Skip to content
Tutorials Center
My learning

Topics

Lesson 11 of 60 ยท Static Testing and Reviews

Reviews, Walkthroughs and Inspections

Compare informal reviews, walkthroughs, technical reviews and formal inspections: who takes part, what roles exist, entry and exit criteria, and a review checklist.

A review is a structured way for people to examine a work product and find problems before the product is used. It is the main form of static testing, as described in Static vs Dynamic Testing. Reviews catch defects that no tool can see, such as a requirement that is clear but wrong for the business, or a design that cannot scale.

Not every review is a formal meeting with forms and timers. Teams choose from a spectrum, from a quick chat over a colleague's shoulder to an inspection with measured rates and written follow-up. This lesson explains the four common types, the roles in a formal review, the criteria for starting and finishing one, and how to build a checklist.

Real versus stand-in. Review meetings, forms and tools are real human practices; nothing in a browser can run them. The TryIt blocks are small stand-ins: one scores a checklist the way a reviewer would tick it, and one turns inspection numbers into a metrics summary. The thresholds shown are example values. Real teams set their own.

Why review at all

Reviews have benefits beyond finding defects:

  • Early detection. A defect found in a requirement review is fixed by editing a sentence.
  • Shared understanding. Developers, testers and the business agree on what is being built.
  • Learning. Authors see how others read their work, and reviewers pick up techniques.
  • Fewer surprises later. Fewer defects reach testing, so test cycles run shorter.
  • Better testability. A reviewer who is also a tester notices what cannot be verified. The next lesson, Reviewing Requirements for Testability, focuses on exactly that.

A review costs people's time, so it should be planned and focused. An unfocused review turns into an argument about style.

The four review types

The four types differ in formality, purpose and who leads.

Four kinds of review

Informal A quick look from a colleague. No process.
Walkthrough The author leads and explains the product.
Technical review Experts judge fitness against standards.
Inspection Formal, measured, led by a trained moderator.

More formality means more cost, and more reliable results.

1 / 2

Informal review

There is no defined process. You ask a colleague to read your document or look at your code, they comment, and you fix what you agree with. It is cheap and quick, and it can be very useful. The weakness is that the results depend on who happens to look, and nothing is recorded, so it is hard to know how thorough it was.

Walkthrough

The author leads the session and takes the participants through the work product, often step by step with example scenarios. The goals are to educate the audience, to gather feedback and to reach a shared understanding. Because the author is explaining, the group sees the product through the author's eyes, which can hide problems. Walkthroughs work well for introducing a design to a team.

Technical review

A group of qualified peers with technical expertise reviews the product to decide whether it is suitable for use and conforms to specifications and standards. A moderator, not the author, usually runs the meeting. Reviewers prepare in advance. The outcome is a recommendation: accept, accept with changes, or rework.

Inspection

The inspection is the most formal type. Its main goal is to find defects, and it follows a defined process with measured data. The author does not lead: a trained moderator does. Everyone prepares individually before the meeting using checklists, the meeting itself only logs defects rather than debating fixes, and metrics are collected to improve the process.

Roles in a formal review

Formal reviews, especially inspections, assign roles. One person may hold more than one role, except that the author should not be the moderator.

Role Responsibility
Manager Decides that a review takes place, assigns people and time, checks that objectives are met
Moderator (facilitator) Plans the review, runs the meeting, keeps it on track and on time
Author Wrote the work product and fixes the defects found
Reviewers Examine the product from their own viewpoint and report defects
Scribe (recorder) Writes down defects, decisions and open questions during the meeting

Reviewers get better results when each has a different perspective: one reads as a user, one as a tester, one as a developer, one as someone responsible for security. Different viewpoints find different defects.

The inspection process

A formal review follows a sequence of activities. The exact names vary between organizations, but the shape is stable:

flowchart LR
    A["Planning"] --> B["Kick-off"]
    B --> C["Individual<br/>preparation"]
    C --> D["Review<br/>meeting"]
    D --> E["Rework"]
    E --> F["Follow-up"]

The author fixes defects after the meeting, and the moderator confirms they were handled.

