A partner compensation system cannot reward what it cannot see.
Where AI costs sit outside the resource model, firms risk continuing to reward the traditional measures: managed fees, supervised revenue and originations. Partners then have little reason to move first on adoption or experimentation.
Our 2026 Equity Partner Contribution & Compensation Survey of more than 140 professional services firm leaders found that only 21% of firms believe their compensation system provides the financial safety needed to encourage partners to experiment with AI and innovate service delivery. A further 36% were neutral.
This matters beyond the partnership. At SuperReturn in Berlin last month, Apollo’s Scott Kleinman told a room of lawyers, accountants and consultants: “you’re going to see a lot of pressure.”
Private capital investors are starting to price AI disruption into professional services the way they already have with software. Partner compensation is likely to be one of the first places where that pressure will be felt.
Here are three changes that can help your firm respond.
1. Redefine gross profit metrics to include AI usage
Most partner compensation systems focus on top-line revenue: managed fees, supervised revenue and originations. They are all revenue-driven. In many firms, particularly mid-sized firms, gross profit receives much less attention in assessing a partner’s contribution to the projects they lead.
There are two big reasons to redefine how you measure gross profit in the age of AI.
Partners share in profit, not revenue. Most partner performance evaluations give revenue considerable weight. This is a problem that predates AI. After all, “revenue is for vanity, profit is for sanity.”
The best read on the profitability of a service is still the gross profit it generates—at group and team level and, depending on the service-delivery model, at partner level too, in relation to the projects each partner leads.
AI is changing the resourcing mix. It changes how many professionals are needed at each level of experience, relative to the work AI agents and similar technology can now perform.
What needs to change
Gross-profit calculations need to include the cost of that AI. You cannot optimise for efficiency without knowing the full cost of delivering the service. Nor can cross-team comparisons, performance discussions, evaluations and reward allocations be reliable if a major cost element is missing.
On paper, the change is simple: base production measures—and what you reward—on gross profit rather than revenue.
Start with net revenue: gross revenue plus premiums, less discounts, write-downs and charge-offs. Then deduct direct people costs, using the cost rates for each level of seniority. Then factor in AI usage.
You need to overlay this cost at the gross profit level because AI licence fees usually sit as overhead costs. Either they’re invisible on the P&Ls of individual teams, groups or partners, or they’re lumped into an overhead allocation, which means they can’t be controlled.
We don’t usually recommend reallocating overheads to groups, because pricing choices should be based on value provided to the client, not an assumed cost of production. Some firms do incorporate an overhead reallocation component, mainly to force the right pricing choices.
Yet the world has changed. Technology is rapidly overtaking real estate as the second-largest P&L item in the legal sector after people costs. AI is now such a significant factor in production that it earns an exception. Allocate AI usage costs back to teams based on usage. The teams that use more AI should bear more of the cost.
Action: Ask whether your current production and reward measures show the full cost—and therefore the true profitability—of AI-enabled delivery. If not, agree how AI usage will be measured and attributed before the next remuneration round.
2. Create incentives for AI adoption and innovation
In many firms, AI innovation is not an explicit part of partner contribution standards, objective-setting or performance evaluations.
Reward the partners who move first. AI-related innovation needs to be a visible and prominent part of the partner compensation framework.
A recent Financial Times report on McKinsey illustrates the direction of travel: the firm is reported to have spent two years overhauling how it pays partners as it moves away from the billable-hour model on which it was built and towards performance-based fees.
That kind of shift does not happen without partners willing to back it before it is proven: taking on client relationships priced differently from how the firm has always priced them, with no guarantee that the model will hold. It is exactly the kind of risk most compensation systems still punish rather than reward.
What needs to change
Fund the risk, not just the reward.
Most firms reward results. Partner A has a good result because they generated good revenue, so Partner A receives more money. But firms also need to allow people to take risks when those risks do not immediately pay off.
At a minimum, compensation should not be negatively affected. Ideally, there should be a financial incentive specifically for taking the risk. That can take different forms: a special bonus allocation or recognition built into how partners move across tiers, bands and ladders.
This is not an argument for rewarding AI activity for its own sake. It is about recognising commercially disciplined attempts to change how the firm delivers work.
Action: Add AI-related innovation to partner contribution standards and agree, in advance, how the firm will protect well-managed experiments from an automatic short-term remuneration penalty.
3. Embed a long-term view of performance and profitability
Annual resets punish long-term bets.
Many partner compensation systems are short-term oriented: eat-what-you-kill and financial-merit firms, and even many holistic firms, unless they are built on some form of lockstep or modified lockstep. Equity or profit share resets every year based on that financial year’s results.
That does not work when AI investment is involved. Most AI investments will not pay off within the financial year. You generally need at least a two- to three-year horizon to judge contribution and performance fairly.
What needs to change
Build patience into the architecture.
Ideally, profit share is sticky enough to reflect that multi-year horizon. If it resets to zero every year, the long-term view does not hold. This is not a fix made to the numbers each year; it must be built into the design of the partner compensation system.
For all of this to work, the compensation committee must be aligned with the approach. It needs a shared view of what merits recognition, how AI costs will be measured, and when a long-term investment should begin to be judged.
Action: Test whether an approved AI initiative that pays off in year two or three would be fairly recognised in your current profit-sharing model. If the answer is no, change the architecture now—not through a discretionary adjustment later.
Is your compensation system AI-ready?
None of this requires waiting for certainty about how AI will reshape your sector. It requires deciding now whether your compensation system is helping or hindering your strategy.
Take these three questions to the next compensation committee meeting:
Do we measure the true gross profit of AI-enabled work, including AI usage costs?
Can partners take well-managed AI risks without being penalised in the remuneration round?
Will partners share fairly in value that takes more than one financial year to emerge?
If the answer to any of these is unclear, the system is not yet giving partners a clear reason to change how work is resourced and delivered.
The Partner Remuneration System Diagnostic™ provides a structured heat map of where the system is helping or hindering AI-enabled delivery, with practical recommendations for closing the gap.
Get in touch to make your plan.
The point is not to reward AI for its own sake. It is to make sure your partner compensation system is not cancelling out the investment your firm has already decided to make.
