Execution stopped being the constraint. Most agencies haven’t noticed.

Kevin Indig published a piece this afternoon that he described as an existential moment. He's right that something shifted, and he's right that most of the industry hasn't caught up to it. But the framing he reached for — "am I still an advisor, or a builder?" — undersells what's actually happened.

The cost of execution didn't just fall. It collapsed into the cost of judgment. And almost every SEO agency, in-house team, and consultancy I look at is still organised around the old constraint.

That's the part worth sitting with. Not the productivity numbers — the org chart implications.

What Indig actually said

The piece itself is worth reading. Indig describes shipping landing pages end-to-end for a major travel brand, building automation apps for keyword research and SEO testing, and running internal agents for research and chart-making. He cites METR's research showing 1.5x to 13x time savings when technical staff use Claude Code. Bain's estimate of 30–50% productivity gains from agentic AI. Anthropic going from $9bn to $30bn in six months.

His core point: when execution collapses in cost, judgment becomes the only thing that doesn't compress. Meanwhile, "most operators are still hiring, budgeting, and measuring as if execution is the constraint."

That last sentence is the whole piece. Everything else is supporting evidence.

The SEO industry built itself around a constraint that no longer exists

Walk into any mid-sized SEO agency and look at the cost structure. You'll find content writers, technical SEOs implementing schema, account managers running reports, juniors pulling data into decks, paid media specialists building campaigns, and somewhere near the top, one or two people who actually decide what the work should be.

The constraint is no longer execution. It's knowing what to execute, why, and whether the output is any good.

That structure exists because executing SEO work is expensive and slow. Writing a 2,000-word piece took a content writer a day. Auditing a site took a senior SEO three days. Building a Looker Studio dashboard took half a week. The judgment layer — what to write about, which audit findings actually matter, which metrics are worth tracking — was a small fraction of total billable hours because the doing was so labour-intensive.

That equation has inverted. A consultant with the right setup can now produce a defensible content brief, draft, schema implementation plan, internal linking map, and reporting dashboard in the time it used to take to write the brief alone.

The constraint is no longer execution. It's knowing what to execute, why, and whether the output is any good.

The agencies that haven't internalised this are essentially selling expensive labour against a market where the labour cost has collapsed. They'll feel it as fee compression first, then as clients quietly bringing work in-house, then as the realisation that their juniors were doing work an agentic workflow can do faster and more consistently.

What "judgment doesn't compress" actually means

Indig's phrasing is good but slightly abstract. Let me make it concrete.

Judgment in SEO work is things like: which of these 47 audit findings will actually move the needle for this specific business. Whether this client's brand is strong enough to compete for a head term or whether they need to win on long-tail before they earn the right to fight for it. Whether a piece of content is genuinely better than what's ranking, or just a polished restatement. Whether a structured data implementation is helping AI systems understand the page or adding noise. Whether a measurement framework is tracking causation or just correlation that will collapse next quarter.

None of that compresses with AI tooling. If anything, the cost of bad judgment goes up, because AI lets bad judgment ship faster and at scale. Ten thousand words of mediocre content used to take weeks to produce. Now it takes an afternoon. The cost of being wrong has collapsed alongside the cost of being right, which means the value of being right has gone up, not down.

This is why I keep telling people the slop tier isn't a content problem. It's a judgment problem. The output is bad because nobody with judgment was anywhere near the work.

The advisor doesn't get replaced. The advisor's leverage just went up by an order of magnitude.

The measurement piece that landed alongside it

It's not coincidence that Google previewed Meridian GeoX, Meridian Studio, and Data Manager updates this morning, ahead of Google Marketing Live on the 20th. The measurement story Google is now telling is explicitly about causal inference — geographic incrementality experiments, marketing mix modelling, first-party data unification.

Read in isolation, that's a Google Ads announcement. Read alongside Indig's piece, it's the same shift surfacing somewhere else.

When execution is cheap, the question stops being "did we ship the campaign" and becomes "did the campaign cause the outcome." Google is positioning Meridian GeoX as the answer to that question for paid media. The same question is unsolved on the organic side, where attribution to AI citations, dark social, brand search lift, and assistant-mediated discovery is still mostly vibes.

The agencies that figure out the organic version of causal measurement first — not "rankings went up" but "we caused this revenue" — will own the next five years. The ones still selling activity reports won't.

Where this gets uncomfortable

The honest implication of Indig's piece, and the one he didn't quite say, is that a lot of the SEO industry's headcount is exposed.

Not the strategists. Not the senior consultants who actually understand what's worth doing and why. Not the genuine technical specialists who can debug a JavaScript rendering issue at 2am. Those people get more valuable, not less.

The exposed roles are the ones built around executing pre-defined tasks. Junior writers producing briefs-to-spec. Mid-level technical SEOs running standardised audits. Account managers assembling monthly reports from templated data sources. The work isn't disappearing — it's being absorbed into agentic workflows that one senior person can supervise.

That's an uncomfortable thing for the industry to say out loud, which is why almost nobody is saying it. The agencies talking publicly about AI are talking about it as a tool that makes their existing teams more productive. That's true at the margins, but it's not the actual shift. The actual shift is that the team you need to do the same work just got smaller, and the team you need to do *better* work just got differently shaped.

What Indig's piece doesn't quite cover

Two limits worth being honest about.

First, the productivity numbers are real but they're concentrated in technical work. METR's 1.5x–13x range is for engineering tasks. The gains in pure writing or pure judgment work are real but smaller. Anyone selling you "10x your content team" is doing the thing where you take the highest number from one study and apply it to everything.

Second, the cost collapse is uneven across business types. If you're a service business that wins on local trust, brand reputation, and word of mouth, the collapse in execution cost matters less to your discovery than it does for a SaaS company chasing organic traffic. The implications scale with how much of your growth depends on produced content versus produced relationships. Most of my clients are closer to the first than the second, and for them, the bigger story is still brand and distribution, not execution efficiency.

Both limits matter. Neither changes the core argument.

What this means for how you should be organising

If you run an in-house team or you're commissioning agency work, the questions to ask are different now.

Stop asking how many pieces of content the agency will produce per month. That's an execution metric, and execution is no longer the constraint. Start asking what their judgment process looks like. Who decides what gets made. How they decide whether something is good enough to ship. How they measure whether the work caused an outcome.

Stop hiring for execution capacity. Start hiring for judgment, taste, and the ability to supervise agentic workflows. The shape of the team you need has changed. A senior consultant with a well-built tooling stack can now do work that used to require a team of four. The cost-to-value ratio of senior expertise versus junior labour has flipped, and a lot of agency pricing models haven't adjusted yet.

Stop benchmarking yourself on activity. Reports showing how many things were done are reports about a constraint that's gone. The new benchmark is causal contribution to outcomes you actually care about — revenue, qualified leads, branded search lift, citation share in the assistants your customers use. If your current reporting can't tell you whether the work caused the result, your reporting is measuring the wrong thing.

The close

Indig framed his piece as an existential moment, and he was being personal about it. But the existential question isn't really "am I an advisor or a builder." It's whether the entire industry is still organised around a problem that's been solved.

Execution was the constraint. It isn't anymore.

That's the loop. And we built it.

The work that survives this isn't the work that produces the most output per hour. It's the work that produces the right output, justified by reasoning the client can trust, measured by causation rather than correlation. Most of the industry is still selling the first thing. The market is going to start paying for the second.