of that “boom” varies: is it just a spark that dies out, or does it blow a hole, or even bring down the whole building? Beyond the size of the problem itself, different products have different tolerance for problems. Setting off a firecracker in a building slated for demolition just means it gets demolished a bit early — tolerance is high, so confidence is high!

Summary

Taking all of the above together, completing testing quickly ultimately comes down to whether you can build delivery confidence as soon as possible, which depends on individual capability, product characteristics, and the development process. It’s really a matter of mindset.

Lastly, it has to be said: nothing can defy its inherent laws. A system that runs stably, is robust and hard-wearing can only be achieved through all kinds of running-in and tempered by all kinds of internal and external problems.

The “fast delivery” and “winning it all in one battle” you imagine — there is no such thing as peaceful times; it’s just that someone else is carrying the weight for you. And the one carrying it might be the developers in the early stage, the testers, or the operations staff — or, in the end, the users who bear it all!

Built with Hugo
Theme Stack designed by Jimmy