Intro
After working at Superpower for 6 months, I realised I wasn't going to get a job. My work there was too boring & low-quality. I didn't have enough context on how to be useful. I couldn't learn quickly enough.
So, I came back to Australia on a mission to build a huge marketing agency. At first, I was genuinely terrified of living back at home, terrified I would have to go to University & so, it was terrific motivation to work hard.
From last July to April (when I finished), was probably the 10 months where I gained the most context on the world (not that I have much). Initially, we spent 2 weeks at Copacabana and paid $2k in rent. We were in 3 rooms that were less than 1 metre from one another. We got our first clients here. We probably had 5-7 clients come from these 2 weeks. It was literally 4 hours of cold-calls & a lot of research on different topics & designing homepages + learning about figma as well. From there, it was 3 months at Pretty Beach. We paid $8-10k/month for this beach-house & we did $200k of revenue here. Good return. Then, we rented a home in Fishermans Parade, Daley's Point for 9 months, spent $5k per month, started to look like an actual company and did $420k of revenue in 7 months from start to finish. It then came to a crashing halt in less than 1 month. I wanted to do advertising as a service, Carnevale didn't care about agencies & Ed wanted control. It ended about 6 weeks ago. It was a really good learning experience, certainly more than one year learning the 4Ps of marketing at USYD.
There are only really two types of businesses created: (1) technical & execution-risk businesses: the superpower's of the world (2) execution-risk businesses. There are marketing agencies in the world valued at $300 million by Private Equity firms. They are gigantic. The enterprise value of the business lies in the ability for the founders to execute, not the idea. We failed. But, I want to write about what I learnt about the industry as a whole, what we did right (very little), what we did wrong & the future of agencies (holding companies, demand + supply curves & impact of AI):
Table of contents:
1. Industry Dynamics: competitors, demand v supply, clientele
2. Selling an agency: valuation numbers, key-man risk
3. Things we did wrong
4. Things we did right
5. Future of agencies: hold-cos & AI
The Marketing Agency Industry:
Like all execution-based businesses, there is generally a formula to find out through reading & experimentation. You can slightly tinker with it, optimise it but just like there are laws of physics, so too are there laws of business that you have to abide by. Here's are a few thing I learnt:
Competitors
Service-based businesses are inherently fragmented. This is because a new entrant comes in the market every couple of days who offer lower prices, do more outreach, take more sales calls. So, there are 3500 advertising agencies in Australia alone. It is still worth categorising them into separate buckets.
You have huge conglomerate agencies. In the agency world, this is WPP, Dentsu, Omnicom, Publicise. These agencies were more financial engineering than marketing. They grew via mergers & acquisitions. They feel & operate like Private Equity companies & holding companies. Sorrell doing his hostile takeover of Oglivy akin to KKR's 'Barbarian at the gates'. A recent one of these is Brainlabs. They essentially started as an Enterprise Google Ads agency after Dan Gilbert worked at Google, got funding from Private Equity & have continued to buy other enterprise clients in different locations, different industry skillsets & different service-offerings: Google Ads to Meta to Tiktok to SEO. These are ultimately the richest people in the industry; Sorrell is worth $800 million USD.
Then, you get individual enterprise firms slightly below this. Eric Siu's firm, Neil Patel Digital, Howatson, Metalab are good examples of this. Have authority via content, build referral pipelines through speaking events & hiring staff of big agencies and layer enterprise clients. They typically stay independent & turnover a couple hundred million; they just don't reach the heights of the biggest agencies, but the founders retain a greater % of equity.
