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.
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.
Which report you are preparing for changes the work, not just the invoice. Preparing for a Type 1 against a Type 2 covers what differs in evidence and timeline.
None of it requires an auditor. 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, which the readiness overview covers.
In your first month, the first thirty days covers the same ground week by week.
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.
What goes on the page
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. The longer version of this decision, including the boundary tests that catch the systems people forget, is on scope and the system description.
The decisions that come up every time
| System | Usually in scope | Why |
|---|---|---|
| Production application and its cloud accounts | Yes | This is the system customers rely on |
| Source control and CI pipeline | Yes | Change management evidence lives here |
| Identity provider and SSO | Yes | Access control evidence lives here |
| Corporate laptops | Yes | Endpoint controls are part of the security criterion |
| Marketing website and CMS | No, usually | No customer data, no service commitment |
| A second product on separate infrastructure | Only if you say so | Include it deliberately or exclude it explicitly in the description |
| An acquired company still on its own stack | Rarely in year one | Carve 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, and what belongs in it is a ten-part list rather than a blank page.
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.
| Criterion | Include when | Added effort |
|---|---|---|
| Security | Always. It is not optional. | Baseline |
| Availability | You publish an uptime commitment or an SLA | Low. Monitoring and capacity evidence you likely already have |
| Confidentiality | Customers send you data classed as confidential under contract | Low to moderate. Classification and disposal |
| Processing integrity | You process transactions where correctness is the product | High. Rarely worth it for a first audit |
| Privacy | A customer explicitly asked, or you handle personal information as a core function | High. 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. What each category costs in work and in CAD is on criteria beyond security.
The Privacy criterion 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.
Comparing firms for this? Tell us what you need and it goes to the ones in the directory that do this work. No charge, and no phone number required.
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.
What a first policy set covers
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.
Four rules that prevent self-inflicted findings
Use "annually" instead of "quarterly" wherever an annual cadence is 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 is itself evidence. Get acknowledgement from every employee at hire and once a year after, in a system that produces a timestamp rather than a verbal yes. The policy templates page has the edit pass as a checklist you can tick off document by document.
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. The items at the top take calendar time to accumulate evidence. The ones at the bottom can be done in a week whenever.
| Work | Why it goes first | Typical time to establish |
|---|---|---|
| Centralized logging with retention past the window | You cannot backfill logs you never kept | 1 to 3 weeks, then continuous |
| Background checks for new hires | Only evidenced by people hired after you start | Immediate to set up |
| Security awareness training | Needs a full cycle inside the window | 1 week to launch |
| Access reviews | At least one complete cycle must fall inside the window | 2 weeks for the first one |
| Vendor register and reviews | Collecting subprocessor reports takes longer than expected | 2 to 6 weeks |
| Change management in the pull request | Every merge after this date becomes evidence | 1 to 2 weeks |
| Risk assessment | Point in time, but feeds several other controls | 1 week |
| Penetration test | Scheduling and remediation both take time | 4 to 8 weeks end to end |
| Incident response tabletop | Point in time, easy to schedule late | 1 day |
| Backup restore test | Point in time, but often reveals a real problem | 1 day |
SOC 2 does not name a penetration test as a requirement. 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. 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 helps 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.
Deciding it in four questions
- Is your infrastructure concentrated with providers a platform integrates with, meaning one cloud, one identity provider, one code host? If it is spread across four vendors and a colocation rack, automated coverage is small and the licence buys you a control framework rather than evidence.
- Are you above about thirty people? Below that the manual collection is a few hours a month, and the licence costs more than the hours.
- Does anyone in the building know what the control list should be? If not, the mapped framework is worth real money on its own, separately from the automation.
- Can you commit to the contract term? Pricing is annual prepay and often multi-year, so the honest comparison is two or three years of licence against the hours it saves, not one.
Two yes answers make it worth pricing. Four make it obvious. One means buy a spreadsheet and a calendar entry instead, and revisit it before the second report.
If the answer is one, you still want the mapped control set that made question three worth asking, and you can get that without the licence. A free compliance workspace supplies the control set, the evidence register, policy templates and a risk register at no cost. TrazTech, which operates this site, runs one, and it checks seven common systems daily and files the evidence against the control. What it does not have is the integration breadth: seven connectors rather than hundreds, no endpoint agent and no HR system. Readiness without a paid platform lays out what each option covers and where the free ones stop.
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.
- Observation window
- The period a Type 2 opinion covers. Real elapsed time, chosen by you, starting when controls are genuinely operating.
- Type 1
- An opinion on control design at a single date. No window, so it can follow readiness immediately.
- Type 2
- An opinion on design and on operation across the window. What most enterprise buyers now specify.
- Fieldwork
- The weeks after the window closes when the auditor requests populations, samples them and asks follow-up questions.
- Exception
- A control that did not operate as described during the window. Noted in the report with your management response. Survivable, and far better than an omission.
- Bridge letter
- A short statement covering the gap between the end of your window and a customer's request date. It is not an audit and carries no opinion.
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 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. If none of these lines are affordable yet, preparing with no compliance budget covers how much of this work costs nothing but hours, and SOC 2 with no security team sets out who does each part when nobody in the company works in security.
| Item | Range | Notes |
|---|---|---|
| Compliance platform | $8,000 to $30,000 per year | Annual prepay, often multi-year term |
| Paid readiness assessment | $5,000 to $20,000 | Report only, no remediation |
| Consultant-led readiness | $20,000 to $60,000 | They run the program, you still make changes |
| Penetration test | $8,000 to $40,000 | Scope dependent, expected by most auditors |
| Security awareness training | $15 to $60 per person per year | Low cost, easy to forget |
| Background checks | $40 to $150 per hire | Only counts for hires after you start |
| Internal engineering time | 150 to 400 hours | The largest real cost and the one left out of quotes |
| Cash out, preparation only, before the audit fee | $9,000 to $130,000 | Low end is a penetration test and training, everything else done internally. High end is a platform plus a consultant-led program plus a broad test |
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 matchedCommon 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.
Should we prepare for ISO 27001 at the same time?
Only if a second buyer has asked for it in writing. The control work overlaps heavily, so the second framework adds roughly a third to your preparation effort rather than doubling it, but certification fees do not overlap at all. Running the two together has the mapping and the CAD arithmetic.
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.