Practice Management · Guide · Working level
The CPA firm's playbook for vetting specialty tax providers
A firm-side diligence process for R&D credit, cost segregation, and energy incentive boutiques: what deliverables to demand, how to read fee structures, reference questions, red flags, and a scoring rubric before you refer a client.
When a client needs a cost segregation study, an R&D credit study, or a Section 179D allocation, most CPA firms refer the work out — and then sign the return that carries the result. That signature is the whole problem. The referring firm keeps the preparer-penalty exposure, the malpractice exposure, and the client relationship, while the boutique keeps the fee. Vetting the provider is therefore not a courtesy to the client; it is the firm's own risk management, and it deserves the same rigor as a new-client acceptance decision.
We have written elsewhere about how a taxpayer should choose a specialty tax provider. This piece is the firm-side companion: the diligence a CPA practice runs before putting its name — and its Section 6694 exposure — behind someone else's study.
Why the referring firm carries the risk
Three exposures survive the referral. First, preparer penalties: under Section 6694, the signing preparer answers for any position on the return lacking substantial authority (or reasonable basis with disclosure), regardless of who computed it — see our companion guide on preparer penalties under Section 6694. Second, professional liability: when a study collapses on exam, the client's suit typically names everyone, and the CPA who "recommended those people" is the defendant with the deepest relationship and often the deepest insurance. Third, Circular 230: a practitioner who relies on another person's work must exercise reasonable care in doing so — Circular 230, §10.22, allows reliance only where the reliance is itself reasonable. The fundamentals of Circular 230 apply to the referring CPA whether or not the boutique's engineers are practitioners at all.
The diligence process below is designed to discharge all three at once — and, critically, to leave a paper trail proving it.
What to demand before the first referral
A complete sample deliverable
Ask for a full, redacted study of the same type and roughly the same size you would refer — not a marketing excerpt. Then actually read it against the government's own expectations. For cost segregation, the IRS Cost Segregation Audit Techniques Guide describes what examiners look for: an engineering-based methodology, unit-by-unit asset detail, reconciliation to total cost basis, cited authority for classifications, and photographs or site-visit evidence. For R&D studies, look for business-component-level analysis mapped to the four tests of Section 41(d), named individuals and their qualified activities, a defensible estimation method for wage allocations, and enough detail to populate the Section G business-component reporting on the redesigned Form 6765.
Red flags in a deliverable: boilerplate narrative that could describe any company; percentages with no visible derivation; no reconciliation between the study total and the return; and a "report" that is mostly a description of the law rather than an application of it to this taxpayer.
The people, not the brand
Boutiques sell a brand and staff with whoever is available. Ask who will perform this client's study, their credentials (engineering degrees for cost seg; relevant technical background for R&D), their tenure, and who reviews their work. Ask what fraction of the firm's staff sits offshore and what the onshore review layer actually does. A shop that will not name its people before engagement is telling you something.
Exam-support terms, in writing
This is the single most differentiating question. Get contractual answers to: Who responds to an information document request about the study? Within what period? Through what stage — exam only, or through IRS Appeals? At what cost — included, capped hourly, or full freight? For how many years after delivery, and does the commitment survive the departure of the people who did the work or a sale of the boutique? Providers with defensible work sell audit support cheaply because they rarely have to deliver it. Providers who charge heavily for it, or promise it only orally, have priced in the expectation of exams going badly.
Track record you can verify
Ask for exam statistics — number of studies examined, sustention rates, adjustments conceded — and treat unverifiable claims ("we've never lost a credit") as puffery. Then ask for references, specifically CPA firms whose clients have been through examination with the provider's study in the file. A reference whose clients have never been audited can only tell you about sales experience.
Reading the fee structure
Fee design tells you how the provider's judgment is bent. The incentives are not subtle.
How common fee structures shape provider behavior:
| Fee structure | Provider's incentive | Firm-side concern |
|---|---|---|
| Fixed fee, scoped up front | Do the agreed work efficiently | Scope may be thin; confirm deliverable spec in writing |
| Hourly with cap | Thoroughness up to the cap | Least biased; verify seniority of hours billed |
| Percentage of credit or deduction found | Find the largest possible number | Rewards aggressive quantification; diligence trigger |
| "Success fee" on amended-return refunds | Maximize the refund claim | Circular 230 §10.27 contingent-fee restrictions may apply |
| Free study, fee only if you file | Get every prospect to file something | Selection pressure toward marginal claims |
Contingent arrangements are common in the boutique market precisely because many boutiques position themselves outside Circular 230's practitioner definition. Whether that positioning holds is a separate question — our brief on contingent fees under Circular 230 walks through §10.27 — but for vetting purposes the point is simpler: a fee proportional to the answer is pressure on the answer. If you refer to a contingent-fee shop, your review of the deliverable must be correspondingly harder, and your file memo should say why the structure did not, in this case, distort the result.
Reference questions that actually discriminate
Skip "were you happy with them." Ask questions with falsifiable answers:
- Has a client of yours been examined on this provider's study? What happened, step by step?
- Did the provider honor its exam-support terms without renegotiating the fee?
- Did the deliverable arrive with everything you needed to prepare the return — Form 6765 Section G detail, Form 4562 class lives, 481(a) adjustments — or did your staff reconstruct it?
- Have they ever told one of your clients the credit or study was not worth pursuing? (A shop that has never turned down a prospect is running a volume model.)
