Startup Lens #01 - Do Things That Don't Scale

What early-stage startup thinking taught us about building a music management company

💡

There is an uncomfortable stage in building something new where almost everything feels inefficient.

You send emails individually. You introduce yourself to people who have never heard of you. You attend shows for one conversation that may lead nowhere. You keep spreadsheets manually. You remember individual artists, photographers, journalists and promoters. You solve problems one by one.

From the perspective of an established company, much of this looks unsustainable.

From the perspective of an early-stage company, that may be precisely the point.

Paul Graham's Do Things That Don't Scale argues that startups often need to begin with work that cannot initially be automated or efficiently reproduced. Early users rarely appear simply because a good product has been created. Founders have to recruit them, work closely with them and pay attention to experiences that would be impossible to provide once the company becomes significantly larger.

Reading this while building Cornerstone made us reconsider some of the work we had previously thought of as temporary inefficiency.

Perhaps some of it is actually how we're learning what the company is supposed to become.

Before there was a network, there were emails

Some of Cornerstone's earliest development came from a very simple process: finding people and contacting them individually.

Around Happy Neighbours' early 100 Club activity, there was no sophisticated industry network waiting for us. Relationships with photographers, journalists, bands, promoters and other people around the project often began with individual messages and cold emails.

Most went nowhere.

A few became conversations.

Some of those conversations became relationships.

And a small number of those relationships opened entirely new parts of the network.

It would be easy to look at this process and conclude that sending emails manually is inefficient. It is.

But efficiency may be the wrong measurement at this stage.

Every unsuccessful venue pitch taught us something about how venues are programmed. Following those questions led us towards external promoters, which led us towards understanding promoter networks, which changed how we thought about entering the East London live ecosystem.

The important output wasn't simply the reply rate.

It was market knowledge.

This suggests an important distinction:

Doing something manually is not automatically valuable. Repeating manual work without learning anything is simply inefficient.

The value appears when the process becomes:

<aside>🔎

manual action → observation → discovery → pattern → framework

</aside>

Eventually, that framework can become a database, an SOP, a service, a training document or something another person can execute.

The unscalable work is therefore not necessarily the business model.

It can be how we discover the business model.

Being small is not just something to survive

Early companies often feel pressure to behave like established companies.

We build departments, write job descriptions, create workflows and try to appear professional. Cornerstone has been doing exactly this as our responsibilities have expanded.

There is value in that structure. Without it, knowledge remains inside individual people's heads and the organisation becomes increasingly difficult to coordinate.

But there is also a danger in professionalising too early.

A five-person company should probably not behave exactly like a fifty-person company.

One advantage of being small is precisely that we can still care about individual relationships in ways that larger organisations may struggle to reproduce.

A journalist isn't simply an entry under "Media".

A support band isn't simply a supplier for a lineup.

An emerging artist isn't simply a lead.

At this stage, we can still remember the conversations behind those relationships.

One experience that made this particularly visible to us was working with TopNote. The relationship wasn't built through one perfectly optimised press email. It developed through communication before the event, attention during the show, conversations between people with overlapping experiences in the industry, and continued contact afterwards.

None of those things are particularly scalable.

But together they create something a CRM entry cannot: trust.

Paul Graham argues that early companies can provide a level of individual attention that would be impossible for a large organisation, and that this intense attention can eventually become part of company culture rather than merely an early acquisition tactic.

For Cornerstone, the question therefore isn't simply:

How quickly can we automate relationship management?

It may be:

Which parts should eventually be systemised, and which parts should remain deliberately human?

Systems should capture learning, not imitate scale

This creates an interesting tension.

At the same time that we're arguing for highly personal, manual work, Cornerstone is building databases, SOPs, function maps and internal workflows.

These ideas aren't contradictory.

The purpose of a system should not be to remove human involvement simply because automation looks more professional.

A useful system captures what repeated experience has taught us.

For example, if we manually contact fifty venues and discover recurring patterns in how their programming works, the next step isn't necessarily to automate fifty more emails.

It might be to change the research process entirely.

That discovery could become a Venue Pitch SOP:

research the venue → identify who actually programmes relevant nights → understand their existing lineups → identify an appropriate route → then pitch.

The system becomes better because the manual work happened first.

This is a useful principle for how we want to build Cornerstone:

Don't systemise assumptions. Systemise discoveries.

Finding artists may require two different acquisition models

Graham also distinguishes between products created to solve a founder's own problem and products aimed at users the founder doesn't already understand. In the first case, early users may naturally be people similar to the founders themselves. In the second, founders may need to recruit more broadly, observe who responds most strongly, and gradually identify the most promising market.

This distinction feels particularly relevant to Cornerstone because our potential services don't all reach people in the same way.

Our band and industry network is largely relationship-led.

We meet artists at shows. One band introduces another. A support opportunity starts a conversation. A photographer introduces a publication. A promoter introduces another promoter.

The acquisition process is predominantly physical and network-based.

