Show Notes

Most agency owners didn’t start a business because they wanted to look at spreadsheets - they started it to do the creative work. But the agencies coming through a hard year are almost always the ones whose owner is closest to the numbers. In this episode, Callum puts Gareth in the hot seat on why your accountant is not watching your numbers for you, the three numbers that belong on every agency dashboard, why the net profit figure agencies quote is usually not the real one, what utilisation actually tells you about capacity, and why the Monday morning version of this job starts with a piece of paper rather than a dashboard.

Subscribe to The STANDOUT Agency Podcast

In this episode

• Why the pattern is so consistent - the agencies that are well run are invariably the ones whose owner is close to the numbers, and where an agency is struggling the finances tend to be the thing being avoided, because they are painful viewing

• The mentality Gareth sees across UK small agencies: the belief that your accountant is keeping an eye on the numbers for you - and why that is fundamentally not true. Accountants are scorekeepers, they are not running the finances of your business for you

• The first number, cash in the bank - what operating cash actually means, why you need a floor you never let it drop below, what big outgoings to build in before you read the figure, and why two months of operating costs is the minimum before the runway gets risky

• The second number, profit - and why the net profit figure agencies report is so often clouded. If the directors are paying themselves in dividends, a reported 30-35% is an accounting margin rather than a true commercial one. 10% is the floor, 20 plus is better, and the 30s if you can get there

• The third number, utilisation - time is the agency’s currency, 70-75% chargeable is what to aim at, and why 100% is not the goal. Plus how embedding AI lifts utilisation for free, because every hour saved on unbillable work goes straight to margin

• What separates the agencies that are good at this from the ones that aren’t - awareness is only half of it. Nothing changes if nothing changes, and the owners doing this well are the ones acting on what the numbers show, including the difficult decisions on people costs

• Why almost every other priority is downstream of the numbers - you can only build a strategy once you know where you actually sit, and knowing which service is the biggest is not the same as knowing which one is the most profitable

• Where to start on Monday morning - make the commitment, start with those three numbers, and don’t make building the dashboard the job. Start with a piece of paper, not with AI

No agency owner loses sleep over a missing acronym. Nobody’s ever said what this business really needs is a Standard Operating Procedure. But ask Callum what comes up in nearly every AI conversation he has with agencies this year, and it isn’t a tool, a model, or a prompt. It’s process, or rather the absence of one, sitting quietly underneath almost every stalled AI rollout he’s seen.

In episode 10, Gareth turns interviewer and puts Callum in the hot seat on why he’s become just as obsessed with SOPs as he is with AI, why creative agencies almost never have them, and why bolting AI onto an undocumented process is how you end up with a very expensive stranger in the building.

In this episode

SOPs aren’t agency language, and that’s the problem. Processes, workflows, tacit knowledge in someone’s head - agencies have all of it, just not written down. In manufacturing, a missed step breaks the production line, so it gets documented. In a creative business it’s absorbed as skill and trust: the department head knows how the work gets done, and everyone else learns it from them. That’s fine right up until you try to scale it, train someone new against it, or work out whether it’s actually efficient.

AI is an alien in the building, and needs everything spelling out. It doesn’t carry the industry’s tacit understanding the way a ten-year team member does, so it needs the process in what Callum calls painful detail. Prompt libraries were the old workaround, and they’re already dying out. A standardised workflow is the upgrade: instead of telling AI what to do every time, you tell it once, properly, and it repeats the process rather than repeating your instructions.

The blueprint session. Callum’s method for building one from nothing: sit down with whoever actually knows the process, usually the senior person who no longer does it hands-on, and go back to the complete first step. An hour to ninety minutes later, there are five to ten defined steps and a clear view of where AI should interject and where it shouldn’t. His rule of thumb: learn through AI, not from it. Feed it the agency’s own quantitative data first, then let it interview you on the qualitative gaps.

Meeting transcription is doing this work already, most agencies just haven’t noticed. Record the onboarding conversation where a department head talks a new starter through how something gets done, and you’ve built documentation and fed AI its context in the same hour, with no extra step required.

