Boldstart Ventures' Shomik Ghosh on Kelly Partners' $KPG.AX unique accounting firm roll-up strategy
Summary
Kelly Partners Group is an accounting roll-up built around retaining, rather than cashing out, the acquired partner. KPG buys 51%, leaves the owner with 49%, and targets succession situations where accountants want liquidity and fewer administrative burdens without abandoning their client relationships. The promise, Shomik Ghosh says, is that joining KPG should let an owner “make more money than you did on your standalone basis.”
The underwriting depends on turning small firms with mid-single-digit margins into businesses approaching 30% margins. KPG directs 9% of acquired revenue to centralized technology, HR, billing, marketing, and sales, freeing highly paid accountants from collections and administration while supporting price increases and new-client growth. Ghosh believes the acquisition hurdle is roughly 20%, with typical prices near 4-5x EBITDA; management’s argument is that buying at even 8x EBITDA can become roughly a 3x acquisition after margin improvement.
Its claimed acquisition moat is a decentralized network of accountant-owners feeding opportunities into KPG. Those owners know nearby practitioners, share in KPG’s equity upside, and can approach succession candidates through trusted professional relationships rather than cold calls. Ghosh cited roughly 1,400-1,700 firms examined and about one-twelfth acquired: the selectivity and “decentralized scouts” matter because KPG targets firms with only $1-10 million of revenue, below the practical scale of many private-equity buyers.
International expansion is being seeded through concentrated Australian-expatriate communities, not broad countrywide entry. KPG began around Los Angeles—Ghosh thought perhaps Burbank—and then targeted selected areas in Florida, North Carolina, and Texas, with similarly specific market selection in the UK and Canada. The intended sequence is to establish an Australian-connected beachhead, develop local owner-scouts, and expand through their US professional networks.
The stock carries genuine compounder expectations despite its modest absolute size. Ghosh estimated KPG’s market value was probably around A$530 million, or “$300-something million” in US dollars, but warned that a displayed 17x EBITDA multiple understates the economics because KPG owns only 51% of acquired firms; his effective estimate was roughly 34-35x. CBIZ ($CBZ), at about $4.8 billion of enterprise value, over $1 billion of revenue, and roughly 20x EBITDA, shows how large an accounting consolidator can become but also how much execution is already priced in.
A potential US listing could improve acquisition credibility and financing, but it also creates temptation. US-listed shares may be easier for American sellers to understand and explain, while US reporting and regulation could broaden the banking pool beyond Australian lenders; acquisition debt currently sits at each acquired entity, as Ghosh described it. His red line is using that access to jump prematurely into a $50 million or $100 million acquisition: “That would scare me.”
Ghosh views AI as a three-to-five-year productivity tailwind for KPG, not an imminent replacement for trusted accountants. He calls AI an “enablement shift”: it can perform research and rules-based work, but client context, judgment, audit risk, and nuanced advice sustain the human relationship. Walker’s useful counter was that productivity gains may not simply compete margins away—the best practitioners could absorb displaced work, serve more clients, and earn materially more.
The cleanest bear case is management losing focus. Wealth management, estate planning, insurance, and other adjacent initiatives could extend the client relationship, yet they can also distract KPG from its repeatable small-accounting-firm playbook. Ghosh’s five-year failure diagnosis was blunt: “Lost focus”—either KPG moves upmarket too early or becomes enamored with an adjacent business while the acquisition engine stalls.
Deep dive
1. Kelly Partners turns accountant succession into a permanent partnership
Ghosh traces the model to Brett Kelly, an Australian chartered accountant who met goals set by his employer but was not made partner. That experience helped prompt him to start Kelly Partners in 2006, after an entrepreneurial youth in which he financed and published a book of interviews with Australian business and government leaders when no publisher would take it.
Kelly Partners now acquires accounting and tax practices, principally in Australia while expanding into the US, UK, and Canada. Its customers are individuals and small businesses—the accountants serving “Andrew and Shomik,” as Ghosh put it—not the firms handling Coca-Cola or similarly large corporate accounts.
The defining term is 51/49 ownership: KPG takes control while the selling accountant retains 49%. For an owner considering succession, the proposition combines liquidity, a succession path, and flexibility to reduce hours without severing client relationships or surrendering all future economics.
