Margin expansion is the question that sits underneath almost every strategic conversation in the industry right now, and it's also the one most agencies talk about in the vaguest possible terms because the honest answers are uncomfortable. The phrase itself usually gets used to mean "make more money without raising prices", which is roughly true but misses most of the structural reality. Real margin expansion in a modern agency requires either charging more for the same work, producing the same work with fewer resources, or changing what the agency sells so that the cost structure looks different. Most agencies are trying to do some version of all three simultaneously, and most are failing at all three because the changes required are deeper than the dashboard optimisations they're actually making.
The starting point is understanding why agency margins have been compressing in the first place, because margin expansion only makes sense as an answer to specific pressures. For most of the industry's history, agency margins were structurally protected by a few things: rate cards that clients didn't have the data to challenge, media commissions that scaled with budgets, production markups that were rarely audited, junior labor that could be billed at senior rates, and the general opacity of how agency work was priced. Each of these has eroded. Procurement teams now benchmark rates aggressively. Media commission is largely gone and the rebate scandals killed what remained. Production markups are visible and contested. Junior labor is either being automated or becoming transparent to clients who know what they should cost. And the whole pricing model is much less opaque than it used to be. The net effect is that the margin cushions the industry used to operate with have thinned, and the agencies that haven't actively rebuilt their cost structure or their revenue model are watching their margins compress by a point or two per year in ways that are hard to stop without structural change.
The first and most obvious lever is using AI to reduce the labour cost of producing the same work. This is the move everyone is talking about, and it's real, but it's also the most competitive of all the margin moves because everyone has access to the same tools. If an agency can produce in two hours what used to take a team two weeks, the cost savings are enormous on paper, but they only show up in margin if the agency continues to charge the old price for the new speed. The moment the client realises (or procurement demands to see) how the work is actually being produced, the pricing conversation reopens and the savings get transferred from the agency to the client. This has already happened in parts of the industry, performance creative production, for example, where clients have largely figured out that AI tools exist and have adjusted their expectations for what things should cost. In other parts, it's still happening, and the agencies that have moved fastest are enjoying a window of temporarily expanded margins that will close as the market catches up.
The honest framing is that AI-enabled production is a margin expansion lever in the short term but a table-stakes capability in the medium term. The agencies that will hold expanded margins five years from now are not the ones that used AI to do the same work cheaper, they're the ones that used the freed-up capacity to take on work they couldn't profitably do before (more variants, more personalisation, more channels, more markets), or that invested the savings into higher-value capabilities that AI doesn't commoditise. The agencies that pocketed the AI savings as margin without reinvesting are likely to give those savings back when clients catch on.
The second lever, closely related but distinct, is changing the labour mix in favour of senior-heavy teams with AI leverage. The traditional agency pyramid had a small number of senior people at the top directing a large base of junior and mid-level labor. This pyramid was the entire economic engine, the margin came from billing juniors at rates that exceeded their cost while the seniors captured attention and relationships. What's happening now is that AI tools make senior people radically more productive at producing the work itself, while the junior layer becomes less necessary because the routine work those juniors used to do is exactly what AI handles well. The margin-expanded agency of the future has an inverted or flattened pyramid: a higher ratio of senior to junior staff, with AI tools substituting for much of what the junior layer used to do, and senior people producing much more output per hour than they used to. This works economically because senior time, even at higher rates, can be enormously more productive when augmented with good tooling, and clients are generally willing to pay for senior attention in ways they aren't willing to pay for junior execution.
The catch is that this inversion breaks the traditional agency training pipeline. The juniors who used to learn the craft by doing the routine work are no longer doing it, which means they're not developing the skills they'd need to become the next generation of seniors. Agencies that lean into the inverted pyramid without solving the training problem are eating their own seed corn, the senior people they depend on now will eventually leave or retire, and there won't be a next generation behind them. The agencies that are handling this thoughtfully are explicitly investing in new kinds of training for junior talent: not teaching them to do the commoditised work, but teaching them to direct AI tools well, to evaluate outputs critically, to have taste and judgment earlier in their careers than previous generations needed. This is slower and more expensive than the old apprenticeship model, but it's the only way the inverted pyramid stays sustainable. Agencies that haven't figured out the training problem are getting a short-term margin boost from the inversion and building a longer-term talent cliff that will catch up with them in five to ten years.
The third lever is shifting pricing away from hours and toward value. This was covered in the outcome pricing discussion, but it's worth connecting directly to the margin question: when you're paid on outcomes or on value delivered, the relationship between hours worked and revenue earned decouples, which is the entire point. An agency that produces a piece of work in two days that's worth $200,000 to the client can legitimately charge something much closer to the $200,000 than to the two-days-of-labor cost. This is how software companies make money, and it's the economic logic that underlies almost every successful move out of the labor arbitrage trap. The margin expansion from this shift can be dramatic, agencies that have successfully moved their best work to value pricing sometimes see gross margins two or three times what they were getting from the hourly model on the same work.
