Let us say you have joined a QA team and your lead asks, “What is our DRE?” If that is a new term, relax. It is a very small idea with one tricky part, and I will take you through it slowly.

The idea in one line

Defect Removal Efficiency, DRE, asks: of all the defects that existed, what share did we find before customers did? The formula:

DRE = defects found before release ÷ (defects found before release + defects found after release) × 100

If testing found 90 bugs and users reported 10 more in the first three months, DRE is 90 ÷ 100, so 90 percent. The number is often attributed in software engineering literature to Capers Jones, who used it widely when comparing the effectiveness of different quality practices.

The tricky part: when do you stop counting?

Here is where beginners get stuck. “Defects after release” keeps growing for as long as the software is used. Count at one week and your DRE looks wonderful, because most bugs have not been found yet. Count at one year and it looks worse.

So the team must agree on a fixed window, say the first 90 days after each release, and use that every time. A DRE is only comparable with another DRE measured over the same window. This also means the figure arrives late. You can only compute release 4’s real DRE three months after release 4 shipped, so by then you are already building release 6. Treat it as a trend for learning, not a gate.

Doing it stage by stage

The overall figure is blunt. The more useful version applies the same arithmetic at each stage. Take an illustrative project at a Jaipur software house, with defects found at each point:

Where foundDefects
Developer testing and code review30
System testing by QA45
User acceptance testing10
Production, first 90 days15

Total known defects: 100. Overall, 85 were found before release, so DRE is 85 percent.

Now look at system testing alone. When it started, 70 defects were still in the product (the 100 minus the 30 already caught). It found 45 of them. Its removal efficiency is 45 ÷ 70, which is about 64 percent. Developer-stage efficiency was 30 ÷ 100, which is 30 percent. UAT, facing 25 remaining defects, found 10, so 40 percent.

Read that table again. QA’s 45 bugs looks like the biggest bar, yet developer-stage checking caught only 30 percent of what was there. If the team wants to improve, the cheapest place may be earlier: code review and unit tests, because a defect caught in review costs far less to fix than the same one reaching UAT or production. That is the point of stage-wise DRE. It shows where defects slip through, sometimes called phase containment.

What DRE does not tell you

Several things quietly break the number, and you should say so when presenting it.

Not all defects are equal. Ninety trivial cosmetic bugs found and ten data-corruption bugs missed gives a handsome 90 percent and a terrible release. Compute DRE separately for the severe defects, which is the number a business actually cares about.

Defects never found are invisible. If users hit a bug but never report it, it is not in the count, so the real DRE is lower than measured. A lightly used feature will flatter the figure.

Classification is slippery. Is a missing requirement a defect? Is a wrong-environment configuration issue in production counted? Different teams answer differently. Settle it once, write it down, and apply it consistently.

It rewards hiding. Once DRE becomes a target, people can reclassify production bugs as “enhancement requests” and the number climbs while quality does not. This is a general problem with metrics, and it is why DRE works best when reviewed by the team itself and not used to rank people.

Defect leakage, its close cousin

You will also hear “defect leakage”, often defined as defects found after release divided by total defects found in the test phase plus after. It is essentially the complement of DRE, expressed as the share that escaped. Some organisations define leakage per stage, for example bugs found in UAT that testing should have caught. Because definitions vary, ask which one your company uses before comparing figures across teams.

To use any of this, you need a defect tracker where each bug records the stage at which it was found and, ideally, the stage at which it was introduced. Without those two fields, you are guessing. Start by adding them, collect a couple of releases, and the first useful conversation will almost certainly be about where your team’s gaps are, not about the percentage itself.

Figures and names in the examples are illustrative.

Leave a Reply

Trending