The bullet train needs tracks. Gareth and Callum’s analogy for the episode: the SOP is the track, AI is the engine, and it used to be a steam train, powered by people. Put a bullet train on tracks that were never properly laid and it derails. Having it written down isn’t what makes a process rigid, either. It’s what lets you deviate from it on purpose, in a specific case, rather than by accident because nobody agreed what the process was to begin with.

Speeding up shouldn’t mean earning less. A callback to the agency habit of charging by the hour: every time AI shaves time off a job, the time-based model quietly punishes you for it. Gareth’s version is the taxi meter versus the train timetable - efficient work needs a set destination and a set time, not a metered wander. Having the process documented is what lets you benchmark the before and after, be transparent with clients about the gains, and decide how much of that speed you keep.

What Callum wants listeners to do this week. Get the whole team onto one AI plan, so whatever gets built once is available to everyone. Then take whichever SOPs already exist, quantitative or not, and get AI to interview you on what’s still unclear until the picture’s complete. For the processes that only live in someone’s head, start with the pain, not the SOP: think about who in the business is constantly repeating the same explanation to someone new, and build the track there first.

Nobody wakes up in the middle of the night wishing they had more design. Nobody wants a website, or SEO, or a retainer. They want what those things get them - more customers, more sales, a business that’s easier to find and easier to trust. Which is obvious, and yet almost nobody in the industry sells that way.

In episode 9, Callum puts Gareth in the hot seat on why the standard agency model - a list of services offered out into a market already full of near-identical lists - is under threat, and why the answer is to stop selling services and start selling named, fixed-scope, outcome-led products.

It’s a conversation that takes in a plumbing leak, a pitch day, and a workshop Gareth ran this week that surprised the agency in the room.

In this episode

The business model is under threat, not just the marketing. Thousands of agencies, more added every day, all offering the same services under the same names. Add AI doing work that only people could do a few years ago, and skills that were once opaque (design, before Canva) now open to anyone. Differentiation is part of the answer, but Gareth’s point is wider than that: it’s a different way of working.

Nobody wants the diagnosis. They want it fixed. Callum on his own STANDOUT AI audit: nobody wants the audit, and they don’t really want AI implementation either. They want more margin, more growth or more control. Gareth’s version is the plumber. When the leak appeared he didn’t want a plumber, he wanted the leak fixed, and wouldn’t have cared who or what did it. A plumber sells an urgent need and never has to dress it up. Agencies solve needs just as real - falling market share, not enough customers - but the need is latent rather than urgent, so it has to be made visible.

Hard to sell, and harder to buy. Every brief gets scoped from scratch. Meanwhile the client is looking at thousands of agencies listing the same services, without fully understanding what those services do or in what order they need them. The pitch day exists because clients are trying to de-risk a decision they can’t otherwise make.

Why vague offers get priced down. A callback to episode 3, Nobody Agreed the Height of the Wall. If the offer is vague, the client compares on price and the agency spends more unbillable time to earn the same money. Callum sees it in estimation: a blank page every time, and every hour of it eroding margin. AI, usefully, forces the issue - you can’t use it to estimate until the service is defined well enough for it to know what it’s estimating. He’s working with a client on exactly that, tiering new projects against the data from old ones.

What a productised service actually is. Not a new service. The skills you already have, packaged around the outcome for the client rather than the service itself: a name, a defined scope, a fixed price, a stated timeframe, and clarity on what the client gets and how it’s delivered. The service still sits at the core. But a client can say “I can buy that” and know their commitment and their risk - rather than signing a retainer whilst quietly thinking “let’s see how it goes”. And it gets better every time you sell it: run the same product three or four times and you can see what worked, what was missing and where the effort was wasted.

Name the problem, not the deliverable. Gareth’s caution: this is not tweaking a few words on the website, it’s a shift in how the business runs. And a productised service has to solve a client problem, so “Brand Identity Sprint” is the wrong name for something whose real job is telling the market who you are, what you stand for and why they should pay. The headline is the problem. The method comes after.

Three types of productised service. The productised service proper - your existing skills, packaged. The technology or AI product - a tool or skill the client uses directly, no human in the loop. And the hybrid in the middle, where technology does the work but you still guide the client through it. The test is whether the client interacts with the technology or whether it’s your internal tool. Two things trip agencies up on the hybrid route: hourly pricing (do it faster, invoice less) and the absence of standard operating procedures - which is an episode of its own, coming soon.

