In late 2025, we surveyed 180 agencies to understand how AI was reshaping their business. The picture that emerged was one of cautious optimism: agencies weren’t collapsing, but they weren’t doubling profits overnight either. They were experimenting, streamlining, and waiting to see where things would land.

So we went back.

In the six months after our initial survey, the technology moved fast. New frontier models launched, AI coding agents went mainstream, and inference costs dropped sharply. So we ran another survey across the same five areas — price cuts, service pricing, revenue impact, staffing, and new revenue streams — to see what changed on the ground.
The results show a market still in motion. More agencies are seeing benefits, but just as many are struggling to measure them.

Here’s what changed six months after our initial survey:

While a lot changed on the technology side, the agency business model didn’t follow at the same pace. As in our first report, the data tells a story not of a tectonic shift, but of incremental change and cautious experimentation.

There’s no pricing collapse, no mass layoffs, and no AI-powered revenue boom. But things are moving — and some of the patterns we spotted six months ago are now even more visible.

Short answer: No. 

Six months ago, we found that most agencies anticipated being asked for AI-related discounts, but weren’t experiencing it at scale. That picture is essentially unchanged. Clients haven’t increased the rate of asking for discounts. If anything, the “not yet but we expect it” group grew slightly, suggesting the feared wave of discount requests still hasn’t arrived.

What has changed is the nature of the conversation. AI is creating new dynamics in client relationships that go beyond simple discount requests.

While some clients see AI as a reason to leave, others see it as a reason to lean in. This polarity is more pronounced now than it was six months ago, and it signals a broader renegotiation of what agencies are for.

Nope. Still no consensus.

Agencies are split four ways on how AI is affecting their pricing: some are keeping or increasing prices, some are lowering prices to compete, some see little impact, and a growing share are still figuring out the right model. That’s practically the same picture we saw six months ago. If anything, the “still figuring it out” group grew slightly, while fewer say it has little impact at all.

The lack of movement is itself a finding. In a period where AI capabilities advanced significantly, agencies didn’t converge on a pricing response. They’re still experimenting.

Value-based billing continues to look like the preferred direction for agencies that are adjusting. The logic is intuitive: if AI helps you ship faster, charging for time makes less sense. Charging for outcomes makes more.

Some agencies are also getting practical about AI costs showing up on the books.

Yes, but it’s getting harder to prove it.

Most agencies are seeing a positive impact of AI on revenue or profits. That hasn’t changed in six months, but it’s interesting what has changed. There’s been a slight increase in agencies saying they can’t estimate AI’s impact on their revenue, and a decrease in those actively working on it.

What this suggests is that more and more agencies are using AI, but a growing number can’t tell you what it’s actually doing for their bottom line. They know it’s helping (things feel faster, lighter, more efficient, etc.), but the measurable revenue impact remains fuzzy.

Those who can point to results are focused on the same areas: streamlining sales processes (background research, proposal writing), streamlining internal processes (faster onboarding, admin work), streamlining deliverable creation (prototypes, content, strategies), and cutting third-party expenses.

Two groups are diverging, and the gap is getting steeper.

There’s an interesting correlation between AI revenue impact and pricing confidence. Agencies seeing real business results from AI are more likely to keep or increase prices. Those still working on implementation are disproportionately more likely to be stuck figuring out their model.

Six months later, that gap has widened.

Among agencies reporting multiple positive AI situations, 52% are now keeping or increasing prices with improved margins (up from 47%). Among those still working on implementation, 61% are still figuring out the right model (up from 47%).

The agencies that figured out how to use AI for real business impact are getting bolder with their pricing. The ones still trying to get there are increasingly stuck. This isn’t surprising, but it is important. It suggests that the window for “we’re still experimenting” is narrowing. The early movers aren’t waiting around.

Not really. But it’s changing who gets hired.

The vast majority of agencies have no plans to reduce staff because of AI. Only 3% report significant reductions. That picture is essentially unchanged from six months ago, and the mass-layoff narrative remains unsupported by the data.

The open-ended responses provide more context, though. Agencies aren’t laying people off, but they are hiring differently. AI is doing the work that a new junior hire or a contractor would have done. The headcount doesn’t shrink; it just doesn’t grow the way it would have.

Pretty much.

Nearly every agency is exploring AI-related revenue streams. A third have launched something, though for most the impact is still small. The rest are still in exploration mode — and only a handful aren’t looking at all. That picture hasn’t changed much in six months: a few more agencies moved from “exploring” to “doing something,” but the something isn’t transformative yet.

The exceptions stand out, though. A few agencies that have launched AI-related services are seeing real demand.

And some are finding new work from the other direction — clients who tried to go it alone with AI and realized they still needed agency expertise.

Updating task statuses based on comments and activity, surfacing action points, sending reminders for time logging and task completion, summarizing project health, generating project or budget proposals from documents or rules.

Six months is both a long time and no time in the AI era. In that window, we got new models, new tools, new capabilities, and an entire new category of AI agents entering the mainstream. On the agency side? The needle moved, but not by much.

There is no doubt that AI is changing the agency business, but it’s changing it slowly, unevenly, and in ways that are harder to measure than the hype suggests. The clearest takeaway is the growing divide. Agencies that have found real value in AI are pulling ahead — more confident in their pricing, more settled in their operations. Those still experimenting aren’t falling apart, but they’re running out of time to figure it out while the leaders keep moving.

This is the second wave of Productive’s ongoing research into AI’s impact on agencies, following the initial survey conducted in September 2025.

Survey: 174 agencies and consultancies who use Productive, surveyed March 17–31, 2026. 12 questions: seven on AI’s business impact, five demographic. The final open-ended question drew 160 written responses.

Respondents: Majority from Europe (54%), followed by Asia-Pacific (21%) and the United States (13%). Most were executive leaders (61%). The sample skews toward digitally mature, process-oriented agencies, typically in the 20–50 employee range.

Analysis: Quantitative data analyzed using descriptive and inferential statistical procedures. All highlighted differences were statistically significant (p < .05). Open-ended responses were thematically analyzed, with selected quotes (anonymized) included throughout for context.