Skip to main content
COOKIE NOTICECOOKIE CHOICE

We use cookies for analytics (Google Analytics) and ad measurement (Google Ads, including remarketing, and ChatGPT Ads by OpenAI). Strictly necessary cookies keep the site working. See our Cookie Policy for what each one does and how to opt out.

We’d like to use cookies for analytics (Google Analytics) and ad measurement (Google Ads, including remarketing, and ChatGPT Ads by OpenAI), and none of them is set unless you accept. Cookie Policy

Learn · SOC 2 Examinations

Changes During Your Observation Period

A Type 2 examination does not require the system to stand still. It requires evidence that controls operated effectively throughout a stated period, and a description telling the reader what changed, when, and how the system differed before and after.

Two questions decide every case: must the change be disclosed in the system description, and did the affected controls operate long enough to be tested across the period? For a significant change, disclosure is nearly always yes — while routine deployments belong in your change-management evidence and stay out of the description. Testability depends on when in the window the change landed.

DC9the criterion governing period changes
5component types a change can move: infrastructure, software, people, procedures, data
300+SOC 2 engagements supported by TCSA

Plain-English explainer · AT-C 205 examination · DC section 200 · Last reviewed August 2026

Changing the system mid-period does not invalidate a SOC 2 Type 2 report. Because the examination reports on controls operating effectively throughout a stated period, a change creates two obligations: management must disclose the relevant details of significant changes in the system description, and the CPA firm must obtain evidence covering both sides of the change. The first is written down. DC9 of the AICPA description criteria (DC section 200) requires a type 2 description to contain “the relevant details of significant changes to the service organization’s system and controls during that period that are relevant to the service organization’s service commitments and system requirements.” The second follows from the examination itself: performed under the AICPA attestation standards (AT-C sections 105 and 205), the opinion speaks to controls operating effectively throughout the period, so every month needs coverage. That is where projects come unstuck — a control first operated six weeks before the period closed has six weeks of population behind it, and drafting cannot create evidence that never existed. The change is also a tested subject: CC8.1 covers whether changes to infrastructure, data, software and procedures are authorised, tested, approved and implemented, and CC3.4 whether the entity identifies and assesses changes that could significantly impact internal control.

Key Takeaways

  • A Type 2 opinion addresses controls operating throughout a stated period, so mid-period change is expected — the work is to describe it and evidence it.
  • DC9 of DC section 200 requires a type 2 description to disclose significant changes to the system and controls during the period, including the date and the before-and-after.
  • "Significant" has no numeric threshold. DC section 200 frames it around the reader: whether an omission has a substantial likelihood of influencing report users’ judgments.
  • When a control is replaced, the auditor tests both — the superseded control over the months it operated and the new one over the remainder, with two result lines in Section 4.
  • Adding a trust services category mid-period rarely works: the added criteria need controls operating throughout the period the opinion covers for them.
  • Market practice, with no published threshold behind it: changes early in the period are absorbable, while a change in the final six weeks lands as a disclosure with little testing behind it.
The threshold

What “significant” means in practice

No percentage or headcount makes a change significant. DC section 200 frames it around the reader: the changes to disclose are those likely to be relevant to the broad range of report users, and management considers whether an omission has a substantial likelihood of influencing the judgments of the specified parties.

Its worked examples are broader than engineering teams assume. Changes to the services provided. Significant changes to IT and security personnel. Significant changes to system processes, IT architecture and applications, and to the processes and systems used by subservice organizations. Changes to legal and regulatory requirements that could affect system requirements. Changes to organisational structure, such as a change of legal entity.

A serviceable screen: would a customer reading the report decide differently if they did not know? Does it move the system boundary? Does it change any of the five component types the description covers — infrastructure, software, people, procedures or data? Does it change a control, or its population? Any yes and you are drafting the disclosure. Management makes that call against the reader test above, and judgement is the only instrument for it.

Note the direction of the risk. Over-disclosure costs a paragraph. Under-disclosure is a misstatement in the description, which is itself something the CPA firm opines on.

Decision table

Eight changes and what each one costs

The right-hand column is market practice, not a rule. It assumes a twelve-month period and a CPA firm told promptly. Your auditor decides.

