Consulting's Divide: Revenue Grows, Billable Hours Fade | SunTzu China

Executive Summary

Deloitte closed FY2025 with $70.5 billion in global revenue, up 4.8% in local currency. KPMG reported $39.8 billion, up 5.1%. EY reached $53.2 billion, up 4.0%. PwC posted $56.9 billion, up 2.7%. In the same period, Accenture's consulting revenue grew 1% in local currency in its fiscal third quarter, new bookings fell 3%, and the stock dropped roughly 18% on the day those numbers were published.

Read together, the two sets of figures describe the real position of professional services in 2026: money is still coming in, but it arrives through a different door. Hourly billing no longer carries the growth. Platform work, managed services and AI-assisted delivery are absorbing the new budgets.

China's data tells a similar story from the other end. The National Bureau of Statistics reported that value added in leasing and business services, the category that contains management consulting, grew 11.9% year on year in the first half of 2026, the fastest of all service sectors and more than double the 5.2% growth of services overall. Demand has not shrunk. What has shrunk is the marginal value of selling people by the month.

Three conclusions:

1. Revenue and headcount have decoupled. The revenue curves at the top firms keep climbing; the headcount curves no longer follow. Accenture cut roughly 11,000 roles in 2025 at a severance cost of $865 million, while publishing its AI and data workforce as a growth metric. Scale is no longer the moat. Revenue per delivered head is.

2. China is reallocating share, not losing demand. With value added in the sector growing at double digits, foreign consultancies face pressure from local firms, consulting teams inside the Big Four, in-house strategy units and AI vendors at the same time. Budgets did not disappear; the signatory changed.

3. Talent structures are turning from pyramid to dumbbell. Junior analyst roles are being eaten at the base by AI tools, while high-value hybrid roles combining industry knowledge, data capability and delivery ownership keep repricing upward. The middle layer has the hardest time: it is neither the cheapest nor the scarcest.

1. Revenue and Headcount Have Decoupled

At the surface level, all four international professional services networks grew revenue in FY2025. None posted a decline, an unusual result in a year of broad macro pressure.

The quality of that growth differs. Deloitte's 4.8% and KPMG's 5.1% are respectable in an inflationary foreign exchange environment, but the second number matters more: headcount is expanding far more slowly than revenue. When revenue outpaces headcount, revenue per person rises by default, and that gap has to be closed by a change in how work is delivered rather than by longer hours.

Deloitte moved from 460,000 people to more than 470,000 in FY2025, a net gain of roughly 10,000, against $70.5 billion in revenue. Each added person corresponds to roughly $7 million in incremental revenue. Arithmetic like that only works when platform products, reusable assets and AI-assisted delivery all pull in the same direction. The old equation of billable rate multiplied by billed days cannot produce it.

2. The Accenture Signal: One Quarter Repriced the Industry

If the Big Four fiscal-year figures show the stock, Accenture's fiscal third quarter shows the flow.

On June 18, 2026, Accenture reported results for the quarter ended May 31: revenue of $18.72 billion, up 6% in US dollars and 3% in local currency; diluted earnings per share of $3.80, up 9%; free cash flow of $3.6 billion; and 104 client bookings of $100 million or more year to date, up 13%. On its own, a steady quarter.

The market reacted the other way. The stock fell about 18% that day to $127.98. The explanation sits in two other lines.

Consulting and managed services are different businesses. The first bills by the day and by the project, so revenue equals people multiplied by days. The second charges for a service cycle, with revenue generated by platforms, processes and automation, which severs the link between headcount and revenue. Inside one company, the gap between 1% and 5% in local currency measures how fast the billable-day model is exiting. Overall bookings fell 3%, which tells you clients are still spending, just moving money from people to systems.

That explains Accenture's 2025 moves: cutting roughly 11,000 roles while absorbing $865 million in severance, and simultaneously reporting its AI and data workforce as a growth metric. In June 2026 the firm agreed to acquire a majority stake in OT security vendor Dragos and to acquire runZero and NetRise outright. Those are platform assets that can be contracted, renewed and replicated, rather than another consulting bench.

For the talent market the implication is concrete. If your output can be replaced by a model response, your pricing power is eroding. If your job is to frame the question, own delivery and turn AI output into a deliverable a client will sign, your pricing power is rising.

3. China: Demand Expanding, Share Reallocating

China's story differs from the global headquarters narrative. The National Bureau of Statistics data shows services value added up 5.2% in the first half of 2026, with leasing and business services up 11.9%, the fastest among service categories. In June alone, the production index for that category rose 9.7% year on year. Across the 14th Five-Year Plan period, value added in the category grew at an average of 10.7% a year, 5.3 percentage points above average real GDP growth.

Demand has not contracted. What has contracted is the logic that priced foreign consultancies in China. For two decades, they priced on three assets: global methodology, a multinational client network and a trust premium attached to a headquarters recommendation. All three are depreciating in 2026.

Start with methodology. When a large model can generate a structurally complete industry framework in ten minutes, the premium clients pay for frameworks falls. Then the client network: decision rights in China are shifting toward domestic companies, outbound players and industrial groups that care more about where you have done the work than where your head office sits. Then the trust premium, which the Big Four have felt most acutely. In 2024 PwC lost dozens of A-share and Hong Kong clients after an audit scandal. According to a tally by Yicai at the time, more than 60% of the clients that disclosed a new auditor moved to the other three firms. Same clients, same budgets, different signature.

