The traditional marketing agency is a staffing model with a markup. You are paying for people, and those people are doing work that is increasingly automated. That is not an indictment of agencies. It is an arithmetic problem that changes what different kinds of clients need to pay for.
How agency cost structures actually work
Agencies operate on a staffing pyramid. Junior talent does the execution. Senior talent reviews and clients. The math: a $75,000-per-year mid-level employee costs the client approximately $150 per hour after salary, benefits, real estate, software, management overhead, and margin. Large agencies run a 4 to 5x multiplier on salary. Healthy agencies target a 55:25:20 split across people costs, overhead, and profit on adjusted gross income. Industry research suggests only one-third of agencies consistently hit all three benchmarks.
Average net margin industry-wide: 13 to 15 percent. Boutique studios under 10 people run closer to 19 percent (less overhead per head). Agencies above 50 people drop to 8 percent as management layers and real estate costs accumulate.
Revenue per employee: the industry average is around $163,000. The large holding companies run $164,000 to $171,000 per employee. The reason that number has only improved modestly despite significant AI investment is instructive: AI reduced the cost of execution, but agencies typically replaced headcount with AI tools rather than reducing headcount proportionally. The productivity gains went to margin protection, not to the client.
This cost structure was rational when execution required people. Writing a meta description, building an audience list, setting up conversion tracking, running A/B tests on ad copy -- each of these took human hours. They still take human time for setup and oversight, but materially less than three years ago.
What AI changed in the execution layer
The shift is directional and measurable. The major holding companies (WPP, Publicis, Omnicom, IPG, Dentsu, Havas) held 44.6 percent of global advertising revenue in 2019. By Q1 2024, that share had dropped to 29.6 percent. WPP reported profits down 71 percent in the first half of 2025. The Omnicom-IPG merger announced in late 2025 was partly a response to a structural efficiency problem: two organizations with overlapping execution capacity combining to cut shared overhead.
On the client side: 60 percent of senior US marketing leaders reported spending less on agencies in 2025 than the prior year, and they named AI adoption as the primary reason. This is not a budget compression story. It is a structural shift in what clients need from external partners.
What changed specifically:
- Ad copy and creative variations: Generating 50 variations of an ad headline now takes minutes with AI tools. It previously took a copywriter hours. The creative brief and the strategic judgment behind it still require experience. The production does not.
- Audience research and segmentation: AI-assisted audience research, persona development, and segmentation work that previously took analysts a week can be done in hours with the right tooling and data.
- Performance reporting: Automated dashboards and AI-generated reporting narratives have reduced the reporting preparation time that junior analysts previously spent on non-analytical tasks.
- SEO and content production: Drafting, optimizing, and publishing content at scale has been substantially accelerated by AI, reducing the per-word production cost while increasing the importance of strategic editorial judgment.
The work that AI has not changed: strategic positioning decisions, brand voice development, complex buying-cycle mapping, organizational alignment between marketing and sales, and the human judgment required to translate data into a course correction. These capabilities do not compress under AI automation. They may even become more valuable as execution commoditizes.
The consulting arms of major management firms recognized this shift. The largest of them now operates at roughly $20 billion in annual marketing and technology revenue, growing at approximately 8 percent year-on-year -- on an opposite trajectory to the traditional holding companies whose CEOs now explicitly name management consultancies as their primary competitive threat, not other agencies.
The AI consultancy model
The AI consultancy model inverts the staffing pyramid. It is primarily senior practitioners doing strategic and architectural work, using AI tooling to compress the execution layer rather than staffing it with junior talent.
The economics differ from agencies in three ways:
- Gross margins are higher: Boutique consulting firms typically run 50 to 70 percent gross margins compared to the 13 to 15 percent agency industry average. This is not primarily because consultants charge more (though day rates tend to be higher). It is because the overhead structure differs: fewer people per engagement, less real estate, minimal junior staffing pyramid.
- Engagements are scoped differently: Agency retainers are typically sold as monthly capacity (x hours per week of y roles). Consulting engagements are sold as outcomes and deliverables. The client is paying for a problem solved rather than hours allocated.
- The leverage structure differs: An agency runs its margin by leveraging senior partner time across multiple junior-staffed accounts. A consultancy runs its margin by leveraging AI tools to compress execution across fewer but deeper engagements.
Day rate benchmarks for independent AI marketing advisors: $600 to $1,200 per day for freelance consultants. $1,500 to $2,500 per day for advisors placed through consulting firms. Monthly advisory retainers range from $2,000 to $5,000 for five to ten hours of engagement, up to $15,000 to $50,000 for embedded strategic advisory covering 25 or more hours per month.
