The channel split is clean and the per-client price gap is roughly 20x: business-facing AI bookkeepers charge 11-67% of a Client Advisory Services practice's revenue per client, firm-facing tools charge 0.4-3.4%, and not one published firm-facing price is per staff seat
The question
"What do Pilot, Receipts AI, Karbon and the newer AI-bookkeeping entrants actually charge per client per month, and do they sell to the end business or into the CPA firm?"
Derivative of [[2026-09-01-cpa-bookkeeper-aging-cohort-tooling-spend]], which found that the displaceable prize in a small accounting firm is labor (28.7-34.4% of net client fees) and not tooling (4-8% at best, and that figure traced to a vendor blog rather than to the AICPA). The channel an entrant sells through determines which of those two lines it is competing for, and the displacement map is different in each case.
What we already know (from the vault)
- The parent brief's ratios are the measuring stick for everything below, with one caution about mixing surveys. For firms in the $500K-$5M net client fees (NCF) bands, the 2025 AICPA MAP Survey puts non-owner salary at 28.7% / 30.3% / 34.4% of NCF. Separately, the 2024 CPA.com / AICPA PCPS Client Advisory Services (CAS) Benchmark Survey puts average revenue per client at $17,867/year ($1,489/month) against a median of 67 clients and median CAS NCF of $1,606,409. These are two different surveys and two different populations; every figure in this brief that combines them is labeled ESTIMATE, cross-survey. [[2026-09-01-cpa-bookkeeper-aging-cohort-tooling-spend]]
- The parent brief recorded Karbon pricing as secondary and asked for verification; its labels were wrong. It carried "roughly $79-$99 per user per month on annual contracts," and explicitly noted Pilot, Bench and Receipts AI pricing "was not located in this pass and is not estimated here." [[2026-09-01-cpa-bookkeeper-aging-cohort-tooling-spend]]
- The firm denominator is small, which caps the firm-facing channel's addressable market. The entire US accounting-firm owner universe is 120,085 firms, and in the acquirable band of NAICS 541211 ($500K-$2.5M receipts) payroll is 41.5% of receipts. [[2026-10-30-cpa-firm-count-and-age-curve-primary-denominators]]
- The vault already holds Pilot's positioning claim, from a third party, in exactly the terms this brief tests. The Acquired episode on Amazon describes Pilot as integrating "with stripe brex Gusto Shopify Square" so they "take literally all of the headache of Finance and Accounting off your plate" - a whole-relationship claim, not a tooling claim. [[2026-04-19-acquired-amazon-com-transcript]]
- The acquisition funnel's bookkeeping/CPA lane is still live and still tax-shaped, which is the weaker side of this map. The Palm Beach candidate carries $150,000/year of monthly accounting across 20-30 clients. [[2026-08-30-shortlist-refresh-scored]], [[2026-10-01-shortlist-refresh-scored]]
What the web says
All vendor prices below were read off the vendor's own live page on 2026-10-31 and are VERIFIED-PRIMARY unless tagged otherwise. Per-client percentages are Ray arithmetic against the $1,489/month CAS revenue-per-client benchmark and are ESTIMATE, cross-survey.
- Pilot sells to both, through two separate products, and only prices one of them. pilot.com/pricing is keyed to the client's monthly expenses via a slider: Essentials $99/month (AI-only, explicitly no human bookkeeper, contact-sales call to action), Core $299/month at $0-12K monthly expenses stepping to $499 / $559 / $679 / $799 / $999 as expenses rise, then "Custom plan" with no price above $200,000 monthly expenses. Core's only stated difference from Essentials is "a US-based bookkeeper." Onboarding is "a charge equal to one month of bookkeeping." Tax is separate (single-member LLC from $1,000/year, C-Corp from $2,450/year) and requires Pilot bookkeeping; CFO services run $1,750-$5,250/month. The direct side is B2B-direct with founder testimonials and a referral program only (/refer-a-friend, $500 gift card to the referrer, $500 off for the referee) - Pilot keeps the client. The firm-facing product is Meridian (meridian.pilot.com, "POWERED BY PILOT," launched June 2026 per CPA Practice Advisor, 2026-06-23, REPORTED), "the AI operating system for accounting firms," explicitly white-label ("Requests go out under your firm's name") and pitched as "One accountant, a $1M book... Revenue per accountant goes up." Meridian pricing is NOT PUBLISHED - the only call to action is "Schedule a call."
