Why are digital agencies so awful?

Not because they’re lazy. Not for a lack of ambition.

I've got five reasons.

1. They have no memory.

Agencies treat every requirement as novel because it’s novel to them.

Because they are structurally incapable of learning.

No institutional learning = No compounding advantage.

Each project is a pilot episode of amnesia.

They’ll tell you your business is unique. What they mean is: we didn’t retain anything from the last six clients who looked exactly like you.

2. No methods. Only vibes.

Agencies invented vibe-based methodologies because real ones require time, investment, and the unbillable sin of thinking.

So instead you get an improv show.

Agile jazz hands.

And an exhausted audience (the client) watching and waiting for an adult to join the Teams call.

But agencies can’t stop the hamster wheel.

3. They don’t hire domain people.

Healthcare, retail, finance, travel products built by people who’ve only ever worked in… other digital agencies.

No one who’s carried regulatory risk.

No one who’s owned a margin.

No one who’s ever had to live with the consequences of a bad decision.

Just glossy decks suspended in the pixel ether of Google Workspace.

4. They don’t hire people who’ve actually used the platforms they’re building on.

This is unforgivable.

Imagine building for a category you don’t truly understand, on a platform you’ve never actually used, for users you cant personally identify with.

It’s insane.

Sales pitches delivered by people who’ve never logged in.

Design constrained by guesswork and hope.

5. They assume the client is an idiot.

Despite knowing less about the domain, the users, and often the technology, agencies default to condescension. Feels more like toxic narcissism than professional services.

⸻

So what did I miss?

Nothing material.

This is the operating model.

And, it survived because it was too large, too normalised, too unexamined to fail.

But not now.

With AI-native tooling, a competent minority of experienced consultants can confidently predict 50–75% of requirements for the majority of digital builds.

Because patterns do exist.

The remaining 25% is nuance. The custom part.

Which makes the old rituals look what they really are: discovery cruises around the exotic islands of ego and whimsy. Months spent rediscovering what should have shipped out of the box.

The antidote has been hiding in plain sight for as long as I can remember. Like Occam’s Razor, its also the obvious one:

Build first then customise what genuinely needs customising.

Not because discovery is useless. But because pretending everything is unknowable is a business model that no longer survives contact with reality.

That gig is over.

The work that remains is not discovery theatre. It’s judgement.

The confidence to say: I already understand most of what you need.

And the humility to design carefully for the part that’s genuinely new.

Build first.

Customise for nuance.

Everything else was just a performance. An act of distraction to sustain agencies’ T&M model.