  1. Planning. Choose the scope, the reviewers and the review type, and decide the entry and exit criteria.
  2. Kick-off. Hand out the work product and related documents, explain objectives and roles.
  3. Individual preparation. Each reviewer studies the product alone and notes questions and potential defects. This is where most defects are found.
  4. Review meeting. Discuss and log findings. Decide whether each is a defect, a question or a suggestion.
  5. Rework. The author fixes the defects.
  6. Follow-up. The moderator checks that fixes are done and that exit criteria are met, and collects metrics.

Entry and exit criteria

A review should start only when it is ready, and end only when it is done. Teams write both conditions down.

Entry criteria answer "may we start?" Typical examples:

  • The work product is complete enough to review, and the author has spell-checked it and run any automated checks (a linter for code, for instance).
  • Reviewers have the supporting documents they need, such as the requirements or style guide.
  • Reviewers have been given enough time to prepare.

Exit criteria answer "are we done?" Typical examples:

  • All logged defects have been fixed, or consciously deferred and recorded.
  • The moderator has verified the fixes.
  • The review metrics were recorded.
  • A decision was recorded: accepted, accepted with changes, or needs a second review.

A review checklist

A checklist stops reviewers from forgetting things and tells them where to look. Keep it short, specific and tuned to the work product. A requirements checklist might include:

  • Is each requirement uniquely identified?
  • Can each requirement be tested, with a clear expected result?
  • Are numbers given with units and limits?
  • Do any two requirements contradict each other?
  • Are error and exception cases described?

Revise a checklist over time: drop items that never find anything and add the questions behind defects that slipped through.

The first stand-in scores a checklist. A reviewer answers each item with yes, no or not applicable. Items have a weight, and the result is a pass or fail against a threshold. It is only a model of how a team might summarize a ticked list.

JS Scoring a review checklist

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Try changing the "no" answers to "yes" one at a time and watch the score and the verdict change.

Measuring a review

Inspections are measured. Two simple numbers are the inspection rate (pages, or lines of code, reviewed per hour) and the defect density (defects found per page). A review done too fast usually finds little, because reviewers skim. A very low defect count in a fast review is a warning sign, not good news.

The next stand-in compares three reviews against example guide values. The numbers are invented for the demonstration.

JS Inspection metrics and exit criteria

Press Run (or Ctrl + Enter) to see the console output here.

No output.

The checkout review looks efficient on paper, since 30 pages were done in three hours. The numbers show it was far too fast and found very little, so the team should schedule it again.

Making reviews work

Reviews fail more from people problems than from process problems. A few habits help:

  • Review the work product, not the person. Phrase comments as "this requirement is ambiguous", not "you wrote this badly".
  • Keep sessions short. Two hours is about the limit of effective concentration.
  • Prepare. An unprepared reviewer wastes everyone's meeting time.
  • Focus on defects, not fixes. Debating solutions in the meeting consumes time. Log the defect and move on.
  • Use the findings to improve. Look for patterns in defects and change how the work is produced.
  • Have management support. Reviews need time in the schedule and results must never be used to judge individuals.

Check your understanding

Quick check ยท 5 questions

of 5 answered You scored / 5 ยท perfect!

  1. 1. Which review type is the most formal and is led by a trained moderator, with defect-finding as its main goal?

    An inspection follows a defined process with a moderator, individual preparation and collected metrics. Its primary goal is finding defects.
  2. 2. In a walkthrough, who typically leads the session?

    In a walkthrough the author leads the participants through the work product to educate them and gather feedback.
  3. 3. Which of these is an exit criterion for a review?

    Exit criteria describe when the review is done. The other options are conditions for starting, so they are entry criteria or logistics.
  4. 4. A 30-page document was reviewed in three hours and only four defects were found. What is the most reasonable conclusion?

    Ten pages per hour with a very low defect density suggests skimming. A short review with few defects is a warning sign.
  5. 5. Which role writes down defects and decisions during the review meeting?

    The scribe (recorder) documents defects, decisions and open questions so the moderator and author can focus on the discussion.

Summary

  • Reviews are people-based static testing, with four common types: informal, walkthrough, technical review and inspection.
  • More formal reviews need more effort and give more reliable results; choose the type that matches the goal.
  • Formal reviews assign roles: manager, moderator, author, reviewers and scribe.
  • The process runs from planning and kick-off through individual preparation, the meeting, rework and follow-up.
  • Entry criteria say when a review may start; exit criteria say when it is done.
  • Checklists focus reviewers, and measures such as inspection rate and defect density show whether a review was thorough.

Next: Reviewing Requirements for Testability.