- "Receipts AI" is two different companies, and the firm-facing one prices per client workspace at the floor of the market. ReceiptsAI (receiptsai.com, operated by PMC Technologies OÜ, Tallinn) publishes a Personal ladder (free / $12 / $24 / $39 per month) and a Pro ladder for practitioners: $79/month for 10 client workspaces and $149/month for 25, keyed to processing credits ("1 processed page = 1 credit") plus workspace and seat caps. That is $7.90 and $5.96 per client per month (ESTIMATE, division). Its pricing page says outright: "Are you a bookkeeping professional? Switch to plans designed for managing more clients." Channel = BOTH. Receipt AI (receipt-ai.com, Big Green Company, the text-a-receipt product) publishes no price on its own site - /pricing and /plans 404 and pricing sits behind the signup wall; the only near-primary figures are its vendor-supplied Intuit QuickBooks App Store listing at $29 / $99 / $299 per month, keyed to pages per month plus member seats, with "Add members for $29 each." Channel = business-direct-led, with an accountants page but no multi-client dashboard or per-client pricing. Third-party directories list it at $19/$20/$39 under plan names that exist in no vendor source - stale, do not cite.
- Karbon is firm-only, per staff seat, and the vault's figure was mislabeled rather than wrong. karbonhq.com is canonical (karbon.com/pricing 404s). Team $59/user/month annual, $79 monthly; Business $89/user/month annual, $99 monthly; Enterprise custom. The widely cited $1,068/user/year is exactly $89 x 12, the Business tier at annual billing, corroborated by Karbon's own return-on-investment calculator ("Cost of Karbon Business Plan $53,400/yr" for 50 employees). So the parent brief's $79-$99 band was the monthly-billing rates, not the annual ones. Client Portal, Document Management, Time Tracking and Billing are included in all tiers; tier gates are usage caps, not feature walls. Karbon AI and Karbon Practice Intelligence carry no published price at all ("pricing depends on your Karbon subscription and firm needs... Contact Karbon"). Channel is unambiguous: "Practice Management Software for Accounting Firms," "Trusted by 40,000+ accounting professionals," and the pricing FAQ states "Pricing is charged per user. Every team member will need to be invited as a user." Clients reach the firm free through the portal. There is no published path for an end business to buy Karbon.
- Across 14 newer entrants, 9 publish a real price and 5 do not. Business-facing published prices cluster tightly: Fondo from $399/month (keyed to close cadence and startup stage, not expense band), Zeni $549 / $799/month (10% off annual), Bookkeeper360 from $400/month, Puzzle's "Software + Service (Expert Reviewed)" $171/month up to $10K monthly expenses on top of software at $30-$360/month, Inkle $99/month up to 30 transactions. Firm-facing published prices: Double (the former Keeper; keeper.app now redirects to doublehq.com) at $10 / $25 / $50 per client per month, plus an Autoclose agent add-on at $500/month firm-wide and a Tax Suite at $200/month; Uncat $9/month/client; Black Ore "starts at $60/return"; Bookkeep $19-$1,199/month per entity (ecommerce journal-entry automation, a per-entity unit rather than a per-client one).
- The five that hide pricing are not randomly distributed. NOT PUBLISHED and demo-gated: Basis (getbasis.ai, "Basis does the work," $100M Series B at $1.15B reported Feb 2026), Truewind ("Your team owns the review. Truewind handles everything else"), Fieldguide ($75M Series C at $700M reported Feb 2026) - all three firm-facing; Numeric, which sells to corporate controllers and internal finance teams rather than to firms; and Decimal, the one business-facing holdout, which still advertises "fixed monthly pricing... no hidden fees" with no figure.