- Would you refer your largest client to them? Your most audit-sensitive one?
Red-flag phrases
Certain sales language correlates reliably with weak work. Treat the following as diligence accelerants:
- "Every business qualifies" or "if you have employees, you have a credit." Section 41 is activity-based; this is the pitch the IRS itself has flagged in its annual scam warnings.
- "Audit-proof." Nothing is. The honest formulation is "documented to withstand examination."
- "The IRS never looks at these." Both false and irrelevant — exam probability is not authority.
- "We only need two hours of your client's time." A study built on a two-hour interview is an estimate wearing a report's clothing.
- "Our fee comes out of the refund, so it costs nothing." A contingent fee dressed as free money.
- Refusal to put exam support, methodology, or sample deliverables in writing.
A scoring rubric for the vetting file
Reduce the process to a form so every partner runs the same diligence and the file shows it. Weight the categories by what actually predicts exam outcomes.
Suggested provider-vetting rubric (score each 1–5; weighted total out of 100):
| Category | Weight | What a 5 looks like |
|---|---|---|
| Deliverable quality (sample reviewed) | 25 | Engineering/analysis visible, reconciles to return, meets ATG expectations |
| Exam-support terms | 20 | Written, through Appeals, included in fee, survives personnel changes |
| Staff credentials and continuity | 15 | Named professionals, relevant credentials, low turnover |
| Fee-structure integrity | 15 | Fixed or hourly; contingent only with documented mitigation |
| Verified references and exam history | 15 | CPA references with actual exam outcomes |
| Responsiveness and preparer usability | 10 | Deliverable maps cleanly to forms; questions answered in days |
Set a floor — for instance, no referral below 70, and no referral at any score if deliverable quality or exam support scores below 3. The rubric is not the point; the discipline and the record are.
Illustrative weights from the rubric above; adjust to your firm's risk appetite.
Papering the referral
Once a provider clears the rubric, the referral itself should be documented in three places.
The vetting memo. One memo per provider per year: what was reviewed, who reviewed it, scores, and the approval. This is the reliance foundation for every return the firm signs on that provider's work.
The client communication. In writing to the client: the firm is referring, not subcontracting; the provider is the client's engagement; the firm will review the deliverable for return-preparation purposes but is not re-performing the study; and fees run directly between client and provider (which also keeps the firm clear of referral-fee and independence complications — see managing boutique relationships).
The engagement letter. The firm's own letter should scope the firm's role — preparation of the return incorporating the specialist's deliverable — and include a reliance-on-specialist clause. Our brief on engagement letters for specialty tax work covers the drafting.
Then, when the deliverable arrives, review it and document the review before signing. What that review file should contain is its own subject — workpaper standards for credit claims you didn't build — but at minimum: tie-out of study numbers to the return, a read of the methodology against the four tests or the ATG, and written answers to the questions the reviewer asked.
When not to refer at all
Neutrality cuts both ways. Some engagements should not go to any boutique: credits too small to survive the fee (a $12,000 credit behind a $9,000 study fee helps nobody but the provider); clients whose documentation habits make any study an estimate; positions on years whose refund statute under Section 6511 is nearly closed; and industries where the case law is running hard against the claim. A specialty provider will rarely tell your client no — the vetted ones sometimes will, which is itself a vetting signal — so the screening function belongs to the referring firm. That screening discipline is covered in client screening for incentive work.
The playbook, compressed: sample deliverable read against the ATG or Section 41, people named, exam support in writing, fee structure understood, references with exam scars, rubric scored, memo filed, referral papered, deliverable reviewed. It is a half-day of work per provider per year. Signing a return without it is lending the firm's license to a stranger.
Frequently asked questions
- What should a CPA firm ask a specialty tax provider before referring clients?
- Ask for a full redacted sample deliverable, the credentials of the people who will actually do the work, the firm's exam-support terms in writing, its fee structure and whether any component is contingent, its exam and appeals track record, and two references from CPA firms whose clients have been through examination with the provider's study.
- Are contingent fees a red flag when vetting an R&D credit or cost segregation provider?
- They are a diligence trigger, not an automatic disqualifier. Circular 230 restricts contingent fees for practitioners in most original-return contexts, and a fee tied to the size of the credit rewards aggressive quantification. Ask how the fee is computed, whether the provider is subject to Circular 230, and whether the same study would look different under a fixed fee.
- Does referring a client to a specialty provider protect the CPA who signs the return?
- No. The signing preparer remains exposed under Section 6694 for unreasonable positions on the return, and reliance on a specialist is reasonable only if the reliance itself was reasonable — meaning the CPA vetted the provider, reviewed the deliverable, and had no reason to doubt its conclusions. A referral without diligence transfers no risk.
- What is a reasonable exam-support commitment from a specialty tax firm?
- In writing, and specific: who responds to information document requests, within what timeframe, at what cost, and for how many years after the deliverable. Strong providers include audit defense of their own study through Appeals at no additional fee. Vague phrases like 'we stand behind our work' with no contractual mechanics are the weakest form.
- How often should a firm re-vet its specialty tax referral partners?
- Annually, and immediately after any client examination involving the provider's work. Personnel, ownership, and quality drift at boutiques faster than at accounting firms — a shop that produced defensible studies three years ago may now run a templated volume model. Treat the relationship like a continuance decision, not a permanent endorsement.