Delivery quality itself has become a new variable. In August 2026, media reported that a cyber security report issued by a professional services firm was found to contain AI-generated content with fabricated concepts and invented citations, and a comparable report released in Canada was retracted. AI raises delivery speed while enlarging the reputational risk attached to a signature. Clients are responding by raising the bar at acceptance: either traceable data sources, or a lower unit price for AI-assisted output. The layer squeezed first is the one that only assembles materials and drafts documents.

4. Where Consultants Go

Pooling this search firm's 2026 placement records, consultant exits fall into four categories. Each has a different buyer, pricing logic and barrier to entry.

Corporate strategy and investment teams. The most traditional exit, and the most crowded in 2026. Industrial companies in new energy, semiconductors, biotech and cross-border e-commerce are building in-house strategy and post-investment teams, and they want advisers who can build models and sit with the business rather than advisers who excel at presenting. Supply far exceeds demand, and employers have added a filter: has this person actually pushed a project to implementation?

Outbound Chinese companies. Expansion abroad requires people who understand international markets and can structure cross-border vehicles and local compliance. Consulting backgrounds are treated as general capability here. Demand clusters in the Middle East, Southeast Asia and Latin America, where the real constraint is language and local relationships rather than analytical frameworks.

AI and data vendors. This category prices most aggressively. Industry solutions, commercialization strategy and delivery consulting roles need people who understand client businesses and can follow technical language, and consultants are a natural pool. Over the past two years, several large model and data services companies have made former consultants their preferred profile for industry sales and solutions roles.

The Big Four and adjacent professional services. Lateral movement across audit, tax, compliance and transaction advisory has accelerated. The direct consequence is that pay anchors have flattened. For candidates at the same level of experience, the spread between competing offers is narrowing, and the window of a 30% raise from a single move has largely closed.

The other half of the flow is the entry point. Junior hiring is contracting most visibly. The old path of elite university, two years as an analyst, then decide where to go is being shaved at the base by AI tools. That outcome follows from the cost structure of delivery regardless of which firm is doing the hiring.

5. Compensation Benchmarks

The table below sets out benchmark bands by seniority, estimated from this firm's 2026 search mandates in professional services and consulting together with commonly observed pay structures in China. Figures are total cash in the China market, covering foreign and leading domestic firms, and exclude long-term incentives.

6. HR Recommendations: How to Rebuild a Consulting Talent Model

1. Replace the volume campus model with a narrow, gated pipeline. If your firm is still hiring junior consultants against a 2024 headcount plan, rerun the arithmetic. Material assembly, first drafts and data cleaning, the tasks AI tools now cover, typically account for more than half of a junior consultant's hours. Cut entry-level intake by 30% to 50% and convert it into a 12-week gated program: weeks 1-4 on business data sources, model use and prompt engineering; weeks 5-8 on industry diligence and client interview transcripts; weeks 9-12 on live project support measured against delivery quality. Offer contracts at the end of the gate rather than at the start.

2. Move performance metrics from utilization to delivery margin and asset reuse. As revenue shifts toward managed and platform work, managing a team on billable hours means doubling down on a business line that is exiting. Run two metrics together: project gross margin as the base, plus a quantified asset list per project covering reusable methodology, templates and data assets, reviewed quarterly. Raise the revenue-per-person target and lower the billable-hours target at the same time, and the team will treat tools and assets as real work.

3. Use asset participation, not only salary, to retain engagement managers. This layer is the scarcest and most mobile group in the industry. On cash alone, domestic firms cannot match corporate employers and foreign firms cannot match AI vendors. A workable structure gives this layer participation in platform assets: carve out a share of revenue from reusable products, tools and industry databases as a team pool, and lift variable pay to 30% to 40% of total cash. The retention argument shifts from what you are paid this year to what the asset in your hands is worth next year.

4. Rewrite the hiring profile and replace storytelling interviews with work samples. Remove vague requirements about prestigious universities and communication skills from the job description and replace them with three testable criteria: can run a data tool chain independently; can define a problem without an existing framework; can audit AI output into a deliverable a client will sign. Add a 48-hour live case covering real industry data with a written judgment and executable recommendation. When hiring laterally from corporate strategy teams, industrial companies and AI vendors, budget a 15% to 25% pay premium. That premium buys implementation experience, not a title.

Data Sources and Limitations

Sources for this report: global firm revenue and headcount figures come from official firm announcements (Deloitte global revenue release of 30 September 2025; PwC global revenue release of 28 October 2025; EY global revenue release of 15 October 2025; KPMG global revenue release of December 2025). Accenture revenue, bookings, margin and headcount figures come from its third-quarter fiscal 2026 earnings release dated 18 June 2026; the single-day share price move and the 2025 headcount reduction and severance figures are as reported by financial media. China macro data comes from the National Bureau of Statistics release of 15 July 2026. Client movements at professional services firms in China come from a 2024 tally by Yicai, with the year noted in the text. Delivery quality incidents come from public reporting in August 2026.

The compensation bands in Section 5 are estimates drawn from search mandates and publicly stated pay structures, without third-party survey data, and should be adjusted for firm type and location. Judgments about talent mobility reflect practical recruiting observation in the professional services sector and are not forecasts of any named firm's workforce policy.