- Not one published firm-facing price is per staff seat. Every one meters per unit of client production work. Per client (Double's core tiers, Uncat, ReceiptsAI Pro), per entity (Bookkeep), or per completed tax return (Black Ore). The single exception within those vendors is Double's Autoclose agent at a firm-wide flat $500/month. Karbon, the incumbent practice-management layer, is the only per-seat vendor in the entire set. No evidence surfaced on whether these entrants ever priced per seat, so this is a description of the current market and not a documented shift.
- Bench is confirmed dead, confirmed acquired, and is the cleanest unit-economics datapoint in the segment. Bench's own site notice, captured in the Wayback Machine on 2024-12-27: "as of December 27, 2024, the Bench platform will no longer be accessible." Employer.com announced the acquisition on 2024-12-30 (press release hosted on bench.co, Jesse Tinsley quoted); Employer.com's own release says operations suspended December 28, one day off Bench's own notice. Bench Accounting Inc. and subsidiary 10Sheet Services Inc. each filed an assignment in bankruptcy in Canada on 2025-01-07, trustee KSV Advisory, largest creditor National Bank of Canada. From the bankruptcy filings via TechCrunch (near-primary): $135M burned 2012 to Sept 2024, revenue $42M (FY to Mar-2023) on a ~$30M loss, then $49M with the loss roughly halved; $51M owed to National Bank, which signed a funding-and-forbearance agreement on 2024-12-12, thirteen days before the collapse. Bench's marketing claimed "35,000+ American small business owners"; Employer.com told TechCrunch the real figure was ~12,000. Pre-shutdown pricing (Wayback of bench.co/pricing, 2024-12-13): Bookkeeping $299/month annual, Bookkeeping and Tax $599/month annual - up from $399 in the 2024-10-10 snapshot, a sharp increase in the final quarter. The site is live and selling today under Employer.com at $199 / $399 / $649 plus a fractional bookkeeper at $55/hour with $1,200 onboarding. Its shutdown notice recommended a rival, Kick.
- Bench is not the only death in this exact shape. Finally (finally.com, $50M Series B plus a $150M credit facility from PeakSpan reported Sept 2024) posted an Assignment for the Benefit of Creditors notice, wound down 2026-09-08, and sold its bookkeeping book all-cash to Inkle effective 2026-10-28 with roughly 100 staff transferred (REPORTED: Accounting Today, CPA Practice Advisor). Two venture-funded, flat-monthly-fee, human-backed bookkeeping providers have now failed roughly twenty months apart.
Convergences and contradictions
- Convergence, and it narrows the parent brief's widest bracket. The parent could only place the firm software line inside a 60x-wide range. Published prices now bound the AI-and-practice-management slice of it. Take a median CAS practice at 67 clients and an implied 8.85 professionals (ESTIMATE: $1,606,409 NCF divided by $181,440 NCF per professional - note that $181,440 is the parent's figure for the technology-committed cohort, which also serves 100 clients rather than 67, so the headcount here is a lower bound and the resulting software percentage is a floor). Buying the fullest published stack available to it - Double Plus at $25/client/month ($20,100/year), Autoclose at $500/month ($6,000/year), Karbon Business at annual billing ($9,452/year) - costs $35,552/year, or 2.2% of NCF. Against non-owner salary of 28.7-34.4% of NCF ($461,000-$553,000), the labor line is 13-16x the published AI-and-practice-management stack (ESTIMATE, cross-survey). This does not refute the parent's unverified "4-8% of revenue" technology figure, because that figure covers the ledger, tax-prep software, hardware and cloud as well; it does establish that the AI layer specifically is a 1-2% line item, far below the labor line it is pitched against.
- Contradiction with the premise of the question: the two channels are not competing in the same market, and the per-client price gap is roughly 20x. Business-facing human-backed providers charge $171-$999/month, or 11-67% of what a CAS practice charges for the same client (excluded: Pilot Essentials and Inkle at $99, which are AI-only or transaction-capped rather than full service, at 6.6%; and note the top Core bands correspond to much larger clients than the CAS average, so the 67% end is not apples-to-apples). Firm-facing per-client tools charge $6-$50/client/month, or 0.4-3.4%. Comparing the dense bands - $299-$799 against $9-$50 - the gap is about 19x at the midpoints and spans 6x to 89x at the edges. Pilot is not trying to displace a firm's software budget; at $299-$799/month it is bidding for the entire client relationship at a quarter to a half the incumbent price. Double and Uncat at $9-$25/client/month are selling the firm a cheap instrument for keeping that relationship.