ChangeDisclose under DC9?Effect on testingAbsorbable if it lands
New feature inside the existing product boundaryUsually — a change to the services providedNo new controls if existing ones govern it. Populations simply grow.Any point in the period
New product on separate infrastructureYes — the boundary must say in or outIn scope, it needs its own control history. Out, it must be named as excluded.First quarter, realistically
Cloud provider migrationYes — architecture and a change of subservice organizationTwo environments in one period: controls tested in both, both providers monitored.First half
Identity provider replacementYes — system processes and architectureProvisioning and revocation tested across two populations, split at the cutover date.First two-thirds
Replacing one control with anotherYes — DC9 reaches controls as well as the systemBoth tested over the months each was in force, and both appear in Section 4 with dates.Any point, if each side has a full occurrence
Acquiring a companyYes — organisational structure, often legal entityThe acquired environment comes in with its own control history, or stays out.Rarely absorbable in-period
Adding a trust services categoryA scoping decision taken before the period opensAdded criteria need controls operating throughout the period the opinion covers.Before the period opens
Significant IT or security leadership changeYes — DC9 names IT and security personnelControls stay, approvers change: authorisation is tested against the role-holder that day.Any point, with a documented handover
New surfaces

Is the module you shipped in June covered?

Two things decide it: whether the module materially forms part of the system as described, and how long the controls relevant to it operated during the period.

Start with the boundary. A platform-shaped description covers the production environment and the controls governing anything deployed into it, so a module shipped there is inside from day one. A product-shaped description names specific services, and anything unnamed sits outside. The same June launch is covered under the first and excluded under the second.

Then the controls. If the module rides on the existing set, populations simply grow — a deployment pipeline that saw 240 production changes in the year might now show 310, and the sample is drawn from the larger list — and coverage holds, because those controls operated all year. If it brought control activities of its own (a new review gate, a separate pipeline, a database with its own access model), those carry only the history they have. A control first operating on 1 June of a calendar year has seven months of history behind it and is testable, while one that starts on 20 November has six weeks and usually falls short. The outcome to avoid is silence: say which it is — in scope and disclosed under DC9 with the launch date, or named as excluded for this period.

Adding a category

Why Confidentiality in month nine rarely works

A prospect asks for Availability or Confidentiality, the period is already running, and the instinct is to add it. Two locks stand in the way, and only one is about you.

The first is operating history. Adding a trust services category brings a new set of criteria, each needing controls that operated for the period the opinion covers. Availability alone reaches capacity management under A1.1, environmental protections, backup processes and recovery infrastructure under A1.2, and recovery testing under A1.3 — and a recovery test run once in week forty-one is a single occurrence, a long way short of a year of operation. The second lock is the CPA firm’s judgement: they must be satisfied the coverage supports an opinion.

Three paths remain. Defer the category and give the prospect a date. Or obtain a Type 1 on the added category as of a date while the Type 2 continues on the original scope. Or shorten the whole period so every category has coverage throughout — rarely worth trading a year of tested Security evidence for a quarter. Whether one report can carry different periods for different categories is a question for the CPA firm.

Dependencies

Changing who you depend on

Replace a hosting provider or a managed service mid-period and both were subservice organizations during the period, so both belong in the description with the dates each was used and the method applied. Under the carve-out method their controls sit outside the description, but your monitoring stays inside it. CC9.2 asks whether the entity assesses and manages risks associated with vendors and business partners, and a migration is exactly when that control is exercised — expect testing of diligence on the incoming provider and of offboarding on the outgoing one.

The inclusive method makes a mid-period provider change materially harder, and this is the part teams discover late. Under it the subservice organization’s controls sit inside your description and inside the scope of the examination, so the outgoing provider’s own management would have to give a written assertion covering the months you used them and submit to the service auditor’s procedures for those months — after you have already left as a customer. Providers rarely agree, which is why a mid-period provider change is almost always handled carve-out.

A carve-out then shifts review of the provider’s own SOC report onto your reader: use one provider until June and another from July, and they need assurance covering both stretches, from reports whose periods never align with yours. The outgoing provider’s last report will usually end before your period does, so the reader needs that report plus a bridge letter from the provider for the tail — and a provider that has already lost you as a customer is the least motivated party in the transaction. Ask during offboarding, while the contract is still live.

Offboarding has one artefact auditors reliably request under CC9.2: dated evidence that your data was returned or destroyed on the terms the contract set, in the form of a certificate of destruction or a written confirmation from the provider. Teams that closed the account and moved on can rarely obtain it afterwards. Check the complementary user entity controls too, since the new provider may assume different things of you than the old one did.

Worked example

A mid-period identity provider migration

An illustrative B2B software company runs a 1 January to 31 December period covering Security and Confidentiality, and replaces its workforce identity provider in May. The dates below are the ones that end up in the report.

  1. 12 Jan

    Business case approved to retire the legacy on-premises directory.

    Nothing to test yet, though the record later shows the migration was itself authorised under CC8.1.

