سیناپس نگار

آخرین اخبار هوش مصنوعی، تحلیل‌ها و ابزارهای AI

ChatGPT drops to 74.5% of marketers as Claude climbs to 72.3%, Basis finds

PPC Land

Source published: Site published:

Original article · فارسی

Marketing and advertising professionals are now using several AI tools instead of relying on just one. ChatGPT is no longer far ahead of its competitors. Basis, a Chicago-based media software company, published the fourth annual edition of its "AI and the Future of Marketing" report on October 19, 2026. The report found that 74.5% of 184 surveyed professionals use ChatGPT at work, down from 88.6% a year earlier. Anthropic's Claude grew from 17.9% to 72.3%. Daily AI use nearly doubled to 61.4%. Meanwhile, 51.1% of organizations now use AI agents, but only 6.5% measure the returns on their AI spending against clear benchmarks.

In Short

Basis surveyed 184 marketing and advertising professionals about their use of AI. Almost all use it, and most use it every day. Their choice of tools has changed significantly in one year, with Claude nearly catching up to ChatGPT. Many companies also let AI agents handle entire tasks, such as reporting and writing campaign briefs. However, very few companies check whether their AI spending produces measurable results. They are spending more without clear proof that the investment is worthwhile.

What Basis published and when

Basis announced the findings from Chicago in a news post titled "Daily AI Use Nearly Doubles Among Marketers as Measurement Lags" and published the full report. The survey included 184 marketing and advertising professionals working for agencies, brands and publishers. According to the report, it ran from July 16 to August 14, 2026. The news post describes the survey period more generally as "in August 2026."

This is the fourth edition of the report. Basis has asked mostly the same questions each year since 2023, making it possible to compare changes over time. The 2023 survey included 202 people, the 2024 survey had 141, the 2025 survey had 140 and the 2026 survey had 184. Some questions were introduced later, so not every measure covers all four years.

The sample size matters when interpreting the results. A sample of 184 means each person represents about 0.54 percentage points, based on a PPC Land calculation. Some important findings come from smaller groups. Questions about leadership included 63 to 65 people at vice-president level or higher, meaning one person's answer changes a result by about 1.6 points. The breakdown of AI agent use includes the 94 respondents whose organizations have built or adopted agents. Basis does not publish a margin of error or explain how it recruited participants.

Basis also has a commercial interest in these findings. The company has sold media software since 2001. In the report, it describes its product as "the only unified media platform to consolidate planning, digital operations, reporting, and financial reconciliation across programmatic, publisher-direct, search, and social." The report argues that measuring results is "close to impossible when planning, execution, reporting, and reconciliation live in separate systems." This argument closely matches the products Basis sells.

Adoption is now close to total

The report found that 98.4% of respondents use generative AI or agentic AI in their marketing or advertising work. A total of 93.5% use AI at least once a week, and just 1.6% do not use it at all. The share of non-users was 5.0% in 2025 and 27.7% in 2023.

The biggest change is how often people use AI. Daily use reached 61.4%, up from 33.6% in 2025 and 10.9% in 2023. A PPC Land calculation based on Basis's figures shows that the share of daily users has grown more than five times in three years.

Research remains the most common use of AI, at 85.9%, followed by generating ideas at 83.2%. Drafting content and creative work accounts for 67.4%, while making processes more efficient accounts for 60.9%. AI use is also growing in later stages of campaigns. Data optimization rose from 35.0% to 53.8%, and image and video production increased from 25.0% to 41.8%. Media planning stands at 39.1%, SEO and keyword work at 39.7%, reusing content at 38.0%, media buying strategy at 30.4%, and creating different ad versions or personalized ads at 29.9%.

"Taken together, the movement suggests AI's role is expanding from helping marketers begin their work to helping them complete it," the report states.

The tool market fragments

A year ago, ChatGPT led other tools by more than 40 percentage points. In the 2025 survey, 88.6% of respondents used ChatGPT, compared with 45.0% for Gemini, 41.4% for Microsoft Copilot and 17.9% for Claude. In 2026, ChatGPT stood at 74.5%, Claude at 72.3%, Copilot at 55.4% and Gemini at 47.8%. The difference between the top two tools had shrunk to just 2.2 percentage points.

Among other tools, Adobe Firefly reached 20.1%, up from 12.1%, while Midjourney rose from 12.1% to 17.9%. Perplexity reached 13.0%, Grok 5.4% and Jasper 3.3%. Salesforce's Agentforce fell from 5.0% to 1.6%, and Stable Diffusion stood at 2.2%.

The survey shows a trend similar to changes in AI website traffic during 2026, although the numbers measure different things. Similarweb data showed ChatGPT's share of generative AI web traffic falling to 52.7%, while Claude's share rose to 8.9% by late May. Comscore's second-quarter data put ChatGPT's share of US AI prompts at 50% in June, down from 70% in January, while Claude reached 11%. These panels measure consumer traffic. Basis measures whether a professional or their organization uses a tool at all. Someone who uses five tools counts as a user of all five. This explains why Claude can be used by 72.3% of marketers in the Basis survey but have a single-digit share of web traffic in another dataset.

