AI Hit Ad Agencies Hard in 2025 - But a Rebound May Be Loading
AI shook the agency model this year. Clients built in-house tools, stocks slid, and fear of margin compression spread fast. Still, this looks like a cycle, not an extinction event. The teams that pair human judgment with smart automation will come out ahead.
The AI-Driven Sell-Off: What Actually Happened
WPP took the brunt of it. Profit warnings in July and October flagged a 3-6% revenue drop, softer pipelines, and new-business pitches falling to roughly one-third of prior levels. Shares hit multi-decade lows and the company was dropped from the FTSE 100 after a 62% slide, prompting leadership changes and heavier AI bets.
Peers felt it too. Omnicom and IPG dipped on similar worries, while Publicis leaned on its CoreAI platform to hold up better. The biggest headline: Omnicom's $13B acquisition of IPG closed in November 2025, with a plan to squeeze out $750M in synergies. By mid-year, Barclays cut ratings across the group, and valuations fell to 2020-like levels.
Client Behavior Turned the Knife
Brands moved routine production inside. Coca-Cola's 2025 holiday campaign was fully AI-generated for the second year, produced with roughly 20 staff instead of 50. It saved money, but consumers called out "uncanny" visuals and a lack of soul. The message: efficiency wins headlines, but taste still matters.
Google's Automation: Useful, Not a Full Replacement
Google rolled out upgrades to Performance Max and launched AI Max for Search in May 2025. The stack automates bidding, creative assembly, targeting, and copy, with claimed gains of 14-27% in conversions. Helpful for in-house teams, yes-but it still struggles with the hardest job: budget allocation across channels with real-world constraints.
That's where experience earns its keep. Agencies are recasting themselves as integrators. WPP is pushing its WPP Open platform to stitch tools together and close the gap between execution and strategy.
Why the Pain May Be Short-Lived
Lower production costs can spark more creative output and bigger tests. If everyone can make more, the new edge is taste, timing, and smart budget moves. Agencies still hold unique data, cross-channel experience, and media muscle memory that's hard to replicate.
Tastes are shifting too: quiet luxury, minimalism, influencer collabs, and resale have more pull than logo-heavy ads. Add ethical AI and regulatory pressure, and the need for expert guidance grows. As hype cools, human insight-amplified by AI-looks set to win.
What Marketers Should Do Next
- Audit your media mix: where AI excels, where it breaks, and what stays human-led.
- Rebase KPIs by channel and funnel stage; add confidence intervals to all lift claims.
- Run controlled tests: holdouts, geo splits, MMM + incrementality for triangulation.
- Stand up a "creative OS": prompts, brand rules, reference libraries, and approval flows.
- Split production: low-stakes assets in-house; high-stakes ideas with your agency bench.
- Make your agency an integrator: one orchestration layer across Google, Meta, retail media, CTV, and search.
- Write data contracts: who owns prompts, models, outputs, and training rights.
- Budget rules > opinions: pre-set reallocation guards based on CAC/LTV and marginal ROAS.
- Add a taste council: creators, strategists, and legal to screen AI work for quality and risk.
- Use a red team for AI: bias, consent, copyright, and model drift checks.
A Simple 30/60/90 Plan
- Day 0-30: Tool and data audit, pause noncritical experiments, define creative and legal guardrails, shortlist 3-5 AI use cases.
- Day 31-60: Launch pilots with clear KPIs across search, social, and CTV; pair in-house work with one agency-led integration test.
- Day 61-90: Scale winners, cut laggards, renegotiate SOWs for integration and measurement, codify your playbook.
The Bottom Line
2025 stung. But this industry isn't getting removed-it's being rewired. If you combine human judgment with smart systems, you'll buy better, create smarter, and waste less.
If you're upskilling your team, consider the AI Learning Path for Business Unit Managers.