This is not about one specific agency. It is a composite drawn from several we have worked with over the past 18 months. The details are real. The pattern is consistent. And the numbers are uncomfortable if you are running a 15-person agency wondering why a five-person competitor keeps winning your pitches.
The team
Five people. Three are client-facing generalists (strategy, creative direction, account management). One is an operations and systems lead. One is the founder, who splits time between business development and senior client work.
No juniors. No project managers. No dedicated copywriters, designers, or analysts. Not because those skills are not needed, but because AI handles the execution layer that those roles traditionally covered.
Everyone on the team is mid-to-senior level. They have the experience to know what good looks like, which is essential because their job is not to produce work from scratch. Their job is to direct, refine, and quality-control the work that AI produces.
The output
This team handles 18-22 active clients across retainer and project work. They deliver content strategies, SEO campaigns, paid media management, brand projects, and ongoing marketing retainers. Their monthly output is equivalent to what a traditional agency would need 18-20 people to deliver.
Revenue per head sits at around £220,000 per year. Total agency revenue is just over £1.1 million with a net margin north of 40%. For comparison, the average UK agency net margin is 12-18%.
The tools and workflows
Here is what makes this possible, broken down by function.
Sales and new business
The founder spends roughly 10 hours per week on business development. AI handles prospect research (pulling company data, recent news, social activity, and competitive landscape into a structured brief in under 5 minutes per prospect). Proposals are AI-drafted from a library of templates and past wins, then reviewed and customised by the founder.
The result: proposal turnaround went from 5-7 days to 1-2 days. Win rate sits at 58%, up from 35% before AI implementation. The founder estimates AI saves 15 hours per week on sales activities alone. For more on this, see how to build an AI-powered sales pipeline.
Delivery
Every client project follows a systematised workflow:
- Briefing. Client calls are recorded and transcribed. AI generates structured briefs from the transcript, pre-populated with brand guidelines, past work, and relevant context pulled from the client’s knowledge base.
- First drafts. AI produces initial versions of all deliverables: content plans, blog posts, ad copy, social calendars, SEO recommendations, audit reports. The team member reviews, adds strategic insight, and refines.
- Client reporting. Monthly reports are auto-generated from analytics data, with AI writing the commentary and recommendations. The account lead reviews and adds strategic context before sending.
- Quality control. AI runs consistency checks against brand guidelines, tone of voice, and previous feedback. The senior team does final review.
Each client engagement that would have required 40-60 hours of human effort per month in a traditional setup now requires 12-18 hours, with equal or better output quality.
Operations
This is where the systems lead earns their salary several times over. They have built an automation layer that connects everything:
- Client onboarding is 80% automated. New clients fill in a structured intake form, which triggers the creation of project spaces, brand guideline imports, reporting templates, and introductory documentation.
- Time tracking and invoicing are automated. No timesheets. The system tracks hours through project management tool activity, generates invoices on schedule, and chases overdue payments.
- Internal communications are streamlined. AI generates daily project summaries, flags at-risk deadlines, and creates the agenda for the weekly team sync.
The operations overhead for 20+ clients is managed by one person spending roughly 15 hours per week on it. In a traditional agency, this would be a full-time operations manager plus a finance person.
What they do not automate
This is as important as what they do automate.
Strategy. The strategic thinking behind every client’s programme is entirely human. AI can analyse data and suggest approaches, but the creative leap, the insight that connects a business challenge to a marketing solution, comes from experienced people who understand the client’s market.
Client relationships. Every client has a named person who knows their business deeply. Calls are not delegated to juniors or replaced by dashboards. The relationship is the product as much as the deliverables are.
Creative direction. AI generates options. Humans choose, refine, and push for better. The team has a strong opinion about what “good” looks like and does not accept AI output that is merely adequate.
Pricing and commercial decisions. What to charge, how to scope, when to push back on a brief, when to invest extra effort for a client who is a strong referral source. These are human judgement calls that draw on experience and instinct.
The pattern is clear: automate the production, keep the thinking. This is exactly why AI will not replace agencies, but it will replace agencies that cannot think.
Lessons for larger agencies
If you run a 15 or 20-person agency, you are not going to shrink to five people overnight. Nor should you. But this model reveals several things worth acting on.
Your middle layer is your biggest inefficiency. The coordination, the status meetings, the passing of work between teams. AI can eliminate most of it. Start by auditing how much time your team spends coordinating versus doing.
Junior roles need redefining. The traditional junior role (do the repetitive work, learn the craft over time) is disappearing. AI does the repetitive work faster and cheaper. New junior roles should focus on AI workflow management, quality control, and learning to direct AI output. See how to train your team on AI for more on this transition.
Revenue per head is the metric that matters. If your revenue per head is below £100,000, you have a structural problem that AI can help solve. Track it monthly. Set targets. Make decisions based on it.
Margin is a choice, not a consequence. This team chose high margins by pricing on value rather than time. Your AI efficiency only becomes profit if your pricing model captures it.
The uncomfortable truth
The reason this is uncomfortable for larger agencies is not that five-person agencies exist. They have always existed. The uncomfortable truth is that these small teams are now competing for the same clients, winning the same pitches, and delivering the same (or better) quality of work.
The playing field has not levelled completely. Large agencies still win on brand recognition, specialist depth, and enterprise relationships. But for the vast middle market, the 5-person AI-first agency is now a genuine threat. Small agencies are using AI to compete with bigger ones, and the gap is closing fast.
The question for every agency owner is not whether to adopt AI. It is how deeply to restructure around it.
This is part of Margin Watch, a series on how AI is reshaping the business of running an agency. Subscribe to the newsletter to get new articles weekly.
