The AI tool stack inside marketing teams just got a lot more crowded, and ChatGPT’s lead has all but disappeared. Basis’s fourth annual AI and the Future of Marketing report, released October 19 and based on 184 marketing and advertising professionals surveyed from July 16 to August 14, 2026, shows Claude climbing from 17.9% to 72.3% adoption while ChatGPT slipped from 88.6% to 74.5%.
That leaves a 2.2-point gap between the two tools, down from a 70-point-plus separation a year ago. For teams that standardized on one model, that moment is over.
The tool race is now a near tie
ChatGPT still leads, but barely. Among the 163 organizations paying for premium AI tools, Claude Pro is the most-used paid product at 33.1%, ahead of Microsoft Copilot Premium at 29.4% and ChatGPT Pro at 25.2%. Basis estimates premium Claude tiers reach nearly two-thirds of paying organizations and ChatGPT’s paid tiers reach 59.6%.
Below the top four, Adobe Firefly climbed to 20.1%, Midjourney to 17.9% and Perplexity to 13.0%. The stack is also getting more bespoke: 43.5% of organizations now run a custom or semi-custom model, up from 28.5%.
AI use is daily and increasingly agentic
Adoption is nearly total: 98.4% of respondents use generative or agentic AI at work, and 61.4% use it daily, up from 33.6% last year. Research at 85.9% and ideation at 83.2% remain the top tasks, but the faster growth is further down the workflow: data optimization jumped to 53.8%, and image and video production rose to 41.8%.
Half of organizations now run AI agents, up from 20%. Reporting and analytics is the most delegated task at 72.3% among agent users, followed by campaign brief creation at 56.4%. Media buying remains the least delegated area at 18.1%, which suggests teams are still cautious about letting AI touch budget directly.
Measurement is the glaring gap
The uncomfortable number is 6.5%: only that share of organizations measure AI ROI with defined metrics and benchmarks. Another 23.4% measure it anecdotally, and the rest either don’t measure, plan to later, or aren’t sure.
Basis connects the gap to weak data foundations. Just 13.0% call their first-party data extensive and well structured. When AI’s role shifts from drafting to decision support, agents running on fragmented data will underperform and budgets become hard to defend.
What to do before you add another tool
- Audit every paid AI seat. Map each subscription to a workflow and kill anything without a clear owner.
- Set one benchmark per use case, such as hours saved, output volume, conversion or cost per asset, before you add agents.
- Review AI ROI monthly, with the same discipline you apply to media spend.
The report’s own warning is blunt: AI budgets are growing faster than almost any other marketing line item, but they are currently the least defensible. Teams that fix measurement now will be the ones positioned to scale agentic AI with confidence later.
Source: PPC Land