Underneath here you have mid-market agencies. KlientBoost is a good example of this. Other examples here are Directive, Online Marketing Gurus, King Kong & First Page Australia. KlientBoost turns over $50M USD largely doing Google Ads + now they've layered some other services. They couldn't quite reach enterprise from the looks of it, struggle a little bit with churn because of it, are pipeline machines through content, sales & ads. They have decent enterprise value & the founder can be have a net worth between $60-100M, but not genreally higher than that. You really need F1000 clients to be able to build proper enterprise value. Because most enterprise decisions are made by CMOs & Marketing directors, you often have to network your way in, get a little lucky & do the work for a long period of time, rather than brute-forcing growth like you can in SMB markets. All SMB & MM customer acquisition machines. Struggle to get above holy grail 96% MOM retention. Partly because their onboarding is built to sell more
Then, you get tonnes & tonnes of small agencies. <10 employees. Cannot really scale. More like a freelance shop. They typically have super high churn & cannot predictably grow. You also do get $5M ARR shops that are decent businesses like Eric Nowoslawski, Stryker Digital, Taylor Haren, ColdIQ.
Demand & Supply Dynamics
Typically, the same kind of agencies all blew up at the same time. Largely, this is because marketing is predicated on where attention & eyeballs go. Therefore, when a new platform comes out like Google, Meta, TikTok, Linkedin — all the attention goes there & there lies a window to become one of the first players to be able to effectively monetise that attention. Typically, if you're a first-mover in service-based businesses you can jump the totem pole to enterprise clients quicker than you otherwise could have because there are simply less options. Here, you get the dynamic where an agency can get to $5-10M ARR within 5 years, and they are now entrenched as the market leader.
In the early 2000s, the internet was created & companies began experimenting with building websites. Firms like MetaLab offered websites for $10k & they quickly gained mind-share, working with enterprise clients like Slack & other VC-backed startups. They were filling the endless demand for an online presence & from that they were able to gain future hundred billion companies & gain authority. While website firms have still grown like BX Studio, they are slower now. Because there are 1000 firms + new GHC Studio's pop up everyday and demand shrinks 2% YOY as companies prioritise other channels & they already have a website and therefore have no need for a new one.
Similarly, once firms had a website in the mid 2000s, they wanted to rank at the top of the page. When users queried something related to their company, they wanted the ability to control mind-share & attention. Tonnes of firms spawned out of this. In fact, I'd argue this was the biggest wave of successful businesses like OMG, FirstPage, Stryker Digital, Sparrow, Neil Patel Digital. Now, the SEO window is not as optimistic. Every business in the world has been pitched with SEO services & been oversold on a trashy service.
Then, Google & Meta had similar fortunes when these platforms became profitable using adspend where Google collects 80% of their revenue & Meta collects 98% of their revenue. Tonnes of agencies proliferated promising X leads or you money back because businesses wanted to establish digital presence & monetise there. A lot of the biggest firms offer this to enterprise clients. W&H, Brainlabs, Klientboost, Megaphone all examples of quality ad agencies.
Recently, two arbitrages have taken place in the last 5 years:
(1) Content, specifically on LinkedIn. Storyarb & Matt Lakajev have both taken advantage of people wanting to build authority on LinkedIn: both SAAS, consultants & agencies + individuals wanting to build their personal brand. Somewhere between 5-10 agencies built quite successfully here. Some agencies in videography & content on Meta did quite well before as well.
(2) Clay & Email automation tools created a huge demand/supply discrepancy. Every agency/SAAS now sends 20-100k emails per month. However, this was predicated on (a) clay building a tool which aggregated all different SAAS tools which collected data (b) Vaibhav allowing you to spam people for $100/month. Eric, ColdIQ, Taylor Haren, Nick Abraham, Spychalski all built agencies out of this.
Essentially, it is not impossible to enter any market. But, typically to grow quickly, you have to acquire an advantage over other people. Typically, it comes in the form of (1) enterprise connections (2) authority via content (3) starting during a time of supply-demand discrepancy. Without those three, you'll probably need to work hard, eat dirt for quite a while and work up the totem pole to gain that through the quality of your work.
Clientele
The secret of the agency industry is that everyone vies for the same enteprise contracts. They're stickier, longer, greater margin, lower churn & genreally more pleasant to work with because they're livelihood is not on the line. We never worked with a proper enteprise client at GHC: HRMS Group has 60 employees and does $10 million, Stanley has 300 and does $50 million and Rural Med-Ed has 30 employees and does $10 million. So, small players. However, these are the takeaways:
1. SMBs are a nightmare. Customers like Joe, Giovanna, TTS, Advance Formwork all paid <$5k for websites. They ask for endless revisions. They expect perfection & magical leads after the purchase of a website. Building a business ontop of other businesses who are sub <$2M revenue is a bad model. They are inevitably going to be high churn. You could have a perfect business & it still would not work because the customers are the problem. However, generally, if you do not have a network you have to start here & work your way up over many years. Even if you have the network, it's probably from many years of effort.