What Gareth wants listeners to do this week. Get your senior people in a room and list the problems your work solves, not the services you offer. Gareth did this with an agency this week and everything on the whiteboard fell into four clusters - and the highest-value clusters weren’t the bulk of the work. Being fast and nice to work with doesn’t count; those are hygiene factors. The value is commercial. Then build the product backwards from that.

Ask anyone selling AI into agencies whether it’s coming for people’s jobs, and you’ll get the same reassuring line every time. Ask them what AI still can’t do - even after swallowing an entire book, framework and philosophy whole - and the answers get a lot more interesting.

In episode 8, Gareth puts Callum in the hot seat to unpack why 300,000 years of human wiring can’t be replicated from a training set, why the right analogy for AI isn’t a robot takeover but a calculator, and the paradox sitting underneath every agency quietly handing its junior-level work to a machine: if AI does the reps that used to build an expert, where does the next expert actually come from?

It’s a candid, occasionally self-implicating conversation (not least because Callum promotes AI adoption for a living, and still doesn’t want to be replaced either).

In this episode

Why Callum trusts AI less the more he learns about it. He’s a genuine advocate, blown away by its utility, and yet the more he understands how it actually works, the less he trusts it unsupervised. The simplest illustration: AI hallucinates because it’s mathematically predicting the next most likely word, not looking anything up — and it’s trained to guess rather than admit it doesn’t know. Even 99% certainty is a bad number for a business to run on.

The apprenticeship paradox. Judgment has always been built through reps - the boring calls, the repetitive spreadsheet trawling, the meetings nobody wanted to sit in. That’s precisely the category of work AI is best at, and precisely why agencies hand it over first. Which raises the real structural question: if a machine is doing the reps that used to turn a junior into a senior, what replaces the apprenticeship as the way judgment gets built at all?

Through AI, not from AI. Gareth and Callum’s answer isn’t to protect the boring work - it’s to make sure senior team leaders build AI systems that actually contain their knowledge and workflow, so juniors are learning the system a senior built, filtered through AI, rather than learning from AI directly with nobody senior in the loop.

Why the felt half of a decision can’t be delegated. AI can read the facts of a situation but not the room. Two people can sense tension, know when to soften something, feel what’s coming.

Calculators didn’t retire mathematicians. Callum’s pushback to the obvious “but shouldn’t AI keep learning too?” challenge: a calculator changed what a mathematician spent their time on, it didn’t replace the maths. But you still need to know the maths to use the calculator well - which is exactly why schools taught mental arithmetic and calculator use as two separate skills. Same split now applies to specialism and AI.

The STANDOUT or Die test, turned on its own author. Feed the entire STANDOUT framework into an AI system and it will recite it back fluently - word for word, on demand. It has never sat in a room, watched a face go quiet, or felt a deal nearly die. A framework in an AI system is a starting point for consistency. It is never a substitute for someone who has actually been in the room.

The line they land on. AI does the arithmetic - finds the hours, calculates the cost, spots the pattern. That part isn’t up for debate; it’s faster and better at it than any person. Humans still do the judgment: what that number means for this client, this relationship, this risk, right now.

What Callum wants listeners to do this week. Write down what your business believes and doesn’t, what it does and doesn’t do - then use AI to pressure-test decisions against that. Never to make the decision itself.

Most agency owners know their own marketing could be better and most people think more time is the answer. In episode 7, Callum puts Gareth in the hot seat to unpack why Morgan Freeman’s character in the movie The Shawshank Redemption explains the agency marketing problem better than any busyness excuse ever will, why hiring a marketing manager can create more problems for the founder that it fixes, and the one brief every agency owner needs to write for their own business before anything changes. It’s a practical, occasionally uncomfortable look at why agencies that are brilliant at client work go quiet the moment nobody’s asking them to be.

In this episode

Why a scene from The Shawshank Redemption, of all things. Gareth’s hook for the episode is Red - the character who spends so long in prison needing permission for everything that, once free, he still can’t use a work toilet without asking the manager first. The bars are gone. The habit isn’t. That’s what Gareth thinks is actually happening to agency owners and their own marketing.