The difficulty is that value pricing requires the client to believe the value is real and unique to the agency providing it. When the work is commoditised or easily replicated, clients won't pay value-based prices because they know they can get the same thing cheaper elsewhere. Value pricing therefore depends on the agency having something distinctive enough to defend the price, which circles back to the moat discussion. Agencies with genuine moats can charge value-based prices because the client has no good alternative. Agencies without moats find that attempts to move to value pricing are rejected by procurement, who correctly observe that the work isn't differentiated enough to justify the premium. The margin lever only works to the extent the positioning has been built first, which is why skipping the positioning work and going straight to "we should charge more" almost always fails.
The fourth lever is productisation, covered earlier, which creates a different relationship between cost and revenue entirely. When an agency productises a service, the marginal cost of serving the next client drops significantly, which means every additional sale of the product contributes much more to margin than an additional custom engagement would. Productised offerings tend to have significantly higher gross margins than custom work, sometimes 60-80% gross margins versus 20-40% for custom services, and the compounding effect is real. The challenge, as discussed, is that productisation is hard to do well, requires real investment, and only works for specific kinds of work. But for the agencies that have successfully productised parts of their business, the margin impact on those product lines is substantial, and the blended margin of an agency with both custom and product revenue streams is meaningfully higher than a pure custom shop.
The fifth lever is eliminating scope creep and non-billable work. This is unglamorous but significant. In most agencies, a meaningful fraction of the work being done, often 15-30% by hour, is either unpaid scope creep, rework caused by poor briefing, internal coordination that shouldn't be necessary, or process overhead that got added without anyone noticing. Each of these is a direct margin leak. The agencies that have gotten serious about margin usually have some version of tight scope management (with formal change requests for anything outside the agreed scope), disciplined briefing practices (that reduce rework by getting the brief right the first time), clear decision rights (that eliminate approval bottlenecks), and regular audits of what people are actually spending time on (to identify and eliminate work that doesn't produce value). None of this is exciting, but the margin impact is often several points, which, in a business operating on 15-25% net margins, is a significant lift. The agencies that don't do this are leaving margin on the table while looking for more dramatic interventions.
The sixth lever is concentrating the client portfolio on accounts where the work is more profitable. Every agency has a distribution of account profitability, and the distribution is usually wider than leadership realises. A few accounts are very profitable, most accounts are marginally profitable, and some accounts are actively losing money, either because the scope has grown without the fee growing, because the client is difficult and requires disproportionate senior attention, because the work has been under-priced from the start, or because operational inefficiencies specific to that account are eating margin. The agencies that expand margin through portfolio concentration do the unglamorous work of actually measuring profitability by account (most agencies don't do this rigorously), identifying the unprofitable ones, and either renegotiating the terms, redesigning the engagement, or walking away. Walking away is the hardest part, there's always a reason to keep an unprofitable account (it's strategic, it has growth potential, the relationship is valuable, the case studies matter), but the agencies that have built discipline here usually see meaningful margin improvement within a year, because cutting the loss-making work has a bigger effect than incremental improvements on the profitable work. The agencies that won't do this usually keep pouring senior time into accounts that will never pay back, which drains capacity that could be serving more profitable clients better.
The seventh lever is reducing the cost of new business development. This is rarely discussed as a margin lever but is often where the biggest hidden costs sit. Most agencies spend a significant fraction of their total capacity on pitches, chemistry meetings, speculative work, and the general overhead of chasing new business, and most of it doesn't convert. An agency that wins one in five pitches is doing well; most win one in seven or one in ten. That means 80-90% of the effort invested in new business is effectively wasted from a direct revenue standpoint. Agencies that have reduced this cost have usually done so by being much more selective about which pitches to enter (refusing to pitch work they're not well-positioned to win, or where the commercial terms don't support profitable work), by building referral and relationship-based growth channels that don't require pitching, and by investing in thought leadership or content that brings inbound opportunities with shorter sales cycles and higher conversion rates. The agencies that pitch everything and treat the hit rate as a cost of doing business are quietly subsidising their new business operation with margin from their delivery work. The agencies that have gotten disciplined here see meaningful margin improvement from simply doing less unpaid work.
The eighth lever, under-appreciated, is using technology to reduce operational overhead that doesn't produce client value. Every agency has significant costs in areas that don't touch client work directly: project management, resource allocation, time tracking, finance and billing, HR, recruiting, internal communications, IT. These functions are necessary but don't generate revenue, and in many agencies they've grown organically over years into bureaucratic weight that consumes senior attention and adds cost without proportional benefit. Modern agencies can automate or streamline much of this, project management software that actually works, resource planning tools that reduce the constant firefighting, finance systems that generate reports without manual work, AI assistants that handle routine internal requests. The savings aren't individually huge but they compound, and the strategic benefit of freeing senior people from internal process overhead is often larger than the direct cost savings. The agencies that have invested in operational infrastructure often find that their margin improves not because they cut anything, but because their senior people are doing more billable-equivalent work instead of shuffling internal paperwork.