2. Small targets create the runway
KPG generally targets firms with $1-10 million of revenue, a fragmented segment too small for many conventional private-equity funds or for CBIZ’s increasingly larger acquisition needs. Ghosh sees that neglected scale as the runway: there are many potential sellers, and succession creates counterparties seeking a transition.
The typical hurdle is about 20%, which Ghosh translated into acquisition prices around 4-5x EBITDA. KPG intends to hold businesses for the long term, so the return is supposed to come from recurring cash flow, operational improvement, and repeated acquisitions rather than a later sale to another sponsor.
Walker’s challenge was fundamental: “You can buy anything if you just come and have enough money.” Unless KPG changes the acquired firm’s economics, a roll-up does not demonstrate the margin-expansion thesis; the thesis therefore lives or dies on whether the company can repeatedly transform under-managed practices.
Ghosh separates relative from absolute valuation. Relative to current earnings, KPG is demanding; in absolute terms, a company probably around A$530 million in market value addressing several countries might remain small beside the opportunity. CBIZ—about $4.8 billion of enterprise value and more than $1 billion of revenue—is his tangible example of how large an accounting consolidator can become.
3. The 9% central charge funds the margin-expansion machine
KPG directs 9% of an acquired firm’s revenue into centralized capabilities spanning technology, HR, marketing, sales, and other functions. The acquired accountants retain 49% and their local relationships, while the group supplies a business system that can improve operations at the acquired practice.
Ghosh’s cleanest example was accounts receivable. His accountant sends a letter and waits for a check; KPG accountants may send a message requesting $200 through Venmo. Standardizing and automating that process can improve cash conversion immediately while eliminating work accountants dislike.
Pricing is another lever. A client whose accountant understands carried interest, extensions, assets, and the surrounding tax context is unlikely to endure “the audit risk” and “brain damage” of switching merely because fees rise 5-10%. That relationship creates room for price increases.
Marketing supplies the growth leg. Many small accountants rely solely on word of mouth and stop accepting clients when personally full; KPG can position a niche practice—“the best VC accountant,” in Ghosh’s example—generate more demand, hire additional practitioners, and show the original partner how the business can increase earnings.
4. Retained ownership answers the post-sale effort problem—partially
Walker’s dental-roll-up analogy supplied the central pushback: after selling control, a practitioner owns only 49%, may work less hard, and could eventually open across the street after a non-compete expires. The relationships and knowledge belong heavily to the individual professional, so centralized systems alone cannot guarantee retention.
Ghosh’s first response was transaction structure. He believed KPG pays roughly one-third upfront and two-thirds in year two, creating a dynamic in which the seller has to keep working for at least two years. During that period, KPG can demonstrate its business system.
The economic bargain is not simply “sell and become an employee.” The accountant keeps 49%, sheds billing, HR, and marketing burdens, and can spend more time on work commanding perhaps $200 an hour. Walker’s example was replacing ten hours of an accountant’s administration with a $40-an-hour office manager, creating both more revenue capacity and more personal time.
Ghosh argues accounting resembles physician care more than dentistry because the perceived downside of switching is severe. A cavity is manageable; losing historical tax knowledge and then facing an audit could threaten “your house,” “your car,” and financial stability. That risk supports very high client retention.
5. Local accountant-owners become KPG’s acquisition scouts
The acquisition advantage is not only the back-office system; it is also a network. A private-equity firm could do something similar, but it would still need to cold-call sellers. A respected local accountant can approach peers encountered through referrals, professional relationships, and CPA meetups.
KPG encourages acquired owners to hold group shares, aligning them with the wider acquisition program. In Ghosh’s example, a Florida accountant serving venture capitalists might know another specialist serving athletes and say, “I just joined Kelly Partners—it’s pretty good”; each successful local deal can therefore produce the scout for the next adjacent niche or city.
Ghosh cited roughly 1,400-1,700 businesses examined and about one-twelfth acquired. That ratio suggests the model depends on selectivity as well as sourcing: the network widens the funnel, but the company must continue choosing acquisitions carefully.
