Accounting outsourcing mistakes get blamed on the provider. The duties that make them expensive never left the firm. The consent nobody re-signed. The review that shrank into a signature.
Work leaves the office. The duties attached to that work stay exactly where they were.
Each one has a decision attached, and most have a rule or a number behind them.
Why Accounting Outsourcing Mistakes Land on the Firm
Nothing in the rules lets a firm delegate its way out of responsibility for the work it files.
Circular 230 governs practice before the IRS by setting rules of conduct for tax professionals, mainly attorneys, certified public accountants, and enrolled agents (IRS, Circular 230 for tax professionals). Section 10.22(a) requires a practitioner to exercise due diligence in preparing or assisting in the preparation of, approving, and filing tax returns, documents, affidavits, and other papers relating to Internal Revenue Service matters.
Relying on someone else is allowed, on terms. Section 10.22(b) says that, "Except as modified by §§ 10.34 and 10.37," a practitioner "will be presumed to have exercised due diligence" for purposes of that section if the practitioner relies on the work product of another person and "used reasonable care in engaging, supervising, training, and evaluating the person, taking proper account of the nature of the relationship between the practitioner and the person."
Those four verbs are also the four places an outsourcing program breaks.
The duty also lands on a named human rather than on the firm in the abstract: section 10.36(a) requires any individual subject to those rules who has, or individuals who have or share, principal authority and responsibility for overseeing a firm's practice governed by that part to take reasonable steps to ensure the firm has adequate procedures in effect for all members, associates, and employees for purposes of complying with subparts A, B, and C.
Mistake 1: Buying Capacity When the Bottleneck Is Review
Adding preparers to a firm whose constraint is review makes the constraint worse.
Every file a new preparer finishes arrives at the same desk that was already full. Offshore or onshore makes no difference to that arithmetic. Firms that skip this step buy seats, watch partner hours climb, and conclude that offshore does not work, when what actually happened is that they widened the top of a funnel with one outlet.
Find the constraint before you buy anything. For two weeks, log where each file waits and for how long: intake, preparation, open client questions, review, signature. The stage with the longest queue is the one to fix.
If files sit in review, your first move is a reviewer or a change in what review has to do, not more preparation capacity. If files sit waiting on client documents, no staffing decision touches the problem at all.
Mistake 2: Comparing an Hourly Rate to a Salary
An hourly rate and a salary are not the same kind of number, and comparing them directly flatters whichever one you are shopping for.
Start from the real market figure for the role. Accountants and auditors had a mean annual wage of $94,750 and a mean hourly wage of $45.56 in the May 2025 Occupational Employment and Wage Statistics survey (BLS, national employment and wage data by occupation).
That is pay, and pay is not the seat. The seat also carries benefits, payroll taxes, software licenses, the desk, supervision time, and the cost of the search that filled it.
The size of that gap is measurable. Across private industry, benefit costs averaged $14.01 per hour worked in March 2026 and made up 30.1 percent of the $46.60 in total employer compensation costs per hour worked (BLS, employer costs for employee compensation).
A firm that compares a provider's hourly rate against a bare salary line is comparing a loaded number to an unloaded one.
Build your own seat cost first, from your own payroll and benefit lines. Then the comparison is between two numbers of the same kind, and the answer stops moving depending on who is presenting it.
Mistake 3: Treating Client Consent as Paperwork You Can Finish Later
Consent under the tax return information rules is a precondition, not an onboarding task.
The timing rule is absolute. Section 301.7216-3(b)(1) states that "a taxpayer must provide written consent before a tax return preparer discloses or uses the taxpayer's tax return information." A consent gathered after the first file moved repairs nothing, because there is no retroactive consent.
The content rule catches firms that reuse an old form. Under section 301.7216-3(a)(3)(i), a consent to disclose has to carry the name of the preparer and the name of the taxpayer, identify the intended purpose of the disclosure, specify the tax return information to be disclosed, and "identify the specific recipient (or recipients) of the tax return information." It must also be signed and dated by the taxpayer.
There is one relief valve, and it applies to your business clients rather than your individual ones. Under section 301.7216-3(a)(3)(iii), a consent for a taxpayer not filing a return in the Form 1040 series may be in any format, including an engagement letter to a client, and in place of naming specific recipients it may allow disclosure to "a descriptive class of entities engaged by a taxpayer or the taxpayer's affiliate" for services in connection with preparing the return. Read that class language before relying on it, because it describes entities the client engaged. The same paragraph also switches off the separate-document rule below for those clients, which is why a business book is easier to re-paper than a 1040 book.