It’s almost never about time. Ask ten owners why their marketing gets neglected and nine say they’re too busy - but the same owner will drop everything for a pitch at two days’ notice. Gareth’s point: capacity shows up the moment something outside the building demands it, which proves the excuse was never really about hours in the day.

The four things client work always has, that your own marketing never does. A brief written by someone else, a budget signed off by someone else, a deadline set by someone else, and a person at the other end you want to impress. Take all four away at once and the work has nothing telling it to start.

What agencies have actually been institutionalised by. Years of receiving a brief, interpreting it, making something, presenting it and getting judged builds real skill at responding to someone else’s requirement but no muscle at all for generating your own. Gareth argues that’s a missing permission structure, not a missing capability.

The harder truth underneath it. Marketing yourself properly means genuine narrowing (committing publicly to a type of client and turning down enquiries you’re capable of servicing). Easy advice to give a client. Much harder when it’s your own revenue, and your own bank account feeling a quiet quarter.

Most owners already agree but do nothing anyway. It’s rarely that owners don’t understand the diagnosis. What’s missing is the discipline to actually deploy the fix.

Briefing yourself, in the exact format you’d demand from a client. Who it’s for, what it costs, who signs it off, when it lands, and - the piece almost always missing - who it actually gets presented to. Gareth argues that last one is what forces the rest of it to happen.

Why hiring a marketing manager doesn’t solve it. Most owners think they’ve fixed “who owns it” the moment they’ve hired someone to run marketing. Gareth pushes back: that devolves the execution, not the accountability. Someone still has to occupy the client role - briefing it, expecting it, reviewing it - or the work stalls exactly the same way.

What Gareth actually wants listeners to do this week. Not a demand to build the whole system overnight, just to acknowledge that this is genuinely the issue in their own business, and commit to changing it.

Shadow AI is already costing agencies real money and real clients - and most leadership teams don’t know it’s happening under their noses. In episode 6, Gareth puts AI governance specialist Callum in the hot seat to unpack why free and personal AI plans are an agency’s biggest hidden liability, the real story of a marketing agency hit with a massive insurance claim after a data leak, and why “all or nothing” AI connectors can expose everything from client contracts to confidential leadership conversations. It’s a practical, occasionally uncomfortable guide to building the governance foundations that let agencies actually use AI without putting the business - or their jobs - on the line.

In this episode

What AI governance actually means. Callum frames it simply: using AI to benefit the agency while mitigating risk — ideally getting it down to no risk at all, so adoption feels safe rather than reckless.

Shadow AI is still the biggest problem. Callum still sees agency staff using free or personal AI plans for company work without leadership’s knowledge. Free plans in particular can use chat data to train models, and without UK GDPR agreements in place, providers can theoretically do what they like with it. The fix is blunt: abolish free and personal plans for business use entirely, move everyone onto paid team/enterprise plans, and reflect the tool and its data obligations in client MSAs.

The horror story. An agency lost a marketing client’s sensitive campaign data after an employee uploaded it into a free AI plan - the data surfaced elsewhere, the client traced the breach back to the agency, and the agency was hit with a massive insurance claim. A personal plan is never a legal “get out” for the business; the client comes to the agency’s door regardless of who used what.

Architecture isn’t enough - you need compliance too. Callum audited a 60-person agency with genuinely excellent AI governance: enterprise Gemini, model training off, updated contracts, a formal tool-approval process. Yet analytics showed two-thirds of the team weren’t using the sanctioned platform at all - still on personal ChatGPT. The lesson: policy without education and enforcement doesn’t change behaviour. Callum’s view is that breaching a “no personal plans” policy should be treated as a fireable offence, given how severe the fallout can be if data leaks.

Give people the “why”, don’t just give them rules. The whippet analogy: you can train a dog to follow a command without it understanding why, but people can understand the reasoning — and will comply far more reliably if they grasp the scale of what’s at stake, rather than experiencing it as leadership being controlling.

Connectors are all-or-nothing - scope them properly. Turning on a connector (Office 365, Google Workspace, etc.) grants the AI everything that the connecting login can see. If that’s the founder’s login, the AI can see financial data, HR records, even confidential leadership meeting transcripts. The fix: work with the agency’s IT provider to create a dedicated, scoped login for the AI tool, starting with access to nothing and adding data in only as it’s needed — bottom-up, not top-down.

