Your AI productivity gains are about to become worthless

Was this newsletter forwarded to you? Sign up to get it in your inbox.

Jules Love's Shift: AI for Agencies, published last week, makes a stark prediction: by 2027, AI-enhanced capabilities will be table stakes. Agencies treating AI as a productivity boost are missing the point. The baseline is shifting beneath them.

According to Spark's latest data, regular AI users amongst agencies jumped from 20% to 35% in just six months, yet only 5% have moved beyond experimentation into what Love calls "AI maturity": building new workflows, hiring for AI skills and developing proprietary IP.

Love identifies three adoption stages: augmentation (AI enhances existing work), automation (systematic scale creates new outcomes), and innovation (new capabilities from unique data or IP).

In our conversation, Love noted most agencies remain stuck in augmentation, focused on productivity gains as the industry baseline shifts around them.

"They're thinking, 'if I can make this faster, I make a little extra margin,'" he says. "But a year or two from now when everybody's working this way, you're just going to be handing that margin back to your client. There's going to be a lot of pricing pressure on these things."

For marketers, the point of differentiation is no longer "are you using AI?" but "what can you do now that you couldn't do before?"

AI tools are already becoming commodities

When clients possess the same tools, agencies can't charge for the capability itself. Value shifts to what the agency does with it.

"Most people are finding that they can do that in a day now," Love remarks. "In fact, your client may even hand you the information because they've done it already themselves."

In Shift, he points to RWS, a media localisation service whose share price declined dramatically as LLMs emerged. The company competed on a capability that language models could replicate instantly.

He contrasts that with Duolingo, whose share price rose 20% in the same period by using identical technology to create AI conversation practice with native speakers. The company identified what only this technology could uniquely deliver. One company's commodity became another's differentiation.

This dynamic is now playing out across agency research, planning and concepting.

Kate Ross, CEO of eight&four, told Love: "We pretty much hand all the AI productivity savings over to the client. Otherwise, what's the point? Might as well do it the traditional way. We're not making extra profit from it. It's about staying relevant and competitive in the market."

Ross's agency built Platform12 rather than surrender margin, exemplifying Love's warning about short-term thinking. Once everyone can work faster through basic AI training, only two value propositions survive: industrial scale that enables measurably different outcomes, or proprietary capabilities competitors can't replicate.

From volume to outcomes

Agencies competing on "we can deliver 1,000 variations instead of 10" face immediate pricing pressure, unless those variations generate measurably better outcomes.

Digital marketing business Jellyfish demonstrates the difference. The agency redesigned its global creative teams around Pencil Pro, which generates channel-ready ads from brand objectives and assets, then uses performance data to optimise future output.

The results: over 1 million creative executions in 12 months, a platform trained on £1bn+ in ad spend, and a shift from selling time to selling outcomes.

Ten variations produce educated guesses. A thousand variations tested continuously enable precision targeting, real-time optimisation, and ROI measurement with statistical confidence.

In Shift, Love warns that "services that can be automated will become commoditised over time, creating severe downward pressure on pricing."

Automation invites commoditisation when agencies compete purely on velocity. Systematic automation combining industrial speed with provable performance outcomes creates differentiation on delivery.

As Spark writes in its report: "While you're experimenting, competitors are systematising. This gap between access and strategic capability is where market leaders separate from the rest."

In the near future, marketers may keep one partner for volume outcomes, one for strategic capabilities, and cut agencies offering neither.

Proprietary data creates moats

Innovation, in Love's terms, refers to fundamentally new capability created by AI. This means unique data competitors can't access, methodologies that took years to develop, or client relationships that create switching costs.

Brave Bison's AudienceGPT achieves 90% behavioural fidelity. Its AI personas mirror real consumer responses because it's trained on proprietary research methodologies. The tool demonstrates how accumulated advantage, rather than access to the model, creates the moat.

Traditional research happens in discrete testing moments, but synthetic tools enable continuous validation. OLIVER's Seance demonstrates 85% correlation with traditional focus groups. The value isn't cost reduction but capability transformation.

The accuracy came from years of OLIVER's internal workshop transcripts: client sessions, strategy development, creative testing. Data accumulated through their embedded agency model that competitors can't easily access.

Eight&Four's Platform12 integrates multiple tools (Coach for brand imagery, Scholar for copy, Loop for performance, Echo for social listening) tuned to specific sectors. This enabled Genesis Motor Europe to visualise vehicles in any environment without location shoots. The sector tuning creates switching costs.

Compare these to agencies claiming proprietary AI tools that are really customised ChatGPT interfaces.

Agencies building defensibility answer with specificity: What customer problem are you solving that AI uniquely enables? What data or methodology do you possess that competitors can't easily acquire? What switching costs make your solution sticky once implemented?

What marketers should actually audit

For marketers evaluating agencies, the questions have shifted. Start with: "What have you built that I can't get elsewhere?" and "Can you prove AI makes my business outcomes better, not just your delivery faster?"

Look for custom tools with proprietary data. Look for capabilities requiring unique inputs competitors can't easily access, not sophisticated prompts anyone could replicate. Look for sector-specific AI systems.

Look for case studies showing conversion uplift, engagement increases, revenue growth, rather than efficiency metrics alone.

Generic ChatGPT usage without custom workflows, or pitches focused on "we can deliver faster or cheaper" should be red flags.

Process indicators matter as much as product. Love recommends treating AI adoption with project discipline: "We try to encourage the teams we work with to treat your AI project like you would a client project. Put it in the same project management system, have a budget associated with it where people can charge time to it."

He suggests a practical starting point: "two hours every two weeks where each team (client services, creative, copywriting) carves out time to work together, not individually, but together on a shared problem."

The window closes faster than agencies think

Within six months, positioning should be visible through training investment and governance frameworks. By mid-2026, proprietary capabilities should be emerging with measurable outcomes.

By early 2027, agencies should have launched new services demonstrating their chosen path: automation or innovation, volume or strategic differentiation.

Agencies that miss this window won't collapse. They'll compete purely on price for work clients increasingly believe they could do themselves.

"Agencies focusing on AI-driven better work and strategic innovation will outperform those chasing only speed and cost cuts," says Love.

For marketers, it's not about auditing agencies on tool usage. It's about strategic commitment visible through operational investment, proprietary capability development, and pricing models that reflect differentiation. If you can't see proof of their positioning today, they haven't chosen one.

The agencies being built now (the ones launching with automation or innovation baked into their business models) won't be competing against agencies stuck in augmentation. They'll be competing for the work those agencies used to do.

Looking for a new agency? DCA cuts through the noise to connect brands with agencies that fit

• We cover all types of projects and disciplines, from strategy and creative to performance and tech.

• Our editorial team vets every agency, so you skip the pitch theatre and get straight to capable partners.

• We offer consultancy, pitch management, or a complimentary shortlisting service if that's all you need. Submit a brief today

Natasha Randhawa, newsletter editor - Department of Creative Affairs.