- Contradiction inside the labor-versus-tooling frame the backlog note asked us to apply: the pricing models are blurring the distinction. Karbon is a tooling seat that markets itself purely on labor ("$39,191 saved per employee, per year," "19.9 hours each week," benchmarked explicitly to a $81,680 median accountant salary), with no software-consolidation dollar claim anywhere on its page. Meanwhile every published firm-facing price meters per unit of client work, which indexes vendor revenue to client volume rather than to headcount. The practical consequence is sharp: a firm executing the capacity plan that 39% of MAP respondents with a capacity plan named ("expand client load without adding staff") holds its Karbon seat spend flat while its Double or Black Ore spend grows linearly with every client added. The tooling line is becoming a variable cost indexed to client volume. One hypothesis worth flagging rather than asserting: the fact that the agent layer specifically is the layer almost nobody prices in public - Pilot publishes six expense-band prices and nothing for Meridian, Karbon publishes three seat tiers and nothing for Karbon AI, Puzzle publishes both software and service prices but leaves its "For Firms" navigation as unbuilt placeholders, and Basis, Truewind and Fieldguide publish nothing - is consistent with a market that has not yet settled what the agent layer is worth.
Synthesis for RDCO
The channel question resolves cleanly, per vendor, and it resolves in favor of the parent brief's reframe. Selling to the end business (B2B-DIRECT): Pilot's pilot.com product, Fondo, Decimal, Numeric (to internal finance teams rather than small businesses), Inkle, and Receipt AI (receipt-ai.com). Selling into the CPA or bookkeeping firm (B2B2B-FIRM): Karbon, Double, Uncat, Black Ore, Fieldguide, Basis, and Pilot's Meridian. Both, through distinct products or tiers: ReceiptsAI, Puzzle (software direct plus an accounting-partners program where Puzzle bills the client and the firm pays nothing), Zeni, Bookkeeper360, Bookkeep, Truewind. The business-facing group is competing for the labor line and competing for it by taking the client outright, not by selling into the firm. The firm-facing group is competing on the tooling line on price while pitching the labor line in its copy. Notably, Pilot built a separate white-label brand rather than adding a partner tier to pilot.com, which is at least consistent with the two channels being unable to share a price list.
For the acquisition lane, the number that matters is not the tooling stack, it is the $299-$799. The Palm Beach CPA practice carries $150,000/year of monthly accounting across 20-30 clients ([[2026-08-30-shortlist-refresh-scored]]), which is roughly $420-$625 per client per month. Pilot Core at its lower expense bands, Fondo at $399, and the reborn Bench at $399 land inside that band as human-backed services, and Pilot Essentials at $99 lands at a quarter to a sixth of it for AI-only books. That is the competitive fact that belongs in diligence on any monthly-accounting book: the price floor for the service is now publicly posted by funded competitors, it sits at or below what the practice charges, and the practice's defense is relationship and tax adjacency rather than price. This strengthens the parent brief's conclusion that tax-heavy books are the safer acquisition while CAS-heavy books are the better margin story - the CAS-shaped revenue is exactly the revenue the business-facing entrants are priced to take.
The two deaths are the most useful finding for RDCO's own positioning. Bench realized roughly $4,083 per customer per year, about $340 per month (ESTIMATE: $49M revenue against the ~12,000 customers Employer.com acknowledged, which mixes a FY-2024 revenue figure with a Dec-2024 customer count), with human bookkeepers in the loop, and burned $135M without reaching profitability. It raised its combined bookkeeping-and-tax price 50% in its final quarter; that is suggestive about where management thought the problem was, but no Bench or Employer.com statement explains the failure, so motive is inference. Finally failed in the same shape roughly twenty months later. Meanwhile a CAS practice earns $1,489 per client per month, and MAP firms overall post a median $252,663 of net remaining per partner (all-firm median, not CAS-specific). The ~4.4x gap between Bench's realized revenue per client and a CAS practice's is the clearest quantity in this brief, and it is a pricing gap, not a demonstrated moat - no firm-failure data was gathered here, so "incumbents survive" is not established. What the evidence does support is narrower and still useful: two well-capitalized attempts to serve this segment at a third of the incumbent price died, which means the segment punishes low revenue per client and acquisition cost at least as hard as it rewards production efficiency. The firm-facing channel, where one sale reaches a firm that already owns 67 relationships, is the structurally better side of that trade even though its per-client price is roughly 20x lower.