  2. 2 Mar

    Pilot: 40 engineering users cut over. Both platforms live.

    The dual-run begins: the access population now has two sources, and each needs its own evidence.

  3. 11 May

    Full cutover. The remaining 380 workforce accounts and all 22 SSO integrations move.

    The split date. Joiner and leaver samples are drawn against whichever platform was authoritative that day.

  4. 11–29 May

    Account, group and entitlement reconciliation, old to new, with 14 entitlements found unjustified and removed.

    The strongest evidence in the file — and the first artefact the auditor asks for.

  5. 30 Jun

    Legacy directory decommissioned, break-glass accounts closed, and final exports archived read-only.

    Proves the old path was closed and its data disposed of — what CC6.2, CC6.3 and, for the platform and its data, CC6.5 ask about.

  6. 31 Dec

    Period ends. Fieldwork opens the following month.

    Section 3 carries the change and its dates, and Section 4 shows two sub-periods of access testing.

Carried through to testing, the example yields four bounded populations for the access criteria: 31 joiners and 24 leavers before 11 May, and 46 joiners and 33 leavers after it. Under the market practice in the table below, the auditor selects from each side separately — a handful on the short side, more on the long — so the year’s access testing runs to roughly two full samples. The 22 SSO integrations are each a configuration item traceable from the old platform’s export to the new platform’s. Section 3 ends up naming both platforms, the pilot and cutover dates, the decommissioning date and what differed before and after. Section 4 shows the access criteria tested twice, plus the migration tested as a change under CC8.1. A common exception in this pattern is a leaver from March disabled in the new platform but still enabled in the legacy one — found because the auditor asked for the retired system’s user listing.

Model DC9 disclosure — Section 3, system description

On 2 March, the service organization began migrating workforce authentication from its legacy on-premises directory to a cloud identity provider, cutting over 40 engineering users while both platforms operated in parallel. On 11 May, the remaining 380 workforce accounts and all 22 SSO integrations were cut over, and the cloud identity provider became authoritative. Between 11 and 29 May, management reconciled accounts, groups and entitlements across both platforms and removed 14 entitlements lacking current business justification. The legacy directory was decommissioned on 30 June. Before 11 May, access was provisioned and revoked through directory group membership requested by ticket and applied by IT operations. From 11 May, provisioning and revocation are executed in the cloud identity provider, triggered by HR joiner and leaver events and approved by the resource owner.

Why this passes. It fixes dates, names both states of the system, and describes how the control itself operated on each side of the cutover. A reader can tell which platform governed access on any given day of the period, and the auditor can bound two populations directly from the text.

Responsibility split

Who does what when the system changes mid-period

Three parties touch a mid-period change, and the boundaries between them are structural. The opinion belongs to the independent licensed CPA firm. Tranquility Cybersecurity does readiness, evidence and coordination, and never audits, attests or signs.

TaskService organization managementReadiness partner (TCSA)Independent CPA firm
Deciding whether the change is significantDecides, and owns the judgementAdvises against DC9 and the reader testEvaluates whether the resulting description is presented in accordance with the criteria
Drafting the DC9 disclosureOwns the assertion in Section 2Drafts the text: dates, both states, the control differenceReads it as the subject matter under examination
Defining and exporting populations before decommissioningRuns and retains the exportsSpecifies which populations, at what cut, with which fieldsTests their completeness and accuracy before relying on them
Selecting samplesSupplies the bounded populationNo role in selectionDesigns and draws the sample (AT-C 205.32)
Concluding on operating effectivenessNo roleNo roleForms and signs the opinion
Responding to an exceptionWrites the response and owns itMay help assemble the factsReproduces the response in the report without opining on it
Timing

The arithmetic of a change that lands late

Everything in this section is market practice. No published threshold fixes it. A change in the first third of the period is absorbed with a disclosure and slightly larger populations. The middle third produces the two-sub-period pattern above. The last third needs the CPA firm consulted before the change is scheduled. And the final six weeks generally cannot support the opinion for anything operating less often than weekly.

The mechanism is arithmetic. Sampling needs a population, and six weeks of a twelve-month period gives a quarterly control none and a monthly control one. Work a single case: replace a quarterly access review on 15 November of a January-to-December period and the new control has 6.5 weeks and zero completed quarterly cycles behind it, so there is no occurrence for the auditor to test at all. The report then carries the DC9 disclosure with an empty result column beside it, while the superseded control is tested across the 10.5 months it did operate. A daily control lands differently on the same date: six weeks is roughly 42 occurrences, which is a testable population on its own.

