Technical Architecture Assessment

Type: Assessment Confidence: 0.86 Sources: 6 Verified: 2026-03-10

Purpose

Evaluates the soundness of a software system's technical architecture across six dimensions: scalability, SPOF resilience, CI/CD maturity, incident response, observability, and code quality. Produces a structured diagnostic identifying where architecture enables or constrains the business. [src1]

Constraints

Assessment Dimensions

Dimension 1: Scalability & Performance

What this measures: System's ability to handle increasing load through horizontal or vertical scaling while maintaining acceptable latency.

ScoreLevelDescriptionEvidence
1Ad hocNo capacity planning; system fails under moderate spikesNo load testing; single server; p95 unknown
2EmergingBasic vertical scaling; known bottlenecks unaddressedManual scaling; shared DB; p95 > 2s
3DefinedHorizontal scaling for stateless services; load testing in release cycleAuto-scaling configured; DB read replicas; p95 < 500ms
4ManagedArchitecture designed for 10x load; data-driven capacity planningProven 5x spike handling; multi-region; p95 < 200ms
5OptimizedElastic architecture handles 100x spikes; cost-optimized scalingReal-time auto-scaling; sub-linear cost growth

Red flags: No one knows current p95; outages during traffic spikes; single-instance database. [src3]

Quick diagnostic question: "What happens if traffic triples tomorrow — when did you last test that?"

Dimension 2: SPOF Resilience

What this measures: Whether single points of failure have been identified and mitigated across infrastructure, data, dependencies, and people.

ScoreLevelDescriptionEvidence
1Ad hocNo SPOF analysis; critical systems on single instances; bus factor of 1Single DB server; no failover; no dependency mapping
2EmergingSome SPOFs identified but not mitigated; basic backupsBackups exist but untested; partial redundancy
3DefinedFormal SPOF audit; critical path redundancy; DR plan tested annuallyFailover for critical services; bus factor >= 2; backup restoration tested quarterly
4ManagedChaos engineering active; automated failover; quarterly DR testsChaos experiments monthly; failover < 60s; RTO < 4h
5OptimizedSelf-healing infrastructure; multi-region active-active; zero customer impact from failures80%+ automated remediation; active-active 2+ regions

Red flags: No SPOF analysis done; single DB with no replication; single engineer owns critical subsystem. [src4]

Quick diagnostic question: "If your primary database dies right now, what happens and how long until recovery?"

Dimension 3: CI/CD & Deployment Maturity

What this measures: Speed, safety, and automation of the delivery pipeline — code commit to production, benchmarked against DORA metrics.

ScoreLevelDescriptionEvidence
1Ad hocManual deployments; risky all-day events; < monthly releasesNo automated testing; change failure > 45%; lead time > 6 months
2EmergingBasic CI; semi-automated deploys; monthly-ish releasesAutomated build only; change failure 30-45%; lead time 1-6 months
3DefinedFull CI/CD; automated testing; weekly-daily deploys; feature flags60%+ coverage; change failure 15-30%; lead time 1w-1m; rollback documented
4ManagedContinuous deployment with canary/blue-green; multiple deploys/day< 15% change failure; lead time 1d-1w; MTTR < 1h; 80%+ coverage
5OptimizedOn-demand zero-downtime deployment; progressive delivery; automated quality gates< 5% change failure; lead time < 1 day; deploy confidence > 99%

Red flags: Deployments only when specific people available; no automated tests; deployment freezes beyond holidays; rollback = restore from backup. [src1]

Quick diagnostic question: "How often do you deploy, and what percentage requires a hotfix or rollback?"

Dimension 4: Incident Response

What this measures: Maturity of processes for detecting, responding to, and learning from production incidents.

ScoreLevelDescriptionEvidence
1Ad hocCustomers find issues first; no process; no postmortemsNo on-call; no alerting; MTTD in hours/days
2EmergingBasic on-call; some alerting; reactive handlingInformal on-call; postmortems for major incidents only
3DefinedStructured process; severity levels; blameless postmortems standardRunbooks for top 10 failures; MTTD < 15min for P1; action items tracked
4ManagedDistributed incident response; automated detection; SLOs with error budgetsAutomated incident creation; quarterly trend review; 90%+ action items completed
5OptimizedProactive prevention; automated remediation; predictive alerting60%+ alerts auto-remediated; incident rate declining QoQ

Red flags: No on-call; customers report outages first; same incident type recurs monthly. [src5]

Quick diagnostic question: "When was your last production incident, how did you find out, and what did you change?"