Lawrence Cunningham reinforces the compounder analogy. Ghosh said Cunningham discovered KPG through a shareholder letter, bought shares, contacted Brett Kelly, and joined the board alongside positions at Marel and Constellation Software. His presence does not prove KPG can replicate Constellation, but it reinforces the comparison with decentralized, acquisition-led organizations.
6. International expansion starts with Australian cultural density
Walker’s pushback was that Australian scouts do not naturally know the best accountants in Florida or California. Porting trusted professional networks is not straightforward. KPG may need to pay fairly for its first several US firms before the local sourcing flywheel can begin.
Ghosh’s answer was geographic precision. KPG did not enter “California” abstractly; it began in Los Angeles and, he thought, perhaps Burbank, choosing areas with high concentrations of Australian expatriates. The first acquired practices could therefore serve familiar cross-border communities and preserve an element of KPG’s Australian culture.
The same approach informed selected entries into Florida, North Carolina, and Texas. Ghosh said the company was also being specific about where it operated in the United Kingdom and Canada, rather than entering those countries broadly. From an Australian-born accountant, the network might eventually extend to a US-born peer who grew up locally and has a good business.
Walker also noted that KPG had, he believed, recently reached the top 10 accounting firms in Australia excluding the Big Four. The unresolved issue is whether the cultural beachhead and decentralized scouting can scale beyond Australia while the company builds its US, UK, and Canadian playbooks.
7. The quoted multiple understates neither complexity nor risk
Ghosh warned listeners not to accept the headline figure that he believed was around 17x EBITDA. Because KPG owns 51% of many operating businesses, non-controlling interests complicate the accounting; his rough economic multiple was closer to 34-35x EBITDA. “It is not cheap by any metric.”
For comparison, he put CBIZ near 20x. KPG’s premium assumes a much longer runway and faster acquisition-led compounding, while CBIZ already operates at materially greater scale. Ghosh nevertheless thinks KPG “could be a 10x-type” investment if international replication works, but he repeatedly conditions that possibility on decentralized M&A and disciplined execution.
Walker noted that KPG’s November 2024 investor-day materials explicitly reconcile the non-controlling-interest accounting. Although the structure is complicated, the company “spoon-feeds” investors the adjustments, making the key debate the assumptions rather than access to the numbers.
Recent acquisitions also suppress reported margins because acquired firms initially enter below KPG’s historical profitability. That means near-term EBITDA growth can look weaker when acquisition activity is high; paying today’s multiple requires trusting historical evidence that the group can lift those practices toward historical norms.
8. A US listing could aid deals while inviting larger bets
Walker initially questioned why a long-term owner-operator would seek a potentially higher US valuation when KPG is not really issuing equity and has bought back shares in the second half of 2024, or so he believed. A less efficient Australian listing could benefit a disciplined repurchaser, while US compliance brings more scrutiny and cost.
Ghosh’s practical answer was seller credibility. An American accountant may find US-listed shares easier to understand and explain to a spouse than Australian shares or an ADR, especially if equity becomes part of the consideration. The listing can therefore function as acquisition infrastructure, not merely stock promotion.
Financing could broaden as well. Ghosh described acquisition debt as residing at each acquired entity rather than the parent; a US listing could give KPG access to US regulations, the Securities and Exchange Commission, and banks such as JPMorgan instead of leaving it dependent only on Australian banking relationships.
His concern is what management does with that access. If better financing leads KPG to conclude it can suddenly acquire a $50 million or $100 million revenue firm, the company enters a more competitive market and weakens its small-practice moat. “That would scare me.” The listing is useful only if it accelerates the existing playbook without redefining it.
9. Adjacencies can extend the model but endanger focus
Walker sees wealth management as a natural extension because tax accountants encounter wealth transfers and investment needs. Ghosh says estate planning and wealth planning are logical extensions, while insurance is more of a stretch. Kelly Partners also has an investment office in which partners and others buy stocks, private companies, or land.
Ghosh’s software analogy captures the trade-off: every new product requires its own team, marketing, distribution, and economics. “Multi-product execution is really difficult,” yet he argues that no successful compounding company has reached massive scale without eventually moving beyond its initial product and comfort zone.