2. Mid-market start to become better. You generally don't talk to the founder, or at least anyone who has equity vested in the business. So, you talk to NFP's like NECA Training & ACWA. You also talk to bigger businesses like Rural Medical Education, Stanley College, Jenagar.
3. Enterprise is where virtually all the money is. That includes people with enteprise-like budgets such as well funded startups. People can work up to here like KlientBoost, OMG if they culled their SMB staff, but generally it's quite hard to do so. They typically start here from working in-house. Content is probably levelling the playing field here though. Jamie Williamson from Umped does a good job of this & could probably go to Enterprise with speaking events, affiliate partners & loads of content.
Service Offerings & Churn
SEO is super sticky, harder to sell, businesses expect to be in it for the long-term, and there is no equivalent of an ad-spend + retainer, only ad-spend. The idea of "earned media" is also appealing to enterprise as it allows marketing managers at F1000 companies to delay results to their bosses, as opposed to the harsh reality of ROAS numbers on Meta & Google. SEO would be the best business to have started in 2010 because you get a $5-10k/month retainer, can stack enterprise clients, now use AI to speed up blog writing quickly.
Content agencies are also an interesting play. They have too sparse reward signals for my liking, meaning it's difficult to tell the ROI of them. There are angles agencies like Storyarb take which is relating them to attributable pipeline, but ultimately, it's harder to tell for consumer brands that emphasis Instagram & Facebook content. This option is not that appealing. It mainly works for enterprise brands & there is no lower tier brands to work with. If you are sub <$10m in revenue, content is not the biggest priority (especially getting it done externally).
Paid Media is a tough slog. If you fail to hit 96% retention rate on it, you basically are just a sales agency. Very few mid-market & SMB paid-media shops will ever properly scale because of this. Enterprise contracts will generally be stickier. But, even then, Smart cancels contracts left & right.
Websites are super tough. Unless, you have very creative chops, this becomes commoditised very quickly. Companies like Skale Studio, Daybreak, Meta-Lab are all the best website design agencies & they are really really design forward. They target brand-forward companies like VC-backed startups. There are some others like BX Studio that leveraged Webflow enterprise affiliate partners, and also Convert Digital who specialise in Shopify solutions & employ 80 staff and turnover ~$10M revenue per year. There are some adjacent design, dev shops like Deep-end who employ 65 people. If you ran a website business for extremely long periods of time, I'm sure it could be a fine business. A business with 30-50 people that produces solid websites, and turns over $6-7m. Overdose digital pivoted to media as soon as they added it because it scaled faster. Overall, they are all relatively small compared to the likes of Klientboost. I chose websites because I thought if they were design-forward & beautiful like Joumana's you could make a good company out of them. We learnt the hard way that if they aren't design-forward, they become commoditised, you do not get referrals & you cannot charge high amounts.
Cold-Email Agencies. Supply-Demand curves are totally out of whack here. Supply of these agencies went bonkers for a little bit & will probably decrease when most of them cannot get clients. They are solid businesses with loads of companies like Eric, Fatty, Belkins, Haren, ColdIQ & Spychalski. Rev-ops is probably the best adjacent play. You can sell retainers to big SAAS companies & not just rely on reply rate % and total meetings booked. Spychalski does this well. It's a fairly small TAM limited to largely software companies, some marketing agencies & random other B2B companies like manufacturing, telco, industrials, commercial-cleaning. However, there is ability to get paid a lot for a single client. This is largely because B2B contracts are high. If you help someone sell $40k of websites per month & they have 50% net gross margins, you can comfortably charge $5k/month. Though, cold-email is really reliant on the offer of the businesses. So, I'd argue there isn't an overwhelming amount of skill here besides operations & maybe some clay-hacking.