Where this changes RDCO's own framing: per-client metering is the pricing model to copy. Double, Uncat, ReceiptsAI and Black Ore all independently landed on metering per unit of client work in a category whose incumbent is per-seat. The inference - and it is an inference, since no vendor states it - is that per-seat pricing caps vendor revenue exactly when the product succeeds: if the agent lets one accountant carry a $1M book, per-seat revenue falls while value delivered rises, and per-client metering captures the opposite. For any RDCO agent product sold into a services firm, that is the directly transferable lesson. It is worth noting the contrast with the flat $12K-plus-$500/month shape in [[2026-05-24-mostlymetrics-pizza-shop-ai]], which is a single-business flat fee rather than per-client metering, and therefore a counterexample rather than corroboration.
Why this is in the vault
It closes the parent brief's own open follow-up with verified list prices, and it supplies two things the acquisition funnel does not currently have: a published competitive price floor ($99 AI-only, $299-$799 human-backed per client per month) to test any monthly-accounting book against during diligence on the bookkeeping/CPA lane in [[2026-10-01-shortlist-refresh-scored]], and a bounded AI-tooling figure (2.2% of net client fees at full attach, from published prices) that keeps axis 1 of the 5-point fit test pointed at labor rather than software. It also retires the Bench uncertainty the parent carried as "recalled, not verified," and adds Finally as a second data point on the same failure shape, which bears directly on whether RDCO should ever price an agent-delivered service book on a flat monthly fee.
Open follow-ups
- What does Pilot's Meridian actually charge a firm, and is it per client, per close, or per seat? It is the only product in the set that is both white-label firm tooling and backed by a provider with published direct prices, so its price would reveal how one vendor values the same production work on each side of the channel line.
- What is the real attach rate for firm-facing per-client tools - what share of a firm's clients get put on Double or Uncat? The 2.2%-of-NCF figure assumes every client is enrolled. At a 30-50% attach rate on the per-client line (the Autoclose and Karbon lines are fixed), the stack falls to $21,482-$25,502, or 1.3-1.6% of NCF, and the labor-to-tooling ratio widens from 13-16x to 18-26x. Nothing public establishes which end of that range is real.
- Did Bench's ~12,000 clients churn to competitors or stay with Employer.com, and which competitors publicly won them? Kick was named in Bench's own shutdown notice; no post-acquisition retention figure has been published, so whether a displaced-demand pool of this size actually moved is unresolved.
- What is customer-acquisition cost for business-facing bookkeeping at the $299-$799/month price point, and was that the variable that killed Bench and Finally rather than delivery gross margin? No official post-mortem exists for either, and the gross-margin and churn figures circulating online trace only to aggregator post-mortems.
