SOC2Prep

SOC 2 preparation: the full walkthrough

Everything between the day a customer asks for a SOC 2 report and the day an auditor can start work, in the order it needs to happen.

Last reviewed 2026-08-16Written by Jacob Masse, TrazTech Inc.

Preparing for a SOC 2 audit means producing four things: a written scope, a policy set that matches how you actually operate, controls that work, and evidence that they worked over time. For a Canadian company of ten to fifty people starting from nothing, that is roughly two to four months of work before the observation window can even begin. This page is the whole walkthrough, in order.

None of it requires an auditor's involvement. Auditors are prohibited from building the controls they later examine, so the preparation is yours either way. The only question is whether you do it internally or pay a consultant to run it, which the readiness overview covers.

Step one: write down the scope

Scope is the first decision and the one most often skipped. It determines what evidence you have to produce for the next year, so an hour of care here saves weeks later.

Write a single page that answers: which product or service is the audit about, which environments support it, which internal systems touch customer data, which third parties process data on your behalf, and which people have access. Anything not on that page is out of scope, and you should be able to defend why.

Common scope decisions and the usual answer
SystemUsually in scopeWhy
Production application and its cloud accountsYesThis is the system customers rely on
Source control and CI pipelineYesChange management evidence lives here
Identity provider and SSOYesAccess control evidence lives here
Corporate laptopsYesEndpoint controls are part of the security criterion
Marketing website and CMSNo, usuallyNo customer data, no service commitment
A second product on separate infrastructureOnly if you say soInclude it deliberately or exclude it explicitly in the description
An acquired company still on its own stackRarely in year oneCarve it out in writing rather than discovering it mid-audit

The system description you draft here becomes a real section of the final report, written by you and reviewed by the auditor. Starting it now rather than at fieldwork is the cheapest scheduling win available.

Step two: choose the criteria, and choose few

SOC 2 is built on five Trust Services Criteria. Security, sometimes called the common criteria, is mandatory. The other four are optional and each one adds controls, evidence and audit fee.

Trust Services Criteria and when to include one
CriterionInclude whenAdded effort
SecurityAlways. It is not optional.Baseline
AvailabilityYou publish an uptime commitment or an SLALow. Monitoring and capacity evidence you likely already have
ConfidentialityCustomers send you data classed as confidential under contractLow to moderate. Classification and disposal
Processing integrityYou process transactions where correctness is the productHigh. Rarely worth it for a first audit
PrivacyA customer explicitly asked, or you handle personal information as a core functionHigh. Overlaps with PIPEDA and Law 25 work

Add criteria later, not now

The instinct on a first audit is to include everything so the report looks strong. Resist it. Every extra criterion is more controls to operate for the full observation window and more evidence to produce. Security alone satisfies the large majority of vendor security reviews. If a specific customer contract names Availability or Confidentiality, add that one. You can widen the scope in year two at far lower marginal cost.

Canadian companies asking about the Privacy criterion should note that it is not a substitute for privacy law. If you handle personal information you are subject to PIPEDA, or to Law 25 in Quebec, whether or not it appears in your SOC 2 scope. Getaudited covers which privacy law applies to you, and that work is worth doing separately rather than bolting onto an audit.

Step three: find out what is missing

With scope and criteria fixed, compare them against reality. That comparison is the gap analysis, and it is the piece of readiness with the best return on a week of effort. The output is a ranked list of things you do not have, sorted by lead time rather than by severity, because a control that takes eight weeks to establish has to start before one that takes an afternoon.

Most first-time gap lists look similar. Access reviews have never been run. Offboarding is done by memory rather than a checklist. There is no vendor register. Logging exists but retention is thirty days when the observation window is ninety. Change approvals happen in conversation rather than in the pull request. Nobody has tested the backup restore. There is no risk assessment document of any kind.

Step four: write policies that describe your company

You need a policy set, and it needs to be approved, dated and acknowledged by staff. What it does not need to be is impressive. An auditor tests you against your own policy, so a downloaded template that promises quarterly penetration tests and a formal change advisory board creates findings you would not otherwise have had.

A first-audit policy set typically covers information security overall, access control, change management, incident response, business continuity and backup, vendor and third-party management, data classification and retention, risk assessment, acceptable use, secure development, and human resources security covering hiring, onboarding, training and offboarding. Some companies combine these into fewer documents, which is fine. What matters is that every commitment inside them is one you will keep every month for the length of the observation window.

Practical rules for writing them. Use "annually" instead of "quarterly" wherever an annual cadence is genuinely defensible, because you will have to evidence every cycle. Name roles rather than people, so a departure does not invalidate the document. Date and version every policy, and record who approved it, because the approval itself is evidence. Get acknowledgement from every employee at hire and once a year after that, in a system that produces a timestamp rather than a verbal yes.

Step five: remediation, in lead-time order

This is the longest stage and the one that needs engineering time. Work it in the order below, because the items at the top take calendar time to accumulate evidence and the items at the bottom can be done in a week whenever.

