Informal Walkthrough
The informal walkthrough is the author's own review. BABOK counts it among its seven review types and defines it in one sentence: the business analyst runs through the work product in its draft state and solicits feedback, and reviewers may do minimal preparation before the joint review session (§10.37.3.2). He is the one who convenes the session, he is the one who picks the guests, he is the one who walks his draft through in front of them and Wiegers puts the contrast plainly: where an inspection is meant to meet the team's quality objectives, a walkthrough principally serves the needs of the author. That is why it is, by some distance, the most capable of BABOK's four informal reviews: Wiegers' table of objectives gives it seven, and only the inspection carries more. It is also where the limit on what comes out of it begins: the scribe's notes are the trace of a session, and the evidential weight of a work product is built with the reviews that log their findings.
Goal
Of BABOK's seven reviews, the informal walkthrough is the only one whose definition bears on a draft (§10.37.3.2), and the author's position in the room commands everything else.
BABOK lists seven possible objectives for a review (§10.37.3.1), and the informal walkthrough serves five of them, more than any other informal review. Remove defects, which Wiegers grants to all six methods in his table, warning that the author's dominance makes this one a less valuable defect filter than the inspection (Table 3-3). Ensure the work product is complete and correct: the same table marks that row for two methods only, the inspection and the walkthrough, and the reason is audible in the room, the group takes the draft end to end and the case that has a step nowhere announces itself. Establish consensus on an approach or a solution: Wiegers reserves the objective for the methods that hold a session, and what this one produces is a shared understanding of what the work product means; Table 3-2 withholds any product appraisal from the walkthrough, and an agreement that has to be evidenced is built in a formal walkthrough (10.37b). Explore alternatives: authors use walkthroughs, he writes, to test their ideas and brainstorm alternative solutions. Educate reviewers about the content: a walkthrough is appropriate, he writes again, when a prime review objective is to educate others about the product, and the room takes more people than an inspection can.
BABOK's objective that carries the exploration of alternatives contains two others, answer a question and resolve an issue, and the informal walkthrough does not take them: a bounded question is put to an available peer and settled in minutes, which is the home ground of the ad hoc review (10.37g). A draft, for its part, has options, and options need people to set them against each other.
Two of the seven escape it. Check conformance to a specification or a standard: both conformance rows of Table 3-3 are empty for the walkthrough, and Table 3-2 gives the reason, no defect checklists are used in it. Conformance is checked by a reader alone, holding the document and the standard: a desk check (10.37e) or a pass around (10.37f), to which Wiegers grants both rows, and a single issue review (10.37c) when it is one standard that has to be settled. Measure work product quality: Table 3-3 reserves that row for the inspection, and Wiegers writes that walkthroughs typically generate no metrics. A measurement presupposes a defect log, so an inspection (10.37a).
Among the objectives BABOK's list does not carry, Wiegers marks one for a single one of the six methods he compares, this one: walk a concrete case through the product. That is the technique's method. Another one is missing from it, and it is the price of its strength: assess how understandable the document is. After a walkthrough, he writes, you are not sure whether you understood the material because it was well structured or because the author presented it clearly. The author's presence is what lets this review educate a group and what stops it testing comprehension, which Table 3-3 grants to the pass around (10.37f), where readers meet the document without him.
The deliverable is the scribe's notes. BABOK lists the scribe among the roles applicable to this review and gives him the task of documenting the defects, suggestions, comments, issues, concerns and outstanding questions raised during the session (Table 10.37.1). Nothing else comes out: no defect log, no product appraisal, no decision. The author corrects, and nothing in the technique closes the loop.
Usage
When to use it
- The work product is a draft: of BABOK's seven definitions, it is the only one that names that state (§10.37.3.2).
- The author wants to test options rather than have defects reported: Wiegers grants the exploration of alternatives to the walkthrough and to one other method (Table 3-3).
- The group has to agree on what the document means: consensus is built in a session, and this technique holds one.
- A group has to be educated about a work product: Wiegers sees the technique's own case there, and the room takes more people than an inspection can.
- A concrete case has to be walked end to end in front of the people who live it: the scenario strategy followed in session, as Wiegers describes it.
- What is missing matters more than what is wrong: Wiegers gives completeness to the inspection and the walkthrough alone (Table 3-3).
- No record is required: nothing comes out of the session but the scribe's notes.
When not to use it
- A record or evidence of a reading is required (a quality gate, an audit, an approval, defect metrics): the session typically follows no defined procedure, nobody validated that the group had read the draft, and BABOK lists no facilitator to do it. An inspection (10.37a) logs the defects and verifies the rework.
- The question is whether the document reads on its own: understandability is judged without the author in the room. A pass around (10.37f) puts independent readers in front of the same text.