Related
- [[2026-09-01-cpa-bookkeeper-aging-cohort-tooling-spend]] - the parent brief; supplies the labor ratio and the CAS per-client benchmark this brief measures against
- [[2026-10-30-cpa-firm-count-and-age-curve-primary-denominators]] - the 120,085-firm denominator that caps the firm-facing channel's addressable market
- [[2026-10-01-shortlist-refresh-scored]] - the live bookkeeping/CPA acquisition lane this brief's price floor applies to
- [[2026-08-30-shortlist-refresh-scored]] - the Palm Beach candidate whose monthly-accounting book is priced against Pilot, Fondo and Bench above
- [[2026-08-30-pre-market-outreach-channel-plan]] - Channel 3 targeting, which this brief argues should favor CAS-shaped books
- [[2026-04-19-acquired-amazon-com-transcript]] - third-party description of Pilot's whole-relationship positioning
- [[2026-05-28-service-book-rollup-agent-first-conversion]] - the service-book conversion math these vendor economics stress-test
- [[2026-05-29-spine-as-a-service-productized-conversion-playbook]] - productized-service pricing shape
- [[2026-05-24-mostlymetrics-pizza-shop-ai]] - flat-fee small-business AI pricing anchor outside accounting; a counterexample to per-client metering
Sources
Vault:
- ~/rdco-vault/06-reference/research/2026-09-01-cpa-bookkeeper-aging-cohort-tooling-spend.md
- ~/rdco-vault/06-reference/research/2026-10-30-cpa-firm-count-and-age-curve-primary-denominators.md
- ~/rdco-vault/01-projects/acquisitions/2026-10-01-shortlist-refresh-scored.md
- ~/rdco-vault/01-projects/acquisitions/2026-08-30-shortlist-refresh-scored.md
- ~/rdco-vault/01-projects/acquisitions/2026-08-30-pre-market-outreach-channel-plan.md
- ~/rdco-vault/06-reference/transcripts/2026-04-19-acquired-amazon-com-transcript.md
- ~/rdco-vault/06-reference/2026-05-24-mostlymetrics-pizza-shop-ai.md
- ~/rdco-vault/06-reference/research/2026-05-28-service-book-rollup-agent-first-conversion.md
- ~/rdco-vault/06-reference/research/2026-05-29-spine-as-a-service-productized-conversion-playbook.md
Vendor pricing pages (VERIFIED-PRIMARY, retrieved 2026-10-31):
- https://pilot.com/pricing · https://pilot.com/faq · https://pilot.com/refer-a-friend · https://meridian.pilot.com (pricing NOT PUBLISHED)
- https://karbonhq.com/pricing · https://karbonhq.com/feature/analytics (pricing NOT PUBLISHED)
- https://receiptsai.com/pricing
- https://quickbooks.intuit.com/app/apps/appdetails/receipt_ai/en-us/ (NEAR-PRIMARY, vendor-supplied listing; receipt-ai.com publishes no price)
- https://puzzle.io/pricing · https://puzzle.io/accounting-partners
- https://fondo.com/pricing · https://zeni.ai/pricing · https://bookkeeper360.com (from-prices only)
- https://doublehq.com/pricing-plans?tab=firms · https://uncat.com · https://blackore.ai/pricing · https://bookkeep.com/pricing · https://inkle.io
- Demo-gated, no published price: decimal.com/pricing · numeric.io/pricing · getbasis.ai · truewind.ai · fieldguide.com
Bench and Finally:
- https://web.archive.org/web/20241227223423/https://bench.co/ (Bench's own shutdown notice, PRIMARY)
- https://web.archive.org/web/20250103144500/https://www.bench.co/ (platform restored)
- Wayback snapshots of bench.co/pricing, 2024-10-10 and 2024-12-13 (pre-shutdown prices, PRIMARY)
- https://bench.co/press-release (Employer.com acquisition announcement, 2024-12-30, PRIMARY)
- https://bench.co (live, 2026-10-31, current pricing under Employer.com)
- https://bench.co/blog/bench-news/employer-com-acquires-mainstreet (May 2025, "over 10,000 small businesses")
- https://ksvadvisory.com/experience/case/bench (bankruptcy trustee case page, 2025-01-07 assignment)
- https://techcrunch.com/2024/12/30/bench-to-be-acquired-after-abruptly-shutting-down/ (NEAR-PRIMARY)
- https://techcrunch.com/2025/02/05/bench-burned-through-135-million-before-shutting-down/ (NEAR-PRIMARY, from the bankruptcy filings)
- Accounting Today and CPA Practice Advisor coverage of the Finally wind-down and Inkle acquisition, Sept-Oct 2026 (REPORTED)
- https://www.cpapracticeadvisor.com (Meridian launch, 2026-06-23, REPORTED)
Paywalled, flagged and not retried: The Information, American Banker, Globe and Mail. GeekWire returned 403.