But services such as artist development consultations, positioning sessions, educational resources or future development programmes may behave differently.

We don't yet know with certainty who will value those products most.

A self-managed artist?

A young manager?

A band that has reached a plateau?

An artist preparing for their first serious release campaign?

Instead of deciding the answer internally, an early-stage approach would be to expose a relatively simple version of the service to a broader group, observe who engages most strongly, and allow those early adopters to help define what the product becomes.

In one part of the company, we are building through network acquisition.

In another, we may need to learn through market acquisition.

Treating both as generic "marketing" would hide an important difference.

The first customer doesn't need to be the richest customer

This also changes how we think about early customers.

The most valuable first customer may not be the one willing to pay the highest price.

It may be the person who immediately understands the problem we're trying to solve.

Imagine two artists.

One pays £500 for a development package but barely engages with the process.

Another pays £60 for an artist positioning consultation, immediately restructures their EPK, tests the framework, returns with questions and shows us exactly where our methodology works and where it fails.

At an early stage, the second artist may be more valuable to the development of the company.

Graham similarly argues that when choosing between users who adopt quickly and those who simply have more money, early adopters can be particularly useful because their engagement helps shape the product.

Revenue matters.

But early revenue and early learning are not always the same optimisation problem.

Cornerstone needs both - and needs to know which one we're pursuing at any particular stage.

You can't outsource understanding how to sell yourself

One of the more uncomfortable ideas in Graham's essay is that founders cannot avoid learning sales simply by hiring somebody else to do it. At the beginning, they need direct exposure to the process themselves.

The same principle has an interesting parallel in artist management.

A management company can sell an artist.

It can pitch promoters, contact journalists, create opportunities, organise campaigns and introduce the artist to industry networks.

But management cannot completely replace an artist's ability to represent themselves.

The artist still has to walk into the room.

They still have to perform.

They still meet the other bands.

They still speak to audiences.

They still participate in promotion.

They still decide whether somebody leaves the evening wanting to work with them again.

An unhealthy relationship would look like:

management creates opportunity → artist consumes opportunity → management finds another opportunity

A healthier cycle is:

management creates opportunity → artist activates it → new relationships and outcomes emerge → management compounds them

The same is true of Cornerstone itself.

As the company grows, individual functions can and should eventually be delegated.

But delegation should happen after we understand the function well enough to know what good performance looks like.

Maybe not everybody needs to become a managed artist

One small note at the end of Graham's essay triggered a larger question for us.

He points out that Google benefited enormously from Yahoo, but Yahoo wasn't Google's partner. Yahoo was its customer.

That distinction made us reconsider how we think about Cornerstone's relationship with emerging artists and managers.

The conventional management model can encourage binary thinking:

we manage you / we don't manage you.

But there may be a much larger space between those two outcomes.

A self-managed artist might use Cornerstone for positioning or development support and remain completely independent.

A new artist manager might use our frameworks, knowledge or resources while building their own roster.

Another artist might work with us on a short development project, achieve a specific objective and leave.

And some relationships may eventually develop into long-term management.

That means creating value for somebody doesn't necessarily require absorbing them into the company.

In fact, early customers who are themselves building something may be particularly interesting. Graham notes that startups can make unusually valuable early customers for other startups because they are flexible, willing to experiment and capable of growing alongside the companies serving them.

Perhaps self-managed artists and emerging managers occupy a similar position for Cornerstone.

Instead of asking only:

Which artists should we sign?

we can also ask:

Who can grow using what we're building?

That creates a very different business.

The point isn't to stay unscalable

There is one important limitation to all of this.

“Do things that don't scale” shouldn't become an excuse for organisational chaos.

If one person remains responsible forever for remembering every relationship, sending every email, fixing every problem and personally delivering every service, the company hasn't discovered an advantage.

It has created a bottleneck.

The purpose of intensive early-stage work is to learn.

Some things should eventually become databases.

Some should become SOPs.

Some should be delegated.

Some should become products.

Some should be automated.

And some, particularly high-value human relationships may deliberately remain personal.

The difficult work is knowing the difference.

For Cornerstone, we currently think about that progression as:

Manual Action → Discovery → Pattern → Framework → System → Delegation or Productisation

Not every action needs to reach the end of that chain.

But if the same manual task keeps appearing, we should understand why.

Early-stage companies don't have the scale, reputation or infrastructure of established organisations.

What they do have is proximity.

Proximity to artists.

Proximity to audiences.

Proximity to problems.

Proximity to feedback.

And proximity to the people they're trying to serve.

Perhaps the goal at this stage isn't to escape that closeness as quickly as possible.

It's to use it while we still can.

Cornerstone Discovery - Startup Lens

Startup Lens explores ideas from entrepreneurship and business thinking through the realities of building an independent music management company. These aren't startup principles copied directly into the music industry, but working reflections on where those ideas hold, where they don't, and what they change in practice.

Reading behind this Discovery: Paul Graham, Do Things That Don't Scale.