That gives the operational rule for discretionary change. Where it can slip past the period end, it becomes a subsequent-event disclosure and costs one paragraph. Where it lands inside, it costs a second bounded population, a second sample and a second result line for every criterion it touches — plus the drafting time to describe both states of the system. Two weeks of patience is usually cheaper than either.

Objection handling

What buyers push back on, and the answer

Six lines a vendor-risk reviewer actually sends, and the instrument that settles each one. Two of them are answered with a bridge letter, which is a management representation and carries no auditor assurance.

“You migrated your identity provider mid-period, so access controls were not tested for the whole year.”

They were tested for the whole year, in two parts: against the legacy directory for the 130 days it was authoritative, and against the cloud provider for the remaining 235. Section 3 carries the disclosure with the pilot, cutover and decommissioning dates. Section 4 carries two control descriptions with operating-date columns and two result lines. Where a reviewer wants more, the 11–29 May reconciliation summary shows every account, group and entitlement that crossed, and what happened to the 14 that failed review.

“Section 3 discloses a significant change but Section 4 shows no exceptions. Did anyone test it?”

Disclosure and testing run on separate rails. DC9 puts the change into management’s description. The tests of controls in Section 4 show what the auditor did about it. Two instruments answer this directly: the operating-date columns, which show each control tested only across the sub-period it was in force, and the change-management results, where the migration itself appears as a selected change under CC8.1. No exceptions means the migration was governed throughout, and that the auditor obtained evidence on both sides of it.

“Your subservice organisation changed halfway through, and their report covers only part of your period.”

Correct, and expected: provider periods rarely align with yours. The description names both providers, the method applied to each, and the dates each was used, so a reader can assemble coverage from two reports. Where a provider’s report ends before your period does, the normal instrument for the tail is a bridge letter from that provider — a management representation carrying no auditor procedures. Ask the outgoing provider for it during offboarding, while the contract is still live. That leverage disappears the day you leave.

“Your report describes a system that no longer exists — you have changed since December.”

By design. The opinion is bound to a stated period and speaks to controls operating throughout it, so every SOC 2 report in circulation describes a system as it stood in the past. The instrument for the stretch between period end and today is a bridge letter signed by our management, stating whether anything material has changed and naming what has. The CPA firm performs no procedures over that gap and gives no assurance on it, and convention limits the letter to roughly three months. Past that, the honest answer is the date the next period closes.

“You added Confidentiality only from October — is that opinion for the full year?”

No. The report should say plainly which categories the opinion covers and for which period: Security across the full twelve months, Confidentiality from the date its controls began operating. Every criterion in an added category needs controls operating throughout the period the opinion covers for that category, so a category added in month ten carries a shorter window, or waits for the next period. Whether one report can carry different periods for different categories is a question for the CPA firm, and the answer belongs in the report before it belongs in a sales conversation.

“The change was made because the auditor found an exception — were you non-compliant during the period?”

The exception stands for the sub-period in which the control failed to operate, and it appears in Section 4 with the auditor’s evaluation of its cause and its effect on the conclusion. The remediation appears in management’s response, with the date it took effect. A reader seeing a deviation, a documented root cause, a dated fix and an unqualified opinion is looking at evidence that the monitoring worked and the organisation acted on what it found. Backdating the fix so the line disappears is the version that damages the report.

Mid-Period Changes — Common Questions

Disclosure, testing, scope, timing and cost, in the order teams ask them.