Paid subscriptions

The share of organizations paying for premium AI tools rose to 88.6%, up from 65.7% in 2025 and 44.0% in 2024. Among the 163 organizations that pay, Claude Pro was the most-used paid product at 33.1%. Microsoft Copilot Premium followed at 29.4%, ChatGPT Pro at 25.2%, Claude Enterprise at 22.1% and ChatGPT Plus at 21.5%. Midjourney reached 19.0%, Adobe Firefly 16.6% and Gemini Pro 14.7%. ChatGPT Enterprise reached 12.9%, and Claude Max reached 9.2%.

Basis combines subscription levels by provider. It says premium versions of Claude are used by nearly two-thirds of paying organizations and that ChatGPT's paid plans reach 59.6%. However, these totals come from adding individual plan percentages. A PPC Land calculation adds Claude Pro at 33.1%, Claude Enterprise at 22.1% and Claude Max at 9.2%, giving 64.4%. Adding ChatGPT Pro at 25.2%, ChatGPT Plus at 21.5% and ChatGPT Enterprise at 12.9% gives 59.6%. An organization paying for two plans from the same provider may be counted twice, so these sums are upper limits rather than counts of unique organizations. Paid Gemini plans reach 24.6% using the same method.

The changes from 2025 are substantial. ChatGPT Plus was then the most-used paid product at 38.0%, followed by ChatGPT Pro at 31.5% and Claude Pro at 15.2%. Copilot changed little, rising from 28.3% to 29.4%.

Custom models

The share of organizations using a custom or partly customized AI model rose from 28.5% to 43.5%. Most of the increase came from partly customized models built on an existing large language model (LLM). Their share rose from 17.1% to 31.5%. Fully custom models increased only slightly, from 11.4% to 12.0%. Another 19.0% plan to introduce custom solutions within 12 months. A further 33.5% use no custom models and have no near-term plans to do so, down from 47.1%.

The measurement gap

The report's main concern is not whether organizations use AI, but whether they can prove it works. According to Basis, just 6.5% of organizations measure the return on investment (ROI) from AI using defined measures and benchmarks. Another 23.4% measure it, but only through informal observations. The remaining 65.1% fall into three groups: 18.5% do not measure ROI but plan to, 22.7% do not measure it, and 23.9% do not know whether anyone in their organization measures it.

In a survey of 184 people, 6.5% represents about 12 respondents, according to a PPC Land calculation.

"For an industry that purports to hold every media dollar to account, the absence of a comparable standard for AI is unlikely to hold much longer, particularly as those investments continue to swell," the report states.

Spending continues to grow. Three-quarters of respondents, or 75.0%, said their organization invested in technology to automate or simplify processes in the previous 12 months, up from 63.6%. Plans for the next 12 months remain almost unchanged: 78.8% intend to invest, compared with 79.3% a year earlier.

Other studies have found similar problems. TransUnion surveyed 100 senior US marketers and found that only 53% reported meaningful returns from AI-powered marketing, while 65% mainly measured AI's impact through estimated savings in time or costs. A study by WSJ Intelligence and Code and Theory involving 801 executives found that one in four large US companies had no one responsible for coordinating AI. Last year, MiQ surveyed 3,169 marketers and found that 72% planned to use AI more, while only 45% felt ready to do so.

Self-reported gains

Because organizations lack clear benchmarks, the report's efficiency results are based on what respondents believe. Three-quarters, or 75.0%, said AI had made them moderately or significantly more efficient over the previous year. The share reporting major efficiency gains rose from 13.5% in 2024 to 25.7% in 2025 and 33.2% in 2026. Only 4.3% saw no efficiency gains, down from 19.9% in 2024.

Quality results are different. A total of 60.9% said AI had made them moderately or significantly better at their jobs, while 16.8% reported no measurable effect, almost unchanged from the previous year. The results also differ by seniority. Among vice-presidents and more senior leaders, 66.2% said AI had made them moderately or significantly better at their jobs. Among entry-level to mid-level employees, 63.9% said it had made them only slightly or moderately better, and 21.0% reported no measurable effect.

Basis suggests this difference may be because senior staff use AI for "synthesis, analysis, and decision support," while junior staff use it "to produce and process." AI may help junior employees work faster without necessarily improving the quality of their work.

Data foundations lag behind

The report connects the measurement problem to first-party data, meaning information an organization collects directly. Only 22.8% of respondents said this data was extensive or a basic foundation of their AI projects, barely changed from 21.4% a year earlier. Another 25.5% called it somewhat important, 18.5% said it played a small role, 14.1% said it played no role, and 19.0% were unsure.