- Conformance to a specification or a standard has to be checked: that is a reader alone with the document and the standard. A desk check (10.37e) gives it; a single issue review (10.37c) settles the matter when one standard is what is at stake.
Description
Walking a case through the draft
BABOK gives this technique no steps. Wiegers names a strategy for it, and it is the whole method: follow a script that describes a specific task or scenario, end to end, to show how the system would function during a sample user session. The thing carries beyond code, he adds, and it reveals errors in early deliverables such as requirements specifications. A draft read out loud, section after section, produces nodding heads. The same draft crossed by a concrete case produces the precise place where the case stops.
One activity delivers three of the five objectives. The case that has a step nowhere is completeness. The step that admits two defensible answers is the options. The word two departments use in two senses is the consensus.
Where the informal walkthrough sits
BABOK lists its seven review types in decreasing formality and calls the first three formal techniques, the next four informal techniques (§10.37.3.2). Two binary questions cut across that ordering: is the session planned, does the reviewer examine the work product alone before speaking? The informal walkthrough answers yes, then no, and it is alone in answering that way. That signature is its boundary with its neighbours: the session is convened, where an ad hoc review (10.37g) is held without being convened, and the reviewer may come into it having done minimal preparation, where a desk check (10.37e) and a pass around (10.37f) consist entirely of that reading.
The name, for its part, is standard: of BABOK's four informal reviews, it is the only one both standards of the field carry, IEEE 1028-2008 among its five types and ISO/IEC 20246:2017 among its ten.
The roles in the room
The author presents. Wiegers puts him in the leader's role and in the material presenter's role at once, and of the three types his table compares, the walkthrough is the only one in that position (Table 3-2); the other review roles, he adds, are usually not defined in it. BABOK for its part lists the scribe among the roles applicable to this review and lists no facilitator for it (Table 10.37.1). Of the seven review types, it is the only one in that position: the session is recorded, and nobody is charged with overseeing it.
| Review type | Facilitator | Scribe |
|---|---|---|
| Inspection (10.37a) | Yes | Yes |
| Formal walkthrough (10.37b) | Yes | Yes |
| Single issue review (10.37c) | May be helpful | No |
| Informal walkthrough (10.37d) | No | Yes |
| Desk check (10.37e) | No | No |
| Pass around (10.37f) | No | No |
| Ad hoc review (10.37g) | No | No |
Setting the two columns side by side is ours, and BABOK does not do it. It does name, though, what the informal walkthrough traded away for its speed: the facilitator, in that same table, has the task of validating that the reviewers have examined the work product before the session and of ensuring that each of them takes part. Those are the two obligations the technique's definition drops, since it allows reviewers minimal preparation, and nobody here answers for them.
The traps
The room goes quiet
Wiegers says it plainly: when you do not fully understand what is being presented, you tend to gloss over it and assume the author is right, and it is easy to be swayed into silence by the rationalization an author gives of his approach, even without being convinced. Yet a reviewer's confusion, he adds, is sometimes a clue that a defect lurks nearby. Ensuring that all the reviewers take part is exactly what Table 10.37.1 gives the facilitator, and this review has none. BABOK also holds that a review is focused on the work product and not on the skills of the participants (§10.37.2): the precaution counts here more than anywhere, with the author standing in front of the group, presenting his own work.
The group sees only what it is shown
The granularity of the material presented is left to the author's discretion (Table 3-2), and Wiegers draws the warning from it: drawing the reviewers' attention to the portions that were changed runs the risk of overlooking the ones that should have been changed and were not.
Taking it for assurance
Six people were in the room, so "the document was reviewed". Nobody validated that a single one of them had opened the draft, and the proof that an independent reading took place is manufactured with a defect log and a verification of the rework, which is to say with an inspection (10.37a).
Sliding into a formal walkthrough without deciding to
The day a facilitator is appointed, individual reading is required before the session and a product appraisal comes out of it, what is being held is a formal walkthrough (10.37b), and it is better that it be a choice than a drift. A finding that has to survive the session belongs, for its part, to item tracking (10.26).
AI considerations
This review's own weakness comes down to one sentence: the author is in the room. A large language model serves exactly there, and it is taken for the room itself just as easily.
Rehearsing the pass
The method consists in following a case end to end, and a model can make the trip first: the step with no alternative branch, the actor cited at step eight and never introduced, the case that has a step nowhere. The group's hour is then spent on what only the group can see.
Being the reader who was not in the room
A model reading the draft cold produces the one reading the session cannot produce, the one where the author explains nothing. That does not make it a pass around (10.37f): what Wiegers has judged there is understandability, by people who occupy different positions and who have something at stake. A model occupies no position. It is a rehearsal, and the test is still to be run.
