The Outsourcing Mistake Owners Blame on the Model When the Handover Was the Problem

A firm signs an offshore provider in January. Week two, somebody sends 40 returns because that was the capacity. The work comes back wrong. Not stupid wrong. Wrong the way work is wrong when someone had to guess. Wrong folder. Wrong file name. Depreciation that doesn’t match last year. A schedule left blank because nobody said it was needed.

By April, the partner decided offshore doesn’t work for firms of his size.

Nothing was wrong with the model or the preparers. The work left the building before anyone wrote down how to do it. The BPO world sorted this out decades ago: transitions fail on knowledge transfer, not on the people doing the work. Accounting firms keep relearning it, one season at a time.

A proper handover takes about 90 days and runs in three phases. Here is the shape of it, and then the 10 things that make each phase work.

Nothing live in the first 30 days. That single rule prevents most of what goes wrong.

 Days 1 to 30: Get Ready

1. Timing and Ownership

A handover eats your senior people’s time for weeks. That is exactly what you don’t have in season. May through August is the real window. November is the smaller one.

Sort the boring blockers in week one, not the week you want work to start. VPN, software logins, permission levels, MFA, the NDA, hardware and USB rules. Provisioning always takes longer than anyone plans for.

Then name one person to own the move. Not the managing partner, who won’t have time. Not whoever looks least busy. Someone senior enough that people answer their emails. Give them four to six hours a week, and take something off their plate to make up for it. A handover with no owner drifts, and in week five you find out nobody wrote the SOPs because everyone assumed someone else had.

2. What Moves Offshore

Everyone says start with simple work. Nobody says how to decide what simple means. Ask three questions of every task on your list. Does it need someone physically present? Does it need a U.S. license? Does it need the client relationship?

Three noes and it moves. That covers data entry, workpapers, reconciliations, month-end close, AP and AR runs, prior-year comparisons, depreciation, and apportionment. Chasing client documents is a half yes: they draft, you send.

Three things never move. Judgment on an uncertain position. The planning conversation. Final review and signature. Run the exercise twice, once for tax and once for bookkeeping, because the answers come out differently.

3. Written Scope

Most scope arguments in month three happen because nobody wrote two pages in month zero. Split the list into phase-two work, phase-three work, and later. Then write down:

• Which clients and which schedules, by name. Not “1040s.”
• Volume per week, and what happens when it spikes.
• Turnaround expected per return type (in days).
• Budgeted hours per return type, so both sides see when something runs long.
• What finished means. Which schedules complete, where the file saves, what it gets named.
• Who reviews, and how fast they commit to turning it around.

That last one matters. If you want a two-day turnaround from the preparer, you owe them one back. Revisit the entire document at day 60, when you know what the work actually takes.

4. Documentation and Workflows

For each task, six things. What kicks it off? The steps in order. The decision points and your decisions. What finished looks like: Which folder it lands in and what it’s named. Who reviews it. Six fields, one page. Three pages is a policy manual, and nobody reads those.

The hard part is the forks. Every process has a moment where it depends, and the depends part lives in one senior person’s head. Ask them what happens when the K-1 arrives late, when the bank feed breaks, when the rental sold mid-year. Write the answers down. Otherwise, you have built a plan where one specific person has to show up.

On the bookkeeping side, four things need to be written down once: the chart of accounts and how you actually use it, recurring entries, bill coding rules, and naming conventions. That is where most bookkeeping rework comes from.

Record instead of writing wherever you can: A three-to-five-minute screen recording, someone talking while they click, kills a question that would otherwise cost a day. Keep it in one place, a workflow-template library or a shared drive folder, so the preparer has it on day one.

Here is the test for whether an SOP is done. Hand it to someone in your own office who has never done that task and watch them try. Every question they ask is a hole. Twenty minutes of that saves a week later.

5. Team Mapping

Say you are bringing on two offshore tax preparers and a bookkeeper. Map each one to a named person in your office. Not “the tax team.” A person who answers their questions, reviews their work, and runs the weekly call. Then name a backup, because people get sick in the last week of March.

