Building for SOC 2 Compliance from Day One
SOC 2 is not a security audit — it is an evidence collection exercise. Teams that understand this distinction ship compliant systems faster and spend less on their first audit. Here is the architecture and process that makes it manageable.

Most early-stage SaaS companies treat SOC 2 as something they will deal with later — when a customer demands it. By then, they are paying a consulting firm $40,000 to retrofit compliance onto a system that was never designed for it. The same compliance posture, built in from the start, costs a fraction of that.
SOC 2 is an attestation framework, not a prescriptive standard. The American Institute of CPAs (AICPA) Trust Services Criteria define what you need to achieve, not how. This flexibility is a feature — you can satisfy the criteria in ways that fit your architecture — but it also means there is no cookbook. This guide is the cookbook.
The Five Trust Services Criteria
SOC 2 audits can cover up to five criteria: Security (required), Availability, Processing Integrity, Confidentiality, and Privacy. Security is mandatory for any SOC 2 report. Most SaaS companies pursuing their first SOC 2 add Availability and Confidentiality. Processing Integrity is relevant for data processing services. Privacy (based on AICPA GAPP) is complex and usually deferred to a second audit cycle.
| Criterion | What It Covers | Typical Evidence | Priority |
|---|---|---|---|
| Security (CC) | Logical and physical access controls, risk assessment, monitoring | Access logs, vulnerability scans, pen test reports | Required |
| Availability (A) | System uptime, performance, disaster recovery | Uptime metrics, DR test results, incident logs | Common add-on |
| Confidentiality (C) | Protection of confidential information | Data classification, encryption evidence, NDA records | Common add-on |
| Processing Integrity (PI) | Complete, accurate, timely processing | Processing logs, error rate metrics, reconciliation records | Situational |
| Privacy (P) | Personal information collection and use | Privacy notices, consent records, retention schedules | Deferred usually |
Infrastructure Controls That Need Day-One Architecture
Some SOC 2 controls are easy to retrofit. Documenting a policy takes an afternoon. Others require architectural decisions that are expensive to change after the fact. Know which is which before you start building.
- Encryption at rest: Every data store that holds customer data must be encrypted. Design this into your storage layer from the start. Retrofitting encryption onto an existing unencrypted database requires migration.
- Encryption in transit: All inter-service communication must use TLS 1.2+. Design your service mesh or API gateway with TLS termination from the start.
- Access logging: Every access to customer data must be logged with who, what, and when. This requires your data access layer to emit structured audit events. Retrofitting this into an ORM or query layer is painful.
- Principle of least privilege: IAM roles with minimal permissions. If you start with overly permissive roles, scoping them down later will break things.
- Multi-factor authentication: Required for all production access. Design your access management to enforce MFA — do not make it optional.
- Data segregation: Customer data must be logically or physically separated per tenant. Multi-tenant architectures that mix data are expensive to remediate.
Processes That Need to Be Running Before the Audit Period
Evidence collection for SOC 2 Type II requires showing consistent execution of controls over time, typically 6-12 months. Processes that need to be running before your audit period starts include vulnerability management (regular scanning and documented remediation), security incident response (a documented process with evidence of execution), access reviews (quarterly review and certification of who has access to what), change management (evidence that code changes were reviewed and approved), and vendor risk management (documented assessment of your critical vendors).
The auditor will ask for evidence from across the audit period, not just from the date of the audit. A vulnerability scan run the week before the audit covers nothing. A vulnerability scan run monthly for twelve months with documented remediation of findings is the evidence the auditor needs.
Day-One SOC 2 Implementation Roadmap
Tools like Vanta, Drata, Secureframe, or Tugboat Logic automate evidence collection, monitor cloud configurations, and track control gaps. At $10-20K/year, they pay for themselves in audit prep time savings alone. Start this on day one — they begin collecting evidence immediately.
SOC 2 auditors want to see that your infrastructure configuration is version-controlled, reviewed, and consistently applied. Terraform or Pulumi with a code review process satisfies this. Ad-hoc console changes that cannot be audited do not.
All production logs — application, infrastructure, access — must flow to a centralized, tamper-evident log store. AWS CloudTrail, GCP Audit Logs, or Azure Monitor are the starting point. Add application-level audit events for all customer data access.
SOC 2 requires documented policies and procedures. Write them as you implement the controls, not retrospectively. A 2-page access control policy written while you set up IAM is far more accurate than a policy written from memory six months later.
At three months, run a gap assessment against the criteria you are pursuing. Use your GRC tool's gap report or hire a security consultant for a half-day assessment. Identify what is not in place, and prioritize. You still have time to close gaps before the audit period evidence matters.