Verticalised agencies seem theoretically cool. Everyone loves to go vertical, following Hormozi. Gym-Launch sold for $46.2M. They had a systematised playbook. Matt Larsen & Charlie Morgan always yap on about this. I think this play is overrated. I haven't seen one other agency do well following this. Most pick one niche & stall at a certain point: Stryker at $5M, Ontologists at $3M. Hormozi says the best option is to conduct a roll-up. Personally, I back going after mid-market & enterprise clients, niching by service, far more than going after a single niche. Partly because most of the spend % in marketing overall comes from enterprises so they care more & you get paid more. If you verticalise, you are still dealing with SMBs? They suck to work with. To be fair, Shaan Hanif has a $100M agency in the creators space & has been able to build products like neutonic & supposedly SAAS for creators as well (though Stan seems to have this space covered).
AI Agencies. We'll see if Alex Lieberman can pull this off. It seems like a software-dev agency for enterprises with some consulting in the mix. I wonder how much enterprises can actually accomplish with AI at the moment, considering context window, token length, memory, computer usage & general reasoning. Enterprise might make sense. The Liam Ottley AI agency seems weird. It seems like hopping on the next wave without thinking through the mechanics of how this kind of business works. Is there a problem? Is the problem repeatable across businesses? Is there genuine value created that SAAS platforms won't intrude on.
Enterprise Agencies. These are the slowest burners, but they are my favourite. Unless you stumble into a $3-5M EBITDA vertical and use that to invest into other ventures like auto shops, software companies, healthcare clinics, home-services, recruitment firms then the rate at which vertical agencies scale seem limited. Morris & Sparrow have 90 staff, have 80 clients who pay $5k on the very low-end, $10-15k on the medium and $20k/month on the high-end. You essentially pay employees $80-100k and take the bit off the top. Enterprise clients stay around for very long periods of time: Morris said 7-8 years. That seems like a better business model.
How to Sell an Agency
An investor wants to essentially come into a low-risk play that continues to spit off cash for long periods of time.
This means: no key-man risk of employees + founders or clients, strong profit margins, >96% MOM retention, sustainable customer acquisition channels, good team more & culture, strong SOPs to fulfil clients & clear-value proposition.
There are specific aspects of agencies that are important.
1. EBITDA valuations vary a lot among businesses. A $20M agency that deals with thousands of SMBS will be less valuable than one with 20 enterprise clients. This is comparing Sparrow to King-Kong. Sabri Suby would probably be richer than Morris, but since his company is solely based on his brand, has incredibly high churn & really bad google reviews, it has no enteprise value. It would spit off cash for him though & they are setting up shop in Santa Monica to do the same thing.
2. Generally, service-based business trade for 1x revenue. If you are sub $2M EBITDA, you probably sell for less than that, a company with between $2-6M EBITDA would probably sell for 4-6x EBITDA, while a company with $10M EBITDA, enterprise clients & low-churn could get 6-10X EBITDA. Agencies EBITDA is particularly low because it is challenging to establish a legitimate MOAT. Anyone with a laptop, enterprise connection can form a formidable business overnight. Investors are very wary of this. Klient-Boost couldn't sell because of lack of enterprise clients. Lindsay can't sell Chello because there is no recurring revenue & no acquisition pipeline system. It is all linked to her.
3. Most huge agencies grow via acquistion. A common myth of agencies is that they're sales beasts with huge pipelines. Some fit this mould, but most of the ones with enterprise value do not. The conglomerates all buy agencies. They leverage debt & use excess cash-flow to buy other agencies. In effect, they are buying the customer list & the staff — not too much else. They grow because you'll never be fired for hiring oglivy. They also grow by hiring people who have good contacts at a firm. Like enterprise sales reps are paid $1-2M for their contact list. Similarly, an account manager could be paid the same here.
4. The risk with developing an enterprise-focused agency is having great founder dependency. Chello is focused on enterprises like Shopify, but all of that comes from Lindsay. It's hard to balance between referral-reliant, but unable to sell because the contracts aren't stick in project-based work land & it's based on Lindsay's network.
