The views and opinions expressed in this article are those of the CIPS member and do not reflect the official position of CIPS.
CIPS Member Article by Senthilvelan Jayachandran I.S.P., ITCP (CIPS Ontario Member)
How architecture practices can preserve institutional knowledge and improve decision-making
A technology decision can outlive the conversation that made it sensible.
The system remains. The diagram survives. A future team can see what was built, but may not know which alternatives were considered or which constraints shaped the choice.
Decisions reflect business priorities, skills, timelines, costs and acceptable trade-offs. Revisiting them as circumstances change can be responsible stewardship. The difficulty is distinguishing a changed need from a forgotten reason.
Architecture can help preserve that distinction. Its contribution extends beyond designing a solution to leaving future teams enough context to question it intelligently.
A decision makes more sense with its context
Example: the report whose purpose changed
Consider a hypothetical community centre using a daily attendance report for monthly programme planning. An overnight update meets that need; the report is not used to allocate staff during the day.
Two years later, a manager wants the same dashboard to guide staffing during busy periods. Why is the information not more current?
A diagram might make the overnight update look restrictive. A decision record would explain its purpose: planning, not same-day operations. The original design is being asked to meet a different need.
The team can now assess the new information requirement and its costs without treating the earlier choice as a mistake.
The lesson: a changed purpose calls for reassessment, not retrospective blame.
Preserve the reasoning, not just the result
A useful decision record explains the choice, its rationale and its limits. A short architecture decision record, often called an ADR, is one way to capture them.
Alternatives matter. An option set aside because of cost is different from one unable to meet a critical requirement. Recording that distinction prevents a temporary constraint from becoming an unquestioned principle.
Keep the effort proportionate. A reversible local choice needs less documentation than a decision creating dependencies across several teams. Preserve the consequential reasoning, not every conversation.
A small record with a useful future
Here is a completed illustrative record. Its identifier, dates and document references are fictional.
Record: ADR-007 | Attendance reporting | Accepted on 15 April 2024. Purpose: Support monthly programme planning; not live staffing decisions.
Choice and alternatives: Overnight update; hourly and on-demand updates considered but not needed for monthly planning.
Reason and trade-off: Daily information is sufficient for planning; the dashboard will not reflect today’s changing attendance.
Owner: Reporting Service Lead, accountable for reviewing this decision.
Review: 15 April 2025, or before any proposed same-day operational use, whichever comes first.
Related record: Reporting Design RD-003, version 1.0. No superseding decision at acceptance.
Figure 1. Preserving context changes the starting point for the next decision.
Make knowledge part of the conversation
A repository gives knowledge a home. Planning, onboarding and architecture discussions give people a reason to use it.
An architecture forum can ask: “Have we considered a similar decision before, and what is different now?” In a smaller organization, a regular conversation among technical and business leads can serve the same purpose.
The conversation should clarify assumptions and ownership, while allowing colleagues to challenge a decision without judging its original authors. Honest records include uncertainty as well as agreement.
Example: a temporary exception with a clear boundary
Imagine a training provider consolidating course-registration tools. One team temporarily retains its existing tool because current courses are already configured there. Its exception specifies the affected courses, an owner and a review before the next enrolment cycle.
A new colleague can see that this is a transition arrangement, not the preferred pattern for future courses. If migration remains impractical, the owner can document a justified extension and a new review point. A review date prompts a decision; it does not force an unsafe cutover.
The lesson: an exception needs a defined scope, an owner and a review point—not an assumed permanent life.
Keep institutional memory open to revision
Preserving a decision should not make it untouchable. “We decided this before” provides context, not sufficient justification for continuing indefinitely.
Standards and approved patterns need the same discipline: where they apply, which assumptions they depend on, and when to reconsider them.
Example: an AI assistant gains a different kind of authority
Consider a hypothetical IT support team whose AI assistant drafts answers from approved guidance. Staff review each response; the assistant cannot change access or close tickets.
Later, someone proposes automatic ticket resolution. The original record shows the boundary: drafting advice is not the same authority as completing an operational action. The proposal needs a fresh assessment of permitted actions, oversight, evidence, failure handling and ownership.
The lesson: approval for one use does not automatically authorize a broader one.
These examples preserve three essentials: purpose, boundaries and review triggers. Together, they make a decision easier to reuse appropriately—and easier to change responsibly.
When a decision changes, retain its rationale and link it to the replacement. Future readers need to distinguish the current direction from its history.
Start with the next meaningful decision
Building this capability does not require documenting the entire past. Start with one upcoming decision that has lasting consequences.
Figure 2. A review can confirm a decision as well as change it. Missing or uncertain context calls for assessment, not automatic reversal.
Capture the purpose, choice, trade-off and review trigger. Name an owner and put the record where affected colleagues can find it. At the next related discussion, use it.
Ask whether the assumptions still hold and whether outcomes are acceptable. Record what changed, or why the decision remains appropriate. Keep sensitive details within suitable access boundaries.
If earlier reasoning is missing, do not invent it. Separate known facts from assumptions and establish a current decision with the relevant people. A labelled gap is more useful than a confident reconstruction presented as fact.
A professional contribution that lasts
For IT professionals, good stewardship includes leaving enough context for others to maintain, question and improve what follows.
The goal is not fewer questions. It is better-informed questions. Architecture becomes institutional memory when the next team can inherit the reasoning without being bound to the answer.
For your next significant decision, ask: if someone reviews this two years from now, will they understand what we chose, why it made sense, and when it should change?
About the Author
Senthil Jayachandran is an Executive Architect with more than 25 years of enterprise architecture experience across banking, insurance and technology organizations. He holds TOGAF 9, Open Group Certified Master Architect and CIPS ITCP/ISP credentials, alongside multiple Salesforce Architect certifications. He writes Architecture Thinking, a newsletter on architecture governance, platform design and applied AI.
The views expressed are my own, shared for professional learning, and do not represent any current or former employer or client. All examples, dates, identifiers and diagrams are illustrative; they describe no specific organization and prescribe no universal technical solution.
The CIPS Blog is where CIPS Members from across Canada contribute and share their thoughts and insights. Send your blog article to blog@cips.ca
Certified Members – Earn Recertification Credits! – CIPS certified members can earn 15 recertification credits for a published non-refereed article. Contributing to the CIPS Member Blog is a great way to share your professional knowledge and insights while earning credits toward recertification.