These rates are higher on a per-hour basis than typical agency blended rates. The relevant comparison is not hourly rate -- it is cost per outcome. An agency that charges $8,000 per month for campaign management and requires 18 months to build attribution infrastructure is not cheaper than a consultancy that charges $15,000 for a 90-day engagement that delivers the same infrastructure plus a trained in-house team.
Comparison: what each model delivers
| Dimension | Traditional agency | In-house team | AI consultancy |
|---|---|---|---|
| Cost structure | Monthly retainer; percentage of spend for media; junior execution pyramid | Salary + benefits; fixed overhead regardless of output; scales with headcount | Scoped engagement fee or advisory retainer; higher day rate, lower total hours |
| Incentive alignment | Fee may increase with managed spend; long retainers favor account stability over performance | Internal team aligned with business outcomes; can be siloed from commercial pressure | Outcome-scoped; incentivized to solve the problem, not extend the engagement |
| Execution capacity | High; dedicated team handles production volume | Variable; depends on team size and resourcing | Limited by design; execution is either AI-assisted or transferred to in-house |
| Strategic depth | Variable; often senior talent at proposal stage, junior at execution | Deep domain knowledge; weak on emerging channels or AI tooling | High; senior practitioners at every engagement stage |
| AI capability | Uneven; adoption varies across agencies and practice areas | Depends on training investment and tool budget | Core capability; AI tooling is how execution compresses |
| Knowledge transfer | Low; learnings stay in agency systems at contract end | High; builds institutional capability | High when designed for it; a good consultancy explicitly transfers capability |
| Typical engagement length | 6 to 24 months; ongoing execution relationship | Permanent; organizational function | 3 to 12 months for defined transformation; ongoing advisory where needed |
| Client satisfaction benchmark | Average agency score: 7.89/10 in a 2025 survey of 138 marketing leaders; top complaint was lack of ROI clarity | High satisfaction with team; low satisfaction with capability gaps | Limited benchmarks available; outcome-scoped engagements reduce ambiguity |
Which model fits which situation
The choice is not about preferring one model. It is about matching the model to the specific problem.
Many organizations need both: ongoing execution managed by an agency or in-house team, with a consultancy engaged for strategic and measurement architecture. The two models are not mutually exclusive, and the most effective marketing operations we observe combine dedicated execution capacity with external strategic oversight that can challenge the assumptions embedded in any single execution relationship.
leapbuzz operates as an AI consultancy for marketing strategy, AI adoption roadmaps, and measurement architecture -- not as an execution retainer. We are built for transformation engagements, not for ongoing creative production. We'll say so if your actual problem is execution volume.
Five-market hiring context
The agency-vs-consultancy decision is also shaped by the talent market in each of the five markets. Senior AI marketing practitioners are not uniformly available.
| Market | Talent market | Agency landscape | Regulatory note |
|---|---|---|---|
| Singapore | Small talent pool for senior AI marketing roles; significant competition from financial services and tech companies for the same profiles. Strong APAC hub for regional roles. | Full spectrum from global holding company offices to boutique regional shops. IMDA and MAS guidance affects financial-services marketing partners specifically. | MAS Guidelines for digital advertising in financial services (effective March 2026) add compliance obligations that affect agency scope in banking and fintech verticals. |
| USA | Deep talent pool; largest market for AI marketing practitioners. High competition and cost. Remote-first culture expands access to national talent. | Broadest agency landscape globally; consolidation among large holding companies; growth in boutique AI-native shops. FTC scrutiny of performance claims increasing. | State privacy law patchwork (CCPA, CDPA, and equivalents) requires compliance consideration in any U.S. marketing engagement. AI-generated content disclosure discussions ongoing at FTC. |
| Canada | Smaller talent pool than US; strong in Toronto and Vancouver. Lower salaries than US equivalents; some US talent accessible through remote arrangements. | Mix of global agency offices and Canadian-owned shops. Quebec market often requires French-language capability and distinct consent compliance under Law 25. | CASL governs commercial electronic messages with opt-in requirement (stricter than US CAN-SPAM). Quebec Law 25 adds automated profiling disclosure requirements. |
| Australia | Strong digital marketing talent concentration in Sydney and Melbourne. Growing AI marketing expertise; competition from financial services and retail sectors. | Full-service agency presence (global and local); growing performance-specialist boutiques. ACCC active in enforcing misleading advertising claims. | Privacy Act reforms (December 2024): automated decision-making disclosure now required where decisions have significant effects on individuals. Spam Act governs commercial email. |
| Malaysia | Growing talent pool; lower cost than Singapore; Malay and English bilingual capability is a distinctive advantage for brands targeting both segments. Kuala Lumpur hub. | Mix of regional agency offices and local Malaysian shops. Cost structures materially lower than Singapore or Australia. Shopee Ads and Lazada Ads are relevant performance channels not present in US or Canada. | PDPA Malaysia amendment in force June 2025; transfer impact assessments now required for data sent to third countries. MCMC Communications and Multimedia Act governs digital advertising. |
Frequently asked questions
What is the real cost difference between a marketing agency and an AI consultancy?