Three mechanical details quietly invalidate consents that firms believe they hold. Under section 301.7216-3(b)(5), a consent that does not specify its duration is effective for one year from the date the taxpayer signed it. Under section 301.7216-3(c)(1), for a Form 1040 series client a single written document cannot authorize both uses and disclosures, so one document has to authorize the uses and a separate one the disclosures. And under section 301.7216-3(c)(3), the preparer must provide a copy of the executed consent to the taxpayer at the time of execution (eCFR, special rules for consents).
Where the receiving preparer sits outside the United States and the client files in the Form 1040 series, section 301.7216-3(b)(4)(i) says a preparer inside the United States may not obtain consent to disclose the taxpayer's social security number, and must "redact or otherwise mask" it before the information goes abroad. Section 301.7216-3(b)(4)(ii) carries the only exception, for disclosure through an adequate data protection safeguard as defined by the Secretary in published guidance.
Pull three signed consents at random this week. Check the recipient name, the signature date against the date work first moved, and whether a duration is stated. The check is quick, and it tells you whether the consents you are relying on still say what you think they say.
Mistake 4: Assuming the Provider's Security Program Counts as Yours
A vendor's security posture is evidence for your program. It is not your program.
A firm in the business of completing income tax returns is inside the FTC Safeguards Rule (eCFR, section 314.2(h)(2)(viii)), and bringing in an outside team is the kind of event the rule expects you to react to. Section 314.4(g) tells covered firms to evaluate and adjust the information security program in light of "any material changes to your operations or business arrangements," and a first offshore team is that change.
Section 314.4(f) then makes oversight standing work rather than one-time diligence: select and retain service providers "capable of maintaining appropriate safeguards for the customer information at issue," require those safeguards by contract, and periodically assess them "based on the risk they present and the continued adequacy of their safeguards."
Write the assessment cadence into the contract at signature, with a date and an owner. A cadence agreed later is a cadence nobody runs.
Mistake 5: Sending Out Work That Was Never Written Down
An offshore team can follow your procedures or invent its own. Those are the only two options, and undocumented firms pick the second by default.
This is the "training" verb from the reliance rule doing real work. A preparer who has never seen your treatment for partial-year depreciation, your fixed asset threshold, or the way you roll forward a specific client's workpapers will make a defensible choice that is not your choice. You then discover it in review, one file at a time, and the rework lands in the weeks you were trying to protect.
Documentation does not mean a manual. Take the three treatments that generate the most review comments in your own files and write each one down as a page: what the firm does, why, and what the workpaper has to show. Say your reviewer keeps returning the same rental property files with the same comment. Write that treatment down once and the comment stops coming back.
Where a firm genuinely has no procedures, the honest sequence is to build them during onboarding rather than to pretend they exist. The effort is front-loaded either way. The only choice is whether it happens before the season or during it.
Mistake 6: Letting the Final Review Become a Signature
Reliance on another preparer's work is permitted. Reliance without attention is not what the rule describes.
The regulation is specific about both halves. A tax return preparer "may rely in good faith and without verification upon information and advice furnished by another advisor, another tax return preparer or other party," and is "not required to audit, examine or review books and records, business operations, documents, or other evidence to verify independently information provided" (eCFR, section 1.6694-1, verification of information furnished).
Then come the conditions that decide cases. The same paragraph says the preparer "may not ignore the implications of information furnished to the tax return preparer or actually known by the tax return preparer" and "must make reasonable inquiries if the information as furnished appears to be incorrect or incomplete" (eCFR, section 1.6694-1, verification of information furnished).
The exposure behind that standard is a preparer penalty, and it follows the signature. Under section 1.6694-1(b)(2), where a firm has a signing tax return preparer, that person generally will be considered primarily responsible for all of the positions on the return or claim for refund giving rise to an understatement. Section 1.6694-1(b)(1) adds that there may be more than one primarily responsible preparer where multiple preparers are employed by, or associated with, different firms, which is the shape of every outsourced engagement.
Where an understatement is due to an unreasonable position the preparer knew or reasonably should have known about, the penalty is the greater of $1,000 or 50 percent of the income derived, or to be derived, from the return or claim. No penalty applies where it is shown that there is reasonable cause for the understatement and the preparer acted in good faith (26 U.S. Code section 6694).
So build review around noticing rather than re-performing. A reviewer who reworks the whole file has bought no capacity at all. A reviewer who reads for the things that do not fit is doing the job the standard describes.
The same paragraph adds a third duty that maps straight onto a review checklist: the preparer "must make appropriate inquiries to determine the existence of facts and circumstances required by a Code section or regulation as a condition of the claiming of a deduction or credit" (eCFR, section 1.6694-1(e)(1)). Read for an unexplained swing against last year, a claimed deduction with no substantiation behind it, and a treatment that changed without anyone saying why.