CPA firms that perform SOC 2 audits vary significantly in price, speed, and expertise. Get three quotes. Evaluate their experience with your tech stack and company stage. Boutique firms that specialize in SaaS often move faster and cost less than Big Four.
Type I vs Type II: The Practical Difference
SOC 2 Type I attests that controls are suitably designed at a point in time. Type II attests that controls operated effectively over a period (typically 6-12 months). Most enterprise customers require Type II. Type I is a stepping stone — useful to show progress to prospects while you accumulate Type II evidence. Do not let vendors or consultants sell you on Type I as a destination.
“Every month you delay implementing compliance controls is a month you cannot use as SOC 2 evidence. Time is the one thing money cannot replace in this process.”
SOC 2 Type I vs Type II: What the Difference Actually Means
SOC 2 Type I is a point-in-time assessment: an auditor evaluates whether your controls are designed appropriately as of a single date. Type II covers a period of time — typically 6 to 12 months — and tests whether those controls actually operated effectively throughout the period. Type I can be completed faster (often 2-3 months from readiness), but most enterprise procurement teams want Type II before signing material contracts. A Type I report is a useful milestone and sales asset, but plan for Type II from the start.
The practical implication: to get a Type II report, you need to run your controls consistently for the entire observation period. That means your access review process, your encryption key rotation, your incident response runbooks — all of it must actually operate, be documented, and leave evidence. Auditors do not accept "we planned to do this"; they need logs, tickets, and screenshots proving the control ran on schedule.
| Dimension | SOC 2 Type I | SOC 2 Type II |
|---|---|---|
| Scope | Design of controls at a point in time | Design + operating effectiveness over a period |
| Observation period | Single date | 6–12 months (minimum 6) |
| Time to complete | 2–4 months from readiness | 9–15 months from start |
| Typical cost (startup) | $15K–$25K | $30K–$50K |
| Enterprise acceptance | Sometimes (often a stepping stone) | Preferred by most enterprise buyers |
| Audit evidence | Policy and design documentation | Logs, tickets, screenshots of control execution |
Implementation Checklist: The Five Trust Service Criteria
SOC 2 is built around five Trust Service Criteria (TSC): Security (CC), Availability (A), Processing Integrity (PI), Confidentiality (C), and Privacy (P). Security is mandatory. The others are optional but commonly added — Availability and Confidentiality are standard additions for SaaS companies. Each criterion maps to a set of controls your organisation must implement and evidence.
- CC6.1: Logical access controls — role-based access, least privilege, access reviews every 90 days
- CC6.2: User provisioning and deprovisioning — offboarding checklist with access revocation within 24 hours
- CC6.3: Multi-factor authentication on all production systems and admin accounts
- CC6.6: Encryption at rest (AES-256) and in transit (TLS 1.2+) for all customer data
- CC6.7: Vulnerability management — monthly scans, tracked remediation SLAs (critical: 7 days, high: 30 days)
- CC7.1: System monitoring — SIEM or log aggregation with alerting on anomalous access patterns
- CC7.2: Incident response plan — documented, tested annually, with defined escalation paths
- CC9.2: Vendor/third-party risk management — annual review of critical vendors, BAAs where applicable
Vendor management deserves particular attention. Auditors will ask for evidence that you have reviewed the security posture of services that handle customer data. At minimum, collect SOC 2 reports or security questionnaire responses from your top 10-15 vendors annually. For AI services that may process customer data, this includes LLM API providers — a concern covered in depth in our HIPAA AI engineering guide.
Automation Platforms: Vanta vs Drata vs Secureframe
Compliance automation platforms have become the standard approach for startup SOC 2 programs. They connect to your infrastructure (AWS, GCP, GitHub, Okta, etc.), continuously test controls, and generate the evidence collection that auditors need. The three dominant players differ in integration depth, audit firm relationships, and pricing model.
| Platform | Best for | Integration depth | Audit firm relationships | Approx. annual cost | Notable limitation |
|---|---|---|---|---|---|
| Vanta | Early-stage startups, fast time-to-audit | Strong (200+ integrations) | A-LIGN, Johanson, others | $12K–$20K/yr | Less flexibility for custom controls |
| Drata | Growth-stage, multi-framework | Very strong (continuous testing) | Multiple major firms | $15K–$25K/yr | Higher cost at entry tier |
| Secureframe | Cost-sensitive, multiple frameworks simultaneously | Good (150+ integrations) | A-LIGN, others | $10K–$18K/yr | Less polished UX than Vanta/Drata |
All three platforms reduce the manual evidence-collection burden by 60-80%. The real time sink is not the platform — it is the policy writing, the control implementation (actually setting up MFA, access reviews, encryption), and the personnel security work (background checks, security training, signed acceptable use policies). Plan for 2-3 months of engineering and operations effort before an auditor steps in regardless of which platform you use.
Realistic Timeline to Certification
Run a readiness assessment against the TSC controls. Identify gaps — typically: no formal access review process, missing incident response plan, inadequate logging, no vendor inventory. Begin implementing missing controls. Set up your compliance platform (Vanta/Drata/Secureframe) and connect your infrastructure.