Insurance worries both speakers more than estate or wealth planning. Financial companies repeatedly attach insurance operations and later separate them; likewise, if Kelly Partners decided to become a real-estate landlord and started buying properties, that would concern Ghosh.
The mitigating feature is sequencing. KPG is testing ancillary businesses in Australia while its newer geographies remain focused on acquiring accounting firms, building the playbook, and building the decentralized scout network. Ghosh expects Brett to reach some level of density before layering on adjacent products and new geographies.
10. AI may widen margins before competition normalizes them
Walker framed the tail risk directly: tax is a rules-based system, so an AI might ingest the equivalent of a W-2, produce a return that is “98% correct,” and leave only review to an accountant. Governments could also simplify filings for most earners, although betting against tax-code complexity has historically been difficult.
Ghosh responded that taxpayers earning under $50,000 are not the core clients of KPG’s acquired firms; it serves higher-net-worth individuals and small businesses with more context and exceptions. He also expects H&R Block and other companies to lobby against simplification, however undesirable that outcome may be for taxpayers.
His broader framing is that AI represents an “enablement shift,” not necessarily a platform shift. Just as ChatGPT lets him investigate why Regeneron was trading where it was and recursively explore unfamiliar issues, AI can support accounting research and rules-based work while humans supply judgment and client context.
Ghosh expects a three-to-five-year productivity and margin tailwind before competitors adopt similar tools and gains normalize. Walker’s pushback is more bullish: lower-tier practitioners may be competed away while top professionals use AI to handle more clients, analogous to lawyers earning far more without being proportionally better at law than predecessors.
11. Transparency must outweigh the compounder rhetoric
Walker’s largest qualitative warning was Brett Kelly’s use of Good to Great, Built to Last, Charlie Munger, Danaher, McDonald’s, and Mark Leonard. That language appeals to value investors but has also been associated, in Walker’s experience, with executives cultivating retail followings before issuing stock, enriching themselves, or presiding over businesses that ultimately failed.
Ghosh’s defense rests on consistency and disclosure rather than quotations. Kelly repeats the same operating ideas across years, while the owner’s manual gives extensive detail about the accounting, Kelly’s ownership, and what he sold each year. Ghosh put Kelly’s stake around 48% and said Kelly expects to remain above 35%, though whether the reduction comes through sales or issuance was unclear. Ghosh separately suggested that shareholders should identify important KPIs and check whether they are reported consistently.
The surrounding actions also matter: Kelly moved to the US to build the geography, has already appointed a successor known to the board, and has trained leaders beneath him. He reportedly imagines eventually becoming like Mark Leonard—growing “a giant beard,” retreating from public view, and leaving a system capable of generational compounding.
McDonald’s is the preferred operating analogy because it combines the franchise model, geographic expansion, ownership of the customer relationship, and attention to real estate. Walker noted that KPG discusses California and Florida, which he said are two of McDonald’s largest markets. But the analogy contains its own warning: if KPG’s investment office turns the accounting roll-up into an undisciplined landlord or investment vehicle, transparency should make that drift visible.
12. The five-year bear case is a broken acquisition engine
Asked what would explain a disappointing stock five years from now, Ghosh answered, “Lost focus.” The likeliest paths are a premature move into $50 million-$100 million firms or excessive attention to wealth management and other adjacent businesses while the acquisition engine stalls.
The acquisition engine is existential because small deals must remain numerous enough to move a growing company. If KPG cannot decentralize sourcing, it cannot indefinitely find sufficient $1-10 million practices at attractive prices; at roughly 34-35x EBITDA, any halt in accretive acquisitions could cause substantial multiple compression as well as slower earnings growth.
Ghosh therefore does not offer a precise fair value. He accepts a high compounder premium because he prefers proven compounders such as Constellation Software, Topicus, and Mastercard to cheap 5x-EBITDA businesses that are supposed to turn around. His past mistake was rejecting Constellation for a decade because the multiple always looked high.
The investable judgment is whether KPG remains early in a repeatable, long-duration system. High client retention, pricing power, low-cost referrals, local M&A networks, and margin improvement make the upside coherent; the premium makes monitoring unforgiving—track the acquisitions and KPIs and whether management continues doing what it said.