Remediation work in the order to start it
WorkWhy it goes firstTypical time to establish
Centralized logging with retention past the windowYou cannot backfill logs you never kept1 to 3 weeks, then continuous
Background checks for new hiresOnly evidenced by people hired after you startImmediate to set up
Security awareness trainingNeeds a full cycle inside the window1 week to launch
Access reviewsAt least one complete cycle must fall inside the window2 weeks for the first one
Vendor register and reviewsCollecting subprocessor reports takes longer than expected2 to 6 weeks
Change management in the pull requestEvery merge after this date becomes evidence1 to 2 weeks
Risk assessmentPoint in time, but feeds several other controls1 week
Penetration testScheduling and remediation both take time4 to 8 weeks end to end
Incident response tabletopPoint in time, easy to schedule late1 day
Backup restore testPoint in time, but often reveals a real problem1 day

The penetration test deserves a note. SOC 2 does not name it as a requirement, but most auditors expect an independent test as evidence for vulnerability management, and enterprise customers ask for the letter regardless. Canadian pricing runs roughly $8,000 to $40,000 CAD depending on scope. Book it early enough that you can fix what it finds and show the fix inside the same window.

Step six: start collecting evidence before you think you need to

A Type 2 report tests whether controls operated across a period. That means the evidence has to exist for the whole period, including the first week, and there is no way to recreate an access review you did not run in March. Set the collection habit up before the window opens, not after.

The workable pattern for a small team is a folder per control, a naming convention that includes the date, and a recurring calendar entry for anything periodic with a named owner attached. Screenshots need the date, the system and the user visible in frame. Exports beat screenshots wherever a system can produce one. Tickets beat both, because they carry a timestamp and an approver you did not have to capture manually. The evidence collection page goes through what auditors accept and what they push back on.

Do you need a compliance platform

Compliance platforms connect to your cloud, identity provider and code repository and collect a portion of the evidence automatically. They also give you a control framework already mapped to the criteria, which is genuinely useful when you do not know what the control list should be.

They cost roughly $8,000 to $30,000 CAD a year, usually billed annually in advance, and the contract tends to be multi-year. They are worth it when your infrastructure is concentrated in one cloud, when you are above about thirty people, or when your team has no time to build collection habits. They are not worth it for a five person company with a single cloud account and one product, where the automated coverage is small enough that a spreadsheet and a calendar do the same job.

What no platform does is decide your scope, write your system description, run your access review, or fix a control you do not have. Roughly half the work on this page is unaffected by whether you buy one.

Step seven: the observation window

Once controls are running and evidence is being captured, you pick the window. Three months is the common minimum for a first Type 2. Six or twelve months is what mature buyers prefer, and what you will move to in year two so that reports run continuously without a coverage gap.

A Type 1 report tests design at a single date and needs no window at all. It is the right answer when a deal needs something now and a Type 2 cannot exist in time. It is not a lesser version of the same thing, it is a different test, and some enterprise buyers will accept it only with a committed date for the Type 2 to follow.

Book the audit firm before the window closes, not after. Good Canadian firms are booked out by a quarter or more, and finishing a flawless nine month window only to wait ten weeks for availability is a common and avoidable delay. The timeline page shows how the dates fit together, and GetSOC2 is where to compare firms once you are at that point.

What preparation costs in Canada

All figures Canadian dollars, all ranges, and none of them include the audit fee itself.

SOC 2 preparation costs, CAD, first year
ItemRangeNotes
Compliance platform$8,000 to $30,000 per yearAnnual prepay, often multi-year term
Paid readiness assessment$5,000 to $20,000Report only, no remediation
Consultant-led readiness$20,000 to $60,000They run the program, you still make changes
Penetration test$8,000 to $40,000Scope dependent, expected by most auditors
Security awareness training$15 to $60 per person per yearLow cost, easy to forget
Background checks$40 to $150 per hireOnly counts for hires after you start
Internal engineering time150 to 400 hoursThe largest real cost and the one left out of quotes

Want a second opinion on your plan

Tell us your scope, your date and where you are stuck, and we will tell you whether the plan is realistic and what is missing from it.

Get matched

Common questions

How long does SOC 2 preparation take?

Two to four months of preparation for a company with no formal controls, assuming one person has real time allocated to it. Companies with an existing security program, cloud infrastructure in one provider and SSO already in place can do it in six to eight weeks. The observation window is on top of that and cannot be shortened.

Can we prepare for SOC 2 without a consultant?

Yes. The work is writing down what you do, closing the gaps between that and the criteria, and keeping records. A consultant buys speed and the knowledge of what an auditor accepts, which is worth paying for when a signed contract sets your date. If nothing external is forcing the date, doing the first pass internally leaves you with people who understand the program.

Which Trust Services Criteria should a first audit include?

Security only, unless a customer contract names another one. Security is mandatory and satisfies most vendor security reviews on its own. Adding Availability or Confidentiality is a modest step up. Processing Integrity and Privacy add substantial work and are rarely the right call in year one.

Do we need a penetration test before the audit?

SOC 2 does not list one as a requirement, but in practice most auditors expect independent testing as evidence for vulnerability management, and enterprise buyers ask for the report separately. Schedule it early enough in the window to remediate findings and evidence the fixes before the period ends.

Should we do Type 1 first or go straight to Type 2?

Go straight to Type 2 if nothing is forcing an earlier date, because a Type 1 costs money and produces a report most enterprise buyers treat as provisional. Do the Type 1 when a deal needs evidence sooner than a window can produce it, and commit publicly to the Type 2 date when you issue it.

What is the most common reason a first audit slips?

Evidence that does not cover the start of the observation window. Teams open the window on paper, get busy, and start collecting properly in month two. The auditor then cannot test the first weeks, and the window either shifts forward or the report carries an exception. Set collection up before the window opens rather than during it.