Client data needs explicit permission. Don’t connect AI to client data unless the client has confirmed they’re comfortable with third-party AI processing, regardless of what UK GDPR compliance the AI plan itself offers.

Client conflicts of interest are a real, distinct risk. Unlike a human team member, AI won’t discriminate between competing clients’ data if it has access to both — so access needs to be separated by client team, not just by role.

A real cautionary tale. An agency on a free Gemini plan hadn’t locked down its Google Workspace. An employee facing an internal grievance searched their own name and surfaced unfavourable meeting transcripts about themselves - which they then used against the company. A direct consequence of an ungoverned connector plus a free plan retaining data it shouldn’t have.

Exports are the blind spot no connector scope can fix. Even with a perfectly scoped connector, anyone can manually export a file and drag it into an AI chat, bypassing every access control in place. The only real defence is policy and communication - “if it’s not already in your AI environment, don’t drag it in” - backed longer-term by IT-side file classification.

For highly regulated clients, start internal-only. With a health comms client holding regulated patient data, Callum’s approach was to leave client-data connectors off entirely and instead build safe, high-value automations on strictly internal data - timesheet processing was the live example, scoped so the AI could only ever see one folder.

Plugins as governance and future-proofing. Packaging an agency’s context and workflows into a version-controlled “plugin” (Callum hosts his on GitHub) creates a single source of truth, stops workflow version-sprawl, and - critically - keeps the agency’s real asset (its context and data) portable if an AI vendor changes terms or hikes prices. The goal is to stay tool-agnostic: the AI is just the processor, the context is the value.

Governance comes first, always. In every roadmap Callum builds, governance is month one - sometimes week one - before any efficiency or ROI work. The analogy: build the house on solid foundations, not sand. Skip it, and agencies risk discovering months into a build that a client never actually consented to AI use on their data, burning significant time and money.

The unexpected upside. Beyond AI safety, the governance process forces agencies - often unstructured, fast-moving businesses with little formal process - to build real structure and documented ways of working, benefiting the whole business, not just AI adoption.

Every agency founder starts as the engine: the seller, the doer, the one everyone depends on. But the skills that get an agency from zero to £1m and a dozen people are exactly the skills that stop it going further. This week Callum asks Gareth about his own experience of running an agency and why agency founders find it so hard to let go. They unpack four intertwined barriers, from simple resourcing to something far more personal, that keep founders stuck being indispensable. If you've ever wondered if you're the reason your agency can't grow, this one's for you.

Your AI sounds clever, but Callum's argument is blunter: it doesn't understand a word it says. It's a prediction machine - and without your agency's context, it can easily predict the wrong things.In this Episode, Gareth and Callum discuss the mechanics behind that claim, the demand amongst agencies for a "company brain”, and why most agencies are simply not set up to feed their brain the information it needs. They then explore the unglamorous Monday-morning starting point: a folder of markdown files.If you've ever pasted a brand doc into a chat window and called it AI adoption, this one's for you.

Gareth wrote a Substack post recently about a garden wall — inspired by a pricing conversation on one of Callum's client projects where nobody, not the agency, not the client, had actually agreed what they were building. That's this episode's strange title, and the starting point for a much bigger problem in how agencies price their work.In this episode Gareth and Callum unpack why marketing agencies have stayed hooked on hourly billing for decades, why AI is finally forcing the issue, and what it actually takes to move to value-based pricing — clarity on the brief, clarity on the outcome, and clarity on what that outcome is actually worth to the client. They get into why hourly pricing quietly punishes agencies for getting faster, and how AI itself can be used to interrogate your own service delivery and surface the value you're not currently charging for.

Callum has noticed a pattern to some of his recent AI engagements with agencies. It’s a surprising one and hardly glamorous - digging meaningful data out of their timesheet system. In this episode Gareth, a 30-year timesheet veteran, interviews Callum on why the AI lead, of all people, wants agencies to keep recording time - and why ‘AI means we don’t need timesheets’ is exactly the wrong lesson to draw. They get into what that timesheet data is actually extracted for, where AI genuinely will make time recording irrelevant, and why an agency with no data on its biggest cost line - human hours - is starting its AI journey from the worst possible place.