The per-hour cost of a consultancy is typically higher: independent AI marketing advisors charge $600 to $1,200 per day; advisors through consulting firms charge $1,500 to $2,500 per day. Monthly agency retainers for comparable strategic scope are often lower on a blended basis but require more total hours because execution is included. The relevant comparison is cost per outcome, not cost per hour. A consultancy engagement that transfers capability and reduces agency dependency over 90 days may cost less in total over a 12-month view than an ongoing retainer that delivers execution without knowledge transfer.
Are traditional marketing agencies in decline?
The holding company segment is under structural pressure: the large holding companies held 44.6 percent of global advertising revenue in 2019 and 29.6 percent by Q1 2024. WPP profits fell 71 percent in the first half of 2025. The Omnicom-IPG merger in late 2025 was partly a response to structural efficiency challenges. Boutique specialist agencies and mid-market independent shops are in a different position -- many are adapting faster than large holding companies. The decline is concentrated at the holding company layer where AI automation hits the staffing pyramid economics hardest.
What does an AI consultancy do that a traditional agency doesn't?
Three things specifically: (1) AI-native execution -- using AI tooling to compress what would be junior-team hours into senior-team hours with AI leverage; (2) Organizational capability transfer -- explicitly building the client's internal capability rather than creating dependency; (3) Strategy-first scoping -- engagement around defined outcomes rather than monthly hour allocations. Traditional agencies are structured around execution volume; consultancies are structured around problem resolution. This makes consultancies better suited for transformation work and agencies better suited for ongoing execution management.
How satisfied are marketing leaders with their agencies?
A 2025 survey of 138 marketing leaders found the average agency satisfaction score was 7.89 out of 10. Only 13 of 138 respondents gave a perfect 10. The top complaint was lack of ROI clarity, followed by insufficient transparency and reactive rather than proactive work. The satisfaction gap is particularly pronounced on measurement and attribution: agencies tend to report the metrics their campaigns perform well on, which may not align with the business metrics the client cares about.
Can you use both an agency and an AI consultancy simultaneously?
Yes, and many mature marketing organizations do. A common model: an agency or in-house team handles ongoing campaign execution and creative production; an AI consultancy handles measurement architecture, AI tooling strategy, and quarterly strategic reviews. The consultancy function can also provide a useful external check on the agency's work -- reviewing performance claims and incrementality, identifying gaps in channel coverage, and flagging when execution is serving the agency's reporting preferences rather than the client's actual goals.
Why do large consulting firms compete with marketing agencies now?
The strategic layer of marketing -- AI adoption roadmaps, data architecture, personalization infrastructure, measurement frameworks -- is adjacent to the technology and business transformation consulting that management firms already sell. As AI made the execution layer of marketing cheaper and more automated, the strategic layer became proportionally more valuable, which is exactly what management consultancies are structured to deliver. The largest consulting firm's marketing practice now operates at approximately $20 billion in annual revenue, growing at 8 percent year-on-year, while holding companies contract. The market is telling a clear story about where the value in marketing services is migrating.
What should I look for when evaluating an AI consultancy vs a traditional agency?
Five evaluation criteria that differentiate them in practice: (1) Outcome scoping -- does the proposal define specific business outcomes or hours of work? Consultancies scope to outcomes; agencies scope to capacity. (2) Knowledge transfer -- does the engagement plan explicitly build your team's capability or create ongoing dependency? (3) Senior practitioner access -- are the senior practitioners who pitched the work the ones doing the work? (4) Measurement independence -- does the partner measure their own impact using methodology they control, or using independently verified incrementality tests? (5) AI tooling transparency -- can they explain exactly what AI tools are in use and what the human oversight layer looks like?
Is in-house marketing a better alternative to both agencies and consultancies?
In-house teams grew substantially: the State of PPC 2025 survey found in-house management grew from 44 to 71 percent of respondents over three years, driven partly by AI reducing execution complexity and partly by a desire for greater brand control and measurement transparency. In-house is superior on institutional knowledge and brand alignment. The gaps it typically has: emerging channel expertise, incrementality testing methodology, AI tooling strategy, and the external perspective to challenge embedded assumptions. Hybrid models (strong in-house team plus targeted external advisory) outperform pure in-house or pure agency in most organizations above a certain marketing maturity level.