The model does not increase the number of people who have read the draft
Nobody validates that the reviewers examined the work product, and a well-kept list of findings, out of a machine, makes that hole look like a closed door. Consensus escapes it for the same reason: the agreement the session produces is the agreement of the people who will have to live with the process, and a model that picks between two options has not weighed them, it has hidden that there were two.
Data protection
A hospital admission process carries patient data. Compliance with the revised Federal Act on Data Protection (revFADP) is settled before productivity, and the organisation's rule, which model, which data, hosted where, is fixed before the session.
Examples
What an informal walkthrough produces is a scribe's notes, taken while the author was walking a case through in front of the group.
| Step in the draft | Raised by | What was said in the session | Kind |
|---|---|---|---|
| Step 1, "the patient reports to the admissions desk" | Nursing | "An emergency patient does not arrive through the desk. The draft has no other entry point." | Open question |
| Step 1, two possible entry points | Admissions · Billing | "Open a provisional file at triage, completed afterwards." / "Enter the admission after the fact, from the care file." The room put both on the table; it did not settle them. | Options |
| Step 3, "read the AVS number from the insurance card" | Admissions | "A newborn has no AVS number yet and is admitted with the mother." | Defect |
| Step 5, "check LAMal cover" | Billing | "Online at admission or deferred to billing." Care does not wait for the cover: both options hold, the cost is not the same. | Options |
| The word "admission" | the whole room | "Administrative registration" for admissions, "the patient's arrival on the ward" for nursing. The room agreed on administrative registration. | Point agreed in the session |
| No step | Social services | "A patient resident outside the canton appears nowhere in the flow, although the stay requires a guarantee of payment from the canton of residence." | Defect: case absent from the model |
| Step 8, financial guarantee | Billing | "The draft freezes at CHF 5'000 the guarantee asked of an uninsured patient. That is a departmental parameter." | Suggestion |
The findings came in the session, from several people, while the author was walking a case through, and two of the rows are options set against each other.
The case absent from the model surfaced because the group heard the flow end to end. The author corrects it himself, and nothing in the technique checks that he did.
Nobody validated that the group had read the draft, and BABOK lists no facilitator for this review. The notes are the trace of a session, they are not the proof that a reading took place.
Visualizations
The formality spectrum is a spatial object: seven types placed on an axis, two marks per cell. It locates and it demonstrates: the informal walkthrough's signature on the two gates belongs to no other cell. Rendered as a table, it would lose the axis and the isolation of the cell.
The roles applicable to the seven reviews are rows and columns, read by scanning, and they come from a single BABOK table. Content made of cells is rendered in cells.
The artefact actually produced is the scribe's notes, qualified by the enumeration BABOK entrusts to the scribe (Table 10.37.1), with one kind as the exception: the options, which that list does not carry and which walking a concrete case through brings out. The choice between the seven reviews is handled in reviews (10.37), which exist for that.
Cost
| Phase | Level | Justification |
|---|---|---|
| Preparation | Medium | Wiegers counts planning in this technique (Table 3-1). A session gets convened: the author picks the guests, fixes the date, prepares his pass. What is not paid for is the reviewers' preparation (Table 3-1, preparation: no), and BABOK writes that they may do minimal preparation. |
| Execution | Medium | One session, several people in the same room at the same time. The cost is the session multiplied by the number of guests, and it is paid in one block. |
| Documentation | Low | The scribe's notes, taken in session. No collation, no defect log, no product appraisal, no metrics (Table 3-2). The author corrects, and nothing in the technique closes the loop (Table 3-1, correction: yes, verification: no). |
Tooling
The informal walkthrough needs a room, the draft on display, a case to walk through and somebody who writes. Wiegers marks the use of a recorder as a maybe (Table 3-2) and BABOK lists the scribe among the applicable roles (Table 10.37.1): the only decision to take is who writes, and it is taken before the session.
Table 3-2, on the other hand, counts no defect checklist for this technique. The day you want one, another review is the one you have chosen, a desk check (10.37e) for instance. And a finding that has to live beyond the session belongs to item tracking (10.26): the scribe's notes track nothing, they say what was said.
Sources
- IIBA, A Guide to the Business Analysis Body of Knowledge (BABOK Guide) v3, §10.37 Reviews (§10.37.2, what the review is focused on; §10.37.3.1, the seven objectives; §10.37.3.2, the seven types; Table 10.37.1, the roles).
- Karl E. Wiegers, Peer Reviews in Software: A Practical Guide, Addison-Wesley, 2002, chapter 3, "Peer Review Formality Spectrum" (Table 3-1, activities by review type; Table 3-2, inspection, team review and walkthrough compared; Table 3-3, objectives and methods).
- ISO/IEC, ISO/IEC 20246:2017, Software and systems engineering: Work product reviews.
- IEEE, IEEE Std 1028-2008, IEEE Standard for Software Reviews and Audits.