Scoping, Report Type, and What It Costs

Lesson 3: Scoping, Report Type, and What It Costs

Scoping is where money is made and lost. Every category you add multiplies requirements; every system you pull into the boundary multiplies evidence; every month you add to the window multiplies sampling. The goal is not the smallest possible scope — it is the smallest defensible scope, because a report that excludes the thing your customer cares about is worse than no report.

Step 1: Draw the system boundary

The system description is management's document, and it defines what the auditor examines. It must describe five dimensions:

DimensionWhat belongs in itWhere teams get it wrong
InfrastructureCloud providers and regions, network topology, physical/environmental protections, key servicesForgetting a secondary SaaS tool that touches production data (observability, support desk, CI)
SoftwareApplications, platforms, operating systems, APIs, databases in the product pathOmitting internal admin tools and data-warehouse copies
PeopleOrg structure, roles with access, who is responsible for which controlContractors and offshore support teams left out
ProceduresHow the controls operate: automated and manual, the actual stepsDescribing the aspirational process instead of the real one
DataWhat data the system holds, where it flows, classification, retentionForgetting backups, logs and derived analytics stores

Two scoping decisions sit on top of that boundary:

  • Entity-wide vs product/service-level. One report covering the whole company is simpler to maintain and satisfies all customers; per-product reports let you exclude a messy legacy system but fragment your evidence and your sales motion. Most companies start entity-wide with a clearly bounded product list.
  • Carve-out vs inclusive method for subservice organisations. If AWS, your email provider or your observability vendor is in the boundary, you generally use the carve-out method: your description names the subservice organisation and the complementary controls it must provide, and excludes the detail of its controls (you don't have them). The inclusive method pulls the subservice organisation's controls into your report — heavier, and rarely worth it. Either way, you must review and document your subservice organisations' own SOC reports annually. That review is itself a control.
CUECs are how you shift responsibility back to your customer. Complementary User Entity Controls are the things the customer must do for your controls to work — provision their own users, enforce MFA on their side, configure SSO, review their own access. Every SOC 2 report lists them. Your sales team should know yours, because sophisticated buyers check whether the CUECs match how they actually deploy you.

Step 2: Choose categories, then report type

Default answer for a B2B SaaS company: Security + Availability, Type II. Add Confidentiality if contracts define customer business data as confidential (most do). Add Processing Integrity only if your output is consumed transactionally as fact. Add Privacy almost never — it is a large scope with its own notice, consent, choice, and monitoring criteria, and it does not substitute for GDPR or CCPA compliance.

DecisionGuidance
SOC 2 vs ISO 27001ISO certifies a management system; SOC 2 reports on control effectiveness for a period. US enterprise buyers ask for SOC 2; European and global buyers increasingly ask for ISO 27001. Controls overlap heavily, so do the first one with the second in mind.
Type I vs Type IIType I only if a deal needs something in under a month. Otherwise Type II directly.
Window lengthAICPA guidance recommends at least six months for a Type II. Three months is accepted but thin. Twelve months produces the cleanest report because annual items (pen test, DR test, risk assessment, training) land inside the window instead of being marked “did not operate”.
How oftenAnnually, ideally with the next window starting the day the previous one closed — no evidence gap.
The “did not operate” trap. A control that only ran outside your review window cannot be attested to. If your pen test happened in June and your window is July–December, that control is excluded or flagged. This single mechanic is the strongest argument for a full 12-month window rather than a neat 3-month sprint: it forces the annual calendar to live inside the period.

Step 3: Pick an auditor, knowing what you're buying

Only a licensed CPA firm can issue the report. Beyond that, evaluate: SOC volume and industry experience, peer-review history, whether they use a proper evidence portal or email spreadsheets, their sampling philosophy and how they handle a deviation, whether the same firm does your readiness work (common, but check independence and, if you care, separate them), and whether the partner will actually talk to your customer's security team. A cheap auditor with a thin process produces a report buyers discount — and a report that generates 100 evidence requests in week three is not cheap in internal hours.

Independence and ethics are under scrutiny. The AICPA has publicly flagged concerns about quick-turn SOC engagements, unlicensed or non-peer-reviewed firms offering SOC services, and business arrangements between CPA firms and compliance-automation tool providers — including referring matters to state boards. Vet your auditor and your automation vendor separately, and keep the auditor's judgment free of commercial pressure.

Step 4: Budget the whole program, not the invoice

Published vendor estimates converge on the same shape, with wide ranges by size and scope. Treat these as indicative planning numbers, not quotes:

Line itemTypical rangeNotes
Readiness / gap assessment~$10k–25kOr free-to-cheap if folded into a compliance platform plus internal work
Formal audit fee — Type I~$5k–20kWeeks of work, mostly documentation review
Formal audit fee — Type II~$7k–100k+Big-4 or complex scope pushes past $150k
Penetration test~$10k–25kAnnual; also satisfies customer questionnaires
Compliance automation platform~$10k–30k/yrScales with employees and frameworks
Remediation (tools, engineering time)~$25k–85kThe variable that actually blows budgets
Internal labourOften the largest costRealistically 0.25–1 FTE for a quarter, then a steady ~10–20% of someone's year

Aggregate program costs of roughly $80k–$350k are commonly quoted for a first Type II, with lean startups landing well below that using a platform, a mid-market auditor and heavy automation. Year two is cheaper — you are paying for evidence collection and the audit, not for building controls.

Timeline

Pre-audit 2 weeks – 9 months Audit window (Type II) 3 / 6 / 9 / 12 months of evidence Fieldwork + report 1 – 3 months scope · gap analysis policies · remediation · readiness controls must operate — no retroactive evidence kickoff · PBC list · sampling draft → management response → signed opinion Realistic first-report elapsed time for a Type II: 6 – 12 months from kickoff to signed report

What you're actually choosing between

FrameworkShapeBest fit
SOC 2 Type IIAttestation on control effectiveness over a period; annualUS B2B SaaS selling to enterprises
SOC 2 Type IDesign only, as of a dateBridge while a Type II window matures
SOC 3Public summary, no control detailMarketing seal alongside a restricted SOC 2
ISO/IEC 27001Certification of an ISMS, 3-year cycle with surveillance auditsGlobal/European buyers, public-sector tenders
HITRUSTPrescriptive control framework with certification levelsHealthcare and payers
AIUC-1 (+ SOC 2)AI agent safety/security/reliability, live adversarial testing, quarterly retestShipping AI agents into enterprise — new joint offering from Oct 2026

🧠 Knowledge Check

1. Your production stack sits on AWS and you use a managed observability vendor that sees customer data. How is this normally handled in the report?

2. Why does a 12-month Type II window usually produce a cleaner report than a 3-month one?

3. Which category should most B2B SaaS companies decline to add to scope in year one?

Further Reading