Dimension 5: Observability

What this measures: Ability to understand system behavior through logs, metrics, traces, and dashboards.

ScoreLevelDescriptionEvidence
1Ad hocNo centralized logging; basic CPU/memory metrics onlyConsole.log debugging; SSH into production; no dashboards
2EmergingCentralized logging; basic monitoring; some app metricsLog aggregation; 1-2 dashboards; no distributed tracing
3DefinedStructured logging; APM deployed; SLI-based alertingThree pillars (logs, metrics, traces); custom dashboards per team
4ManagedFull distributed tracing; anomaly detection; observability as codeEnd-to-end tracing; < 5% false positive rate; Terraform-managed
5OptimizedAI-assisted root cause analysis; predictive observabilityAIOps correlation; sub-minute root cause ID; cost-optimized telemetry

Red flags: Engineers SSH into production to debug; no one can answer current p95; monitoring is infrastructure-only. [src3]

Quick diagnostic question: "If a user reports slowness, what tool does your engineer open first and how fast can they find root cause?"

Dimension 6: Code Quality & Maintainability

What this measures: Structural health of the codebase — technical debt, test coverage, documentation, and onboarding speed.

ScoreLevelDescriptionEvidence
1Ad hocNo standards; no review; < 20% test coverage; high couplingNo linting; no PRs; no docs; every change risks regressions
2EmergingBasic standards; inconsistent review; 20-40% coverageLinting configured; some unit tests; onboarding 2-3 months
3DefinedEnforced standards; mandatory review; 40-70% coverage; ADRs2-reviewer minimum; integration tests; onboarding < 30 days
4ManagedAutomated quality gates; > 70% coverage; tech debt budgetedSonarQube gates in CI; quarterly debt reduction; clear module boundaries
5Optimized> 85% coverage; architecture enforces modularity; continuous refactoringMutation testing; onboarding < 2 weeks; health metrics trending positive

Red flags: No code review; "only one person understands this"; test suite hours long or skipped; engineers afraid to refactor. [src6]

Quick diagnostic question: "How confident is your team making a significant change to a core module without regression?"

Scoring & Interpretation

Formula: Overall Score = (Scalability + SPOF Resilience + CI/CD + Incident Response + Observability + Code Quality) / 6

Overall ScoreLevelInterpretationNext Step
1.0 - 1.9CriticalArchitecture is a business risk — outages and slow delivery constrain growthAddress lowest dimension; invest in CI/CD and observability first
2.0 - 2.9DevelopingBasic systems in place but significant gaps; breaks at 3-5x scaleClose biggest gap; priority: observability > CI/CD > incident response
3.0 - 3.9CompetentSolid foundation with room for optimizationInvest in chaos engineering, SLOs, progressive delivery
4.0 - 4.5AdvancedEngineering enables, not bottlenecks; mature DevOpsFine-tune cost efficiency; advanced observability; platform capabilities
4.6 - 5.0Best-in-classArchitecture is a competitive advantageMaintain; evaluate emerging paradigms

Dimension-Level Action Routing

Weak Dimension (Score < 3)Fetch This Card
Scalability & PerformanceCloud Migration Playbook
SPOF ResilienceBusiness Continuity Planning
CI/CD & DeploymentTechnology Stack Decision Framework
Incident ResponseCyber Risk Quantification
ObservabilityTechnology Stack Decision Framework
Code QualityTechnology Stack Decision Framework

Benchmarks by Segment

SegmentExpected Average"Good" Threshold"Alarm" Threshold
Seed/Series A (1-5 eng)1.82.51.2
Series B (6-20 eng)2.83.32.0
Growth (21-50 eng)3.54.02.8
Scale/Public (50+ eng)4.24.53.5

DORA Metrics Quick Reference

MetricEliteHighMediumLow
Deployment FrequencyOn-demand (multiple/day)Daily to weeklyWeekly to monthly< monthly
Lead Time for Changes< 1 day1 day - 1 week1 week - 1 month1 - 6 months
Change Failure Rate< 5%5-15%15-30%30-45%
Mean Time to Recovery< 1 hour< 1 day1 day - 1 week> 1 week

Only 16.2% achieve on-demand deployment; 23.9% deploy less than monthly. [src2]

Common Pitfalls

When This Matters

Fetch when a user asks to evaluate engineering architecture, prepare for technical due diligence (fundraising, acquisition), diagnose declining delivery velocity, assess readiness for a scaling phase, or onboard a new CTO/VP Engineering needing a baseline.

Related Units