Does a change during the observation period invalidate a SOC 2 Type 2 report?
No. A Type 2 examination reports on controls operating throughout a stated period, and systems change across six or twelve months. The change creates two obligations: management discloses the relevant details of significant changes in the description, as DC9 of DC section 200 requires, and the CPA firm obtains evidence covering both sides of it. It becomes a problem in three situations only — where the change was ungoverned, where it went undisclosed, or where it landed so late in the period that it left no testable population behind it.
Do we have to disclose every change made during the period?
Only significant ones — those relevant to the service commitments and system requirements. DC section 200 frames significance around the reader: changes likely to be relevant to the broad range of report users, judged by whether an omission has a substantial likelihood of influencing their judgments. Its examples run from the services provided to IT and security personnel, system processes, IT architecture and applications, the processes and systems used by subservice organizations, changes to legal and regulatory requirements that could affect system requirements, and organisational structure. Routine deployments belong in your change-management evidence and stay out of the description.
Is a feature we launched in month eight covered by the report?
It depends on the boundary and the controls. Where the description covers the production environment and the module was deployed into it, the module is inside the described system from launch, governed by controls already under test: the deployment population simply grows, so a pipeline that saw 240 production changes might now show 310, and the sample comes from the larger list. Where the description names specific products, an unnamed one sits outside. Where the module brought control activities of its own — a new review gate, a separate access model — those carry only the history they have. First operated 1 June in a calendar year is seven months and testable, while 20 November is six weeks and usually falls short.
Can we add a trust services category in the middle of the observation period?
Rarely, and it is a scoping decision taken before the period opens. Every criterion in the added category needs controls operating throughout the period the opinion covers for it, and the CPA firm must be satisfied the coverage supports an opinion. Availability alone reaches capacity management under A1.1, environmental protections and backup and recovery infrastructure under A1.2, and recovery testing under A1.3. The workable paths: defer the category and commit to a date, obtain a Type 1 on it as of a date while the Type 2 continues, or shorten the whole period.
What happens when we replace a control halfway through the period?
Both are tested — the superseded control over the sub-period it was in force, the new one over the remainder — and Section 4 shows both with operating dates and results. Sample sizes follow each control’s own frequency and sub-period population, so the year’s total lands near two full samples. On an 11 May split of a calendar year, a monthly control has four completed months on one side and eight on the other. A quarterly control has one and three, and the single-cycle side is tested in full, sample equal to population. The auditor will also ask whether a window existed in which neither control operated, so document the cutover date, the owner and the handover.
We migrated cloud providers mid-period. What will the auditor need?
Both environments described with the dates each was used, both providers named as subservice organizations under the method applied, and infrastructure controls tested in both. The migration itself is tested as a change under CC8.1, and your management of the incoming and outgoing providers under CC9.2 — including dated evidence that your data was returned or destroyed on the contract’s terms, which providers supply readily during offboarding. The item teams miss is the export. Once the outgoing environment is decommissioned, the completeness of any population drawn from it can never be re-performed, so configuration baselines, user listings and log extracts must go to durable storage while the environment still runs, and stay readable until the report is issued.
How late in the period can we make a significant change?
As market practice, with no published threshold behind it: changes in the first third are absorbed with a disclosure and larger populations, the middle third produces the two-sub-period testing pattern, the last third needs the CPA firm consulted before the change is scheduled, and the final six weeks generally cannot support the opinion for anything operating less often than weekly. Arithmetic drives it. Replace a quarterly access review on 15 November of a January-to-December period and the new control has 6.5 weeks and zero completed cycles, so nothing exists to test, while the superseded control is tested across the 10.5 months it did operate. A daily control on the same date has roughly 42 occurrences and survives.
What if the change happens after the period ends but before the report is issued?
That is a subsequent event: something occurring after the period the description covers but before the date of the service auditor’s report, significant enough that a reader would be misled by its omission. Management discloses material ones in the description, and the service auditor asks about them as part of the examination under AT-C 205 — the management representation letter, dated as of the report date, reaches them as well. They fall outside the period the opinion covers, so they are disclosed and discussed but never tested. Expect the auditor to ask directly, and expect a customer to ask about anything that made the news.
Do we need a bridge letter if we changed something after the period ended?
Usually yes, where a customer is asking about today and your report period closed months ago. A bridge letter is a short statement signed by your own management, covering the stretch between the end of the report period and the date it is written, and saying whether anything material has changed — a significant change like this one should be named in it, with its date. The CPA firm performs no procedures over that gap and provides no assurance on it, so the letter carries your representation and nothing more. Convention limits it to roughly three months. Past that, most vendor-risk reviewers ask when the next period closes.
Does a mid-period change increase the cost of the examination?
Usually a little, and the driver is testing volume. A replaced control means close to two samples for the criteria it touches, and a migration adds populations from a second environment, so the CPA firm’s hours move with it. TCSA’s readiness and coordination work is quoted as a fixed fee after scoping, from $2,500 for a Type 1 and $3,000–$10,000 for a Type 2, all-in, including the independent CPA attestation. The expensive version is the unplanned one: populations that were never exported, a decommissioned system nobody can query, and fieldwork stalled while someone reconstructs a March user listing from memory.

Written By Expert Auditors

Surendra Pal SinghChief Information Security Officer & Data Protection OfficerCISODPOCISAMCSEITILISO 27001 Lead AuditorISO 27701 Lead AuditorISO 42001 Lead Auditor
Saundhi ChauhanLead AuditorISO 27001 Lead AuditorISO 27701 Lead AuditorISO 42001 Lead Auditor
Last reviewed: August 2026 • Content verified by certified lead auditors

Get in touch

Book a free consultation or send us your requirements. We respond within 24 hours.

Send us your requirement.

01Your details
02What is it about?
SOC 2
03Send
Asking about: SOC 2