Things I did wrong?
Choosing Service Offerings & Recurring Revenue
Websites were a bad choice for us. They are design-forward, project-based work & low perceivable ROI for the end consumer. None of us are designers or seemed remotely interested in design. It would've been better to do advertising or SEO and learn that skill.
We probably would've made less money. Perhaps, we would not have been able to move out of home & certainly not paid for the Daley's Point home. But, ultimately, Tom & I took home $80k of profits which is quite negligible. It is demotivating to have to chase new business every single month, and dealing with the ebbs & flows of the business.
Finding consistent customer acquisition
Cold-calls were the initial acqusition channel. They didn't make sense, but enough hustle + grit made them work. If you do 100-150 calls every day for 4-6 months, you can basically sell anything with a good enough front-end lead magnet and it'll work to some extent. Though, I personally did $300k of sales in 7 months so would've been $550-600k for the year. After you pay $100k for a manager, $130k for sales-rep and $200k for designers + developers, it leaves $100k of margin. That means to make $1M EBITDA you need 40 employees.
For some reason, I chose cold email as our second channel. This was generally quite stupid. A demand capture offer through a demand generation channel. The sheer volume & unique messaging meant it broke-even as we spent $45k-ish on it and sold $90k of websites. However, if we put $45k into google ads, it would've yielded far far better results. Other businesses pay for google ads meaning they are probably profitable, more repeatable and less dealing with tonnes of unqualified leads who do not care about buying. Big mistake on my part. Poor allocation of capital which was tough after grinding for 6 months on calls and never wanting to go back.
Quality of the work
The websites we produced are not that good. Some are 7/10 in today's climate like ACWA, SouthCoast Sports Medicine. However, lots of the websites produced are just not of high quality; the design choices are worse than templates on the Webflow website. BX Studio is the minimum-viable calibre of websites built today. While they are starting to get better, websites like Medilife, Spraysmart, Robinson, Expert Training Group, AVA, Warmset, Safety Adserv are all pretty rubbish. Like 2-5/10 from my perspective. The quality of the design is basically the product. If that design is not good, you are not really producing a good product. This is also my fault. My thesis going in was about beautiful websites. Ed wanted to hire Philippines, make gross margins 80% from Day 1 & I gave up the right because I wanted to scale sales instead. Our company would've collapsed earlier if I had remained steadfast. Nevertheless, this was a huge mistake of mine and one I didn't realise till it was too late.
Delivering Good Service & Building SOPs
Projects took far too long. There are projects that are still not finished, 8 months later. Ed did not hustle nearly enough. They all dragged on. He lost motivation at some point, did not care & built an expectation that the money was going to continue coming in. We did not get one referral & I think this certainly contributed. Again, the onus is on me to (1) take some control of clients (2) motivate him to do the same if I do not have control.
Choosing Cofounders & Team Dynamics
There are 2 comments here:
(1) I should not have partnered with Ed & Tom in the first place.
(2) The team dynamics from the start felt wrong and I couldn't address them. Tom & Ed seemed to repeatedly gang up on me, but maybe I was sensitive. Perhaps, they did so in response to my mood being quite volatile due to nicotine & caffeine. I will never hang out with both of them again together. It is torture. I certainly contributed to this with my mood swings + spending a lot of time in Sydney at points + not being communicative when I was on holiday or in Sydney.
Capital Allocation
Generally, we spent money on things we shouldn't have. The house was probably okay. But, too many software tools for cold email. Posters for the house. Cold email in general was a poor allocation of resources. We spent far too much on designers & developers. It cost nearly $180k by the end making gross margins 60% before factoring in project managers.
Things we did right?
Hustled
Basically, the company got off the ground because I sold $300k worth of websites between June till December 15th. I did this first through making 100-150 cold calls everyday and sitting on 2/3 sales calls per day & then harassing people over email. Then, we spent 1000-5000 emails per day via SmartLead & sold 80k there. Essentially, we got 45 people to put in an initial deposit of money in 6 months from a standing start with no warm network, 1 referral the entire time. Not a bad display of consistency & grit.