At three or four per anchor the anchor becomes the bottleneck and review slows for everyone.

Pick the anchor on reachability, not seniority. A Slack or Teams channel per pairing, and questions go there, never to whoever happens to be online. That one rule kills most of the conflicting-answers problem.

6. Client Disclosure

The AICPA Code, at ET 1.150.040 and 1.700.040, requires that you tell the client, preferably in writing, before sharing confidential client information with a third-party service provider. You either contract with the provider for confidentiality or get the client’s specific consent. Most firms do both. On tax work, §7216 sits on top. The Tax Adviser has a clear write-up.

In practice: add the language to your engagement letter for anyone signing new engagements, send a short standalone notice to existing clients, and keep the email acknowledgment. Do it before any live file moves.

7. The Review Loop

This is the piece firms build in March, badly, under pressure. Build it now. Four templates and one habit:

• A preparer note on every file. Three lines: what I found, what I assumed, what needs your confirmation. Cuts review time more than anything else here.
• Four labels on your review notes. Correction, information, preference, teaching. A preparer who can’t tell a preference from an error reads every note as a failure and stops telling you when they’re unsure.
• A rating out of five on every returned file. Accuracy, completeness, and whether the notes saved you hunting for context. Fifteen seconds, and after a month you have a pattern instead of an impression.
• A stuck rule. Fifteen minutes of their own research, 10 in the knowledge base, then park it in a fixed format and move to the next return. Never sit waiting.

The habit is the weekly one-to-one, same slot, held even when it feels pointless. How all this runs day-to-day is a longer subject on its own right.

Days 31 to 60: Hand Over

8. Phased Release

The failure is always identical. Five files come back fine, so next week somebody sends 40. Reviewers drown, feedback stops, and the preparer repeats the same mistake because no one corrects it.

So, step it up. Two practice returns from last year first, marked by the anchor. Then five to 10 live files, simplest clients only, with every line reviewed. Then 20 to 30 with more complexity, and review moves from every line to targeted. Then normal load.

Move to the next batch by number, not by date. If you built the review loop above, you already have the numbers:

• First-pass clearance, meaning files that clear review with nothing sent back. You want 70% and climbing.
• Files needing two or more review cycles. Under 15%.
• Actual hours against the budgeted hours in your scope doc. Within 20%.
• Queries left open longer than one shift. Should be zero by the end of each overlap.
• Average file rating. Four or better, with no twos.

Pick numbers you can pull in five minutes. A scorecard nobody updates is worse than none.

Talk every day in this phase, even for 10 minutes. That is when habits set. If the preparer has gone quiet, that isn’t confidence; that’s guessing.

Days 61 to 90: Settle In

9. Client Contact Ladder

Most firms treat this as binary. The offshore person never speaks to clients, or one day they do. Move them one rung at a time, on evidence.

Client disclosure has to be done before rung 3, not rung 7. Most firms should stop at rung 3 or 4 in year one, and that is fine.

10. Season and Off-Season

After 90 days, your loop should change shape twice a year. Off-season is when you add scope, onboard people, cross-train onto new return types, rewrite SOPs, and let people dig when they get stuck. In season, freeze all of it: no new scope, no new people, nobody working outside what they know, no process changes. Research gets time-boxed and parked, the one-to-one drops to 15 minutes and covers only blockers, and nobody moves up the client ladder between January and April.

If It Already Went Badly

Before you fire the provider, check four things. Was the scope written down? Were the SOPs written down? Did every offshore person have a named onshore anchor? Did the work go out in batches?

If any answer is no, you don’t have a provider problem. You have a handover problem, and switching providers just restarts the same experiment with new people. Do the unglamorous first 30 days in the summer, hand over through the autumn, and let filing season be the thing you are ready for instead of the thing you are learning during.

Sam Roger is the co-founder of Acculink CPA, which builds dedicated offshore accounting and tax teams for U.S. CPA firms.

Photo courtesy Karolina Grabowska for Unsplash+