Mistake 7: Starting in February
A team hired in February learns your software during the weeks it was supposed to be producing.
Ramp is real work: software access, your naming conventions, your review checklist, a run of practice files that nobody files. Start it inside the season and every hour of it competes with a deadline, which is why the first season with a new team so often costs more partner time than the one before it.
Our own ramp runs 3 to 4 weeks of training before a season starts, and it exists because the alternative is training on live returns.
Pick the quiet months and work backwards from the first deadline the new capacity is meant to relieve. If the calendar no longer allows a full ramp, the honest options are a smaller scope this season or a start date after it, not a compressed onboarding.
Mistake 8: Scaling on One Good Month
A good first month is a small sample taken during the period when everyone is paying attention.
Scale on measured output instead. Three numbers per preparer answer the question: review time per file, rework rate, and on-time delivery against the agreed window. Watch them across a full cycle of the work type you moved, including a stretch where volume is heavy, then add seats to the roles the numbers justify.
Scale before you have measured and the second failed offshore attempt writes itself: a firm doubles headcount on a strong start, review time per file climbs because supervision thinned out, quality slips, and the conclusion drawn is about the country rather than about the sequence.
Ask what happens when a preparer rolls off while the numbers still look good. A continuity plan agreed in a calm month is a handover; the same conversation in March is a gap.
When Outsourcing Is the Wrong Answer
Some work should not leave, and a provider who tells you otherwise is selling.
Work that cannot be described cannot be delegated. If a treatment exists only as partner judgment and changes per client, writing the procedure is the project, and outsourcing before that is finished exports the ambiguity.
A book made mostly of one-off, judgment-heavy engagements rarely repays the setup. The economics of outsourcing come from repetition, and a firm with no repeating work is paying ramp costs for a benefit it will not collect.
A firm with no reviewer who has time is not ready either. The whole model depends on a person inside the firm who can reject work, and where that person is already the constraint, adding preparation capacity makes the queue longer rather than shorter.
There is also the client-commitment case. Where your engagement terms promise that no third party touches the file, that promise governs until it is renegotiated with the client, not around them.
Questions Firms Ask About Accounting Outsourcing Mistakes
Who Is Liable if an Outsourced Accountant Makes a Mistake?
Your firm is, in the ways the client and the IRS see it. The preparer penalty for an unreasonable position falls on the tax return preparer, and due diligence is presumed where the firm used reasonable care in engaging, supervising, training and evaluating the person it relied on. Contractual remedies against a provider are a separate, private matter, and they do not change who answers to the client.
What Is the Most Common Accounting Outsourcing Mistake?
Treating the decision as a purchase rather than a change to how work moves. Priced as a purchase, the only variable is the rate. Sequenced as a change, four things happen before the first file moves: diagnose the bottleneck, write the treatments down, sign the consent, and keep the review layer the savings were meant to fund.
Do Clients Have to Be Told Their Work Is Outsourced?
For tax return information, consent is the operative requirement, and where the recipient sits outside the United States it is required before any disclosure. Professional ethics add an earlier step: the "Use of a Third-Party Service Provider" interpretation requires a CPA firm to inform the client, preferably in writing, that a third party may be used on the engagement, before confidential information is provided to that third party (Journal of Accountancy, working with third-party experts).
How Do You Test a Provider Before Committing Client Files?
Give them a fixed block of your own representative work, prepared on your software and your procedures, and grade the output yourself against a file you already know. Anonymized or mock files keep client information out of it entirely. If you use real historical files instead, they are still tax return information, so the consent has to be signed before the block moves, and for a client filing in the Form 1040 series the social security number has to be redacted or masked before the file goes to a preparer outside the United States (eCFR, section 301.7216-3(b)(4)). Either way the test shows you the two things that matter: what the team can do unaided, and what the provider's review catches before the file reaches you.
Fix the Sequence, Not the Vendor
The pattern behind these mistakes is order. Firms buy capacity before finding the constraint, move files before the consent is signed, and scale before anything has been measured. Every one of those is cheap to fix in the quiet months and expensive to fix in April.
So run the sequence the other way round. Find the constraint, write down the treatments, get the consent right, put the oversight in the contract, then test a small amount of real work and let the numbers decide the size.
If you are carrying volume you cannot staff, don't trust us, test us. Our Free 40-Hour Proof Pilot puts a block of your own representative work through the full review chain on your software and your procedures, so your reviewer grades real output before a client file is at stake.