Write or formalise: information security policy, acceptable use policy, incident response plan, change management policy, vendor management policy. Collect signed acknowledgements from all employees. Complete background checks if not already in place. Run security awareness training.
If a customer is asking for evidence before you can complete Type II, a Type I audit provides a report based on current control design. Choose an auditor who will also do your Type II to reduce re-scoping overhead.
The 6-month observation period begins. Controls must run consistently: access reviews on schedule, vulnerability scans on schedule, incident response tested, vendor reviews completed. Every control execution must leave evidence — automate where possible.
Auditor pulls evidence for the observation period and tests controls. Typical fieldwork takes 3-6 weeks. Respond promptly to evidence requests — delayed responses are the most common cause of audit timeline slippage. Report issuance follows 4-8 weeks after fieldwork completion.
Cost Breakdown for Startups
| Cost component | Type I estimate | Type II estimate | Notes |
|---|---|---|---|
| Audit firm fees | $8K–$15K | $18K–$35K | Varies by scope, firm, and observation period length |
| Compliance platform | $10K–$20K/yr | $10K–$20K/yr | Vanta/Drata/Secureframe annual subscription |
| Engineering time | 1–2 months FTE | 2–4 months FTE | Implementation, remediation, evidence collection |
| Policy/legal work | $2K–$5K | $2K–$5K | Can use templates; reduce with compliance platform policies |
| Penetration test | $5K–$15K | $5K–$15K | Required by most enterprise customers; annual |
| Total range | $25K–$55K | $35K–$75K | Ongoing annual cost drops after first certification |
The numbers are real. A seed-stage startup with 10 engineers, AWS infrastructure, and a 6-month observation period should budget $40K-60K all-in for a first Type II report. The cost drops significantly for subsequent annual audits — the platform is already set up, controls are running, and auditors spend less time on design evaluation. For teams managing multiple compliance frameworks simultaneously, the AI governance and NIST RMF guide covers how SOC 2 intersects with emerging AI-specific frameworks.
Continuous Compliance in Practice
SOC 2 compliance is not a one-time event — it is an ongoing operational practice. After achieving Type II certification, you must maintain compliance continuously for your next audit period. This means: access reviews every quarter (verify that terminated employees have been deprovisioned, that access levels match job roles), vulnerability scans on a regular cadence (monthly at minimum), evidence collection for every control (automated through Vanta/Drata or manual through a shared drive), and incident response drills annually.
The most common audit finding in renewal audits: access reviews that did not happen on schedule. An employee left 3 months ago and their AWS access was never revoked. A contractor's VPN access was never terminated. These findings are easy to prevent with automated access review tools (Cerby, Opal, custom scripts that compare HR records against identity provider groups) but surprisingly common in teams that treat compliance as a project rather than a process. For teams also building for HIPAA compliance, the access control requirements overlap significantly — a single access management process can satisfy both frameworks.
- Quarterly access reviews: compare active accounts against HR records, revoke orphaned access
- Monthly vulnerability scans: infrastructure and dependency scanning, remediation tracking
- Continuous evidence collection: automated screenshots of configurations, policy acknowledgments
- Annual penetration test: third-party assessment covering OWASP Top 10 and infrastructure
- Annual risk assessment update: review threat landscape, update risk register, adjust controls
- Incident response drill: tabletop exercise with documented findings and action items
Common Mistakes That Delay Certification
The three most common mistakes that delay SOC 2 certification: starting evidence collection too late (you need 3-6 months of continuous compliance evidence for Type II — starting evidence collection 2 months before the audit is too late), over-scoping (including every system in scope when only customer-facing systems need to be covered — more scope means more controls, more evidence, more audit cost), and treating compliance as a project instead of a process (building controls for the audit and then letting them decay until the next audit cycle).
The over-scoping mistake deserves emphasis. Your SOC 2 scope should include only the systems that store, process, or transmit customer data. Internal tools, development environments, and employee-facing systems typically do not need to be in scope. A smaller scope means fewer controls, less evidence, faster audits, and lower cost. Work with your auditor to define the boundary precisely before starting implementation. For teams that also need to handle health data, the HIPAA compliance requirements have similar scoping decisions around what constitutes a covered system.
Need this kind of thinking applied to your product?
We build AI agents, full-stack platforms, and engineering systems. Same depth, applied to your problem.
Enjoyed this? Get the weekly digest.
Research highlights and AI news, delivered every Thursday. No spam.
Related articles

HIPAA and AI: A Practical Engineering Guide

Before You Deploy an AI Agent: 12 Things Engineers Skip

The Time We Cached Everything and Served the Wrong Data to the Wrong Customer

What Adding AI to Your Existing Product Actually Costs (Nobody Tells You This)