Ed did a good job of managing the process, and ensuring clients didn't churn like what happened with International Fitness & Joe at the very beginning. He presented well with a beard, handled changes decently.
Future of Hold Cos & Agencies
Hold-Cos
I am going to write a longer piece about this because I thought quite deeply about this area given this is the natural extension of virtually all service-based businesses.
There are countless people playing in this field at the moment with some notable players being Jess Mah, Codie Sanchez, Peter Kang, Alex Hormozi, Brent Beshore & Enduring Ventures. They are all started from the same position off bootstrapping/raising small amounts of money for a service-based business, scaling that & rather than re-investing the excess cash-flow into the existing business, you see opportunities to buy/invest into other businesses, roll them up.
Effectively, this decreases risk by not having all your eggs into one basket with AI posing potential dangers to a decent amount of businesses. But, mainly because after a certain point the additional $$$ you get from re-investing excess profit flows into the existing model diminish. Most good businesses compound overtime without needing the founder to continually pump away: content gains more authority, ads reach ROAS max before it's not worth scaling, product/service reaches asymptote. So, it makes sense to diversify the portfolio and invest into other types of businesses: recruitment, healthcare clinics, construction firms, home-services businesses like electricians, plumbers.
AI
We thought about this quite deeply at the agency. Overall, my thesis is that agencies won't go away because the purpose of good agencies is not execution, but strategy. Companies could always hire in-house, but they choose the agency because they are seeing data across hundreds of clients.
Some thoughts/questions:
1. AI will disrupt execution-based marketing quite a lot. Performance Max campaigns & Meta+ will make digital media-buying irrelevant. ChatGPT will continue to mean blog-writers are used less & less for one piece. There are already AI-logo generators, mediocre website builders like SquareSpace AI. Basically all Relevance AI agents are build for marketing workflows.
2. There is the obvious argument that marketers will just make MORE stuff. Or be able to focus MORE on their existing work. I'm not convinced. I think there are asymptotic limits to the creation of work that means an extra 100 hours on a blog post won't necessarily make it far more effective. Max & superpower believe in convex returns on effort though. This could be true & mean more employees devoted at getting the end-product from 95% great to 99% will become what more humans spend their time on.
3. Taste & creativity still seem super important. I am not sure if this is hackable via machine learning. If you train a model based on past data, can it come up with net-new concepts? Can Icon make new ads? Can AI-website come up with novel website concepts? Can you generate net new UGC creative? I am not convinced. Next-token prediction seems based on past data which is useful. But, great marketing is very creative & novel.
4. SMBs could be eliminated as customers. If a customer has sub $3M in revenue. They might just use Meta AI to create their ads. Even if they could improve ROAS by 25% with an agency, it's not worth the ad-retainer + hassle of relying on someone else. This could mean agencies go out of business because SMBs are not available for quick cash hits. There is also a world where SMBs are just too lazy to do the ads themselves and hence why they are SMBs in the first place.
5. Do AI agencies emerge as another form? To me, they are a BI/Data Analysis/ML/Software Dev type of agency & not really a marketing agency. But, interested to see how Tenex does. Side Note: my sense is that running a software dev agency is firstly not a great business because project-based work, need to crack enterprises because they have custom requirements that no one system can crack & it seems weirdly. Though, the big 5 IT firms obviously do quite well. But, I don't think they're great for learning to build software companies anyway because most are enterprise clients whose software is built on their systems, not other people's systems.
6. Marketing seems like a slightly overrated skill. Alex Lieberman & Hormozi are rich because of marketing. There are a lot of CPG brands rich off marketing to be fair. But, the idea of "honing your craft" seems a little overrated. Tinkering with Google Ads landing pages + keywords, Meta Creative, Cold-Email copy all seem very specific to the type of business you run rather than neccessarily having to build that skill. Building a brand online seems the exception to this & one to be constantly tinkering with. Though, that is deeply personal. I doubt you neccessarily get 10x better by controlling a lot of other people's brands. To be fair, a lot of