The ninth lever is geographic and talent arbitrage that's more sophisticated than the old offshore model. The traditional version of this, opening an office in a lower-cost country and staffing routine work there, has been around forever and produces some margin benefit but comes with real coordination and quality costs. The modern version is more nuanced: distributed teams where senior talent can live anywhere, specific functions located where the talent is best rather than where the cost is lowest, access to global freelance networks for specialised work rather than full-time hiring, and explicit decisions about which work stays close to clients and which can operate remotely. The agencies that have gotten this right are usually the ones that rebuilt their operating model around distributed work during the pandemic and kept the changes rather than reverting. The ones that snapped back to the old "everyone in the London or New York office" model are paying premium labor costs without the productivity benefits that used to justify them.
A few honest observations about margin expansion that most agency conversations skip.
The easy levers have mostly been pulled. Every agency has already looked at the obvious places, tightening scope management, using AI for basic production, trimming some staff, pushing for higher rates. The margin gains from these moves are real but small, and they're one-time rather than sustainable. The agencies expecting to expand margins by two or three points per year indefinitely through operational optimisation are going to be disappointed. The durable margin expansion comes from structural changes to what the business is and how it makes money, not from running the existing business more efficiently.
Margin and growth are often in tension. Agencies under pressure to grow revenue often accept work at margins that dilute their average profitability, telling themselves the growth is worth the margin hit. In some cases it is, if the growth builds capability or establishes a beachhead in a strategic category, but in many cases the growth is just growth, and it masks the underlying margin compression while postponing the reckoning. The agencies that have actually expanded margins sustainably are often the ones that were willing to grow more slowly in exchange for better quality revenue. This is politically hard because growth numbers look better in a board presentation than margin numbers, and because holding company reporting often weights growth heavily, but the agencies that resist the growth-at-any-cost pressure end up with healthier businesses.
The biggest margin lever is usually positioning, not operations. Agencies that have expanded margins durably usually did so because they moved into a position where they could charge more for the same work, not because they figured out how to do the same work cheaper. The positioning move, becoming the agency that specific clients have to hire for specific kinds of work, with no good alternatives, is what sustains value pricing, which is what drives structural margin expansion. Agencies that try to expand margins through efficiency alone while staying in the same commoditised positioning are running up a down escalator.
Leadership compensation is usually aligned the wrong way. In many agencies, especially holding-company-owned ones, senior leadership is compensated primarily on revenue growth, with margin as a secondary consideration. This creates a predictable pattern: leadership chases top-line growth, accepts lower-margin work to hit revenue targets, and leaves it to the next leadership team to deal with the margin compression. The agencies that have actually expanded margins sustainably usually have leadership whose compensation is weighted toward margin and account quality rather than pure revenue, and those agencies are less common than the industry pretends, because changing compensation structures is harder than complaining about them.
Technology investments look expensive until they don't. Building the operational infrastructure, AI tooling, and data systems that enable margin expansion costs real money upfront, and the payback period is often longer than a single budget cycle. Agencies that can't absorb short-term cost increases in exchange for medium-term margin benefits usually can't make the investments that would matter, which is why private-equity-backed and founder-led agencies often outperform holding-company agencies on this dimension, they have the patience that the quarterly reporting cycle doesn't allow. This is one of the structural reasons that the margin-leading agencies in the industry are often not the largest ones.
The agencies that have expanded margins the most are usually the ones that got smaller first. This is the most counterintuitive observation and the one leadership teams most want to avoid. Meaningful margin expansion often requires letting go of business that was dilutive, reducing headcount that was over-allocated to work that couldn't support it, and shrinking the footprint before rebuilding on better foundations. The agencies that have done this successfully usually look worse by revenue metrics for a year or two while the margin profile improves, and then grow again from the higher-margin base. The agencies that refuse to shrink first, trying to expand margins while maintaining or growing revenue simultaneously, usually achieve neither, because the existing revenue base includes too much unprofitable work to hit meaningful margin targets without cutting something.
The deeper point about margin expansion is that it's not primarily an operational question, even though agencies mostly treat it as one. It's a question about what kind of business the agency is choosing to be. An agency selling commoditised services on hourly pricing into a buyer's market has limited margin levers available, no matter how efficiently it runs. An agency selling distinctive services on value pricing into a specific positioning has structural margin advantages that compound over time. The move from the first kind of business to the second is what "margin expansion" ultimately requires, and it's a multi-year strategic transformation, not a cost-cutting exercise. The agencies that have expanded margins durably in the last five years are the ones that made this transformation explicitly, invested in the positioning, rebuilt the cost structure, and accepted the short-term pain of the transition. The agencies that haven't are finding that incremental efficiency improvements aren't enough to outrun the structural pressure, and the margins keep drifting down despite everyone working harder.
The uncomfortable implication for most agency leaders is that the margin question isn't really about margins, it's about whether the current shape of the business can support the margins leadership wants. If the answer is no, the real work isn't optimising the existing business; it's becoming a different business. The agencies that understand this are making the hard changes now. The agencies that don't are running the same plays year after year and wondering why the numbers keep getting worse. The difference between those two trajectories will be one of the most visible dividing lines in the industry over the next decade, and the gap is widening faster than most of the conversation acknowledges.