Respondents also described problems with data quality. Only 13.0% said their organization's first-party data was extensive and well organized, down from 17.9%. Another 31.0% said it was sufficient but separated across teams, 26.6% called it limited and fragmented, 12.0% said they hardly used first-party data, and 17.4% were unsure.

The report links these findings to the growth of AI agents. "While half the industry now runs agents, just 13.0% of marketers describe their organization's first-party data as extensive and well-structured, most likely resulting in agents underperforming their potential," it states. This is Basis's interpretation. The survey did not directly measure how well agents perform.

Agents arrive in volume

The share of organizations that have created or started using AI agents for marketing or advertising rose from 20.0% to 51.1%. Basis describes this as an increase of 155.5%. The report defines agents as systems that "pull data, make decisions, and take action to meet a predefined objective, all with limited or no human involvement." This places them within the broader category of agentic AI.

Among the 94 respondents whose organizations use agents, reporting and analytics was the most common task, at 72.3%. Creating campaign briefs followed at 56.4%, then creating content at 45.7%, improving creative work at 38.3%, media planning at 35.1%, personalization at 25.5% and media buying at 18.1%.

Media buying, which comes closest to directly spending money, was the least common of the listed tasks assigned to agents. AppsFlyer data published last December showed that 57% of AI agent deployments in app marketing focused on technical automation, such as checking configurations.

Expectations for the future are ahead of current use. Among 180 respondents, 77.2% expected digital advertising to become at least partly automated or managed by agents within three to five years. Of these, 50.0% expected partial automation, 25.0% expected most of the work to be automated, and 2.2% expected full automation. Leaders were more confident than other staff. Among 63 leaders, 36.5% expected advertising to be mostly automated and 3.2% expected full automation. Among 117 entry-level to mid-level respondents, the figures were 18.8% and 1.7%.

Views have also changed about which part of marketing AI will affect most. In 2024, 57.4% said content creation was the area most likely to change. In 2026, that figure fell to 23.4%, while analysis became the leading answer at 27.2%, up from 11.3% two years earlier. Creative optimization followed at 15.8%, media planning at 14.7%, and 6.5% named team size.

Jobs: tasks move faster than headcount

The replacement of human tasks with AI nearly doubled. Basis found that 47.8% of respondents said their organization had replaced human tasks with AI solutions in the previous 12 months. The figure was 27.9% in 2025, 21.3% in 2024 and 10.4% in 2023.

Replacing employees showed a different pattern. Among leaders, 20.6% said their organization had used AI to replace employees or planned to do so. This was down from 25.4% in 2025 but above 18.6% in 2024 and 11.5% in 2023. Since 63 leaders answered this question, the fall from 25.4% to 20.6% represents about three people, according to a PPC Land calculation.

Basis interprets the difference as organizations "putting the additional capacity toward more work, rather than fewer people." However, it adds that "it remains unclear whether that reflects a durable division of labor, or simply a lag." The issue remains important across the industry. Earlier this month, HubSpot cut about 660 jobs while reorganizing its product teams around customer outcomes, while saying AI-driven efficiency was not the reason for the decision.

Respondents were more concerned about job losses across the industry than about losing their own jobs. Just over half, or 50.6%, expected AI to cause major job losses in digital marketing within three to five years. That was down from 58.6% in 2025 but above 41.1% in 2024. At the same time, 70.0% saw little or no risk that AI would replace their own job during that period, up from 66.4%. The share who saw no risk at all rose from 17.1% to 25.0%.

Leaders reported that their teams had largely accepted AI. According to the report, 96.9% said their teams had embraced AI, up from 74.6% in 2024. A total of 28.6% said their teams had embraced it strongly, twice the 14.0% recorded two years earlier. No leader reported strong resistance.

Content quality, GEO and brand safety

Confidence in AI-generated content continues to fall. Basis found that 76.6% said AI-generated content did not match the work produced by people in their organizations. That was up from 75.0% in 2025, 70.2% in 2024 and 70.3% in 2023. The report links the change to people becoming more critical after four years of daily AI use and to "the proliferation of low-effort AI content across the open web." Earlier this year, an NP Digital survey of 565 marketers found that 47.1% encountered AI errors several times a week.

For online search, 53.4% of 178 respondents said their organization had started optimizing content specifically for AI search. The report calls this "generative engine optimization" or "answer engine optimization." The practice is often shortened to GEO. Another 24.7% had not started, while 21.9% were unsure.

All respondents in a group of 180 agreed that AI creates a brand safety and misinformation risk for digital marketers. The share describing the risk as moderate or significant rose from 88.6% to 92.2%. Half, or 50.0%, called the risk significant, up from 40.0%, while 42.2% called it moderate. Concerns about AI-generated advertising content also relate to spending. An analysis by TAG and ANA found that low-quality AI-generated content accounted for 1.3% to 2.4% of open-web programmatic advertising spending and received a premium-quality rating more than 70% of the time.

Energy and regulation

Concern about AI's environmental impact rose to 81.1%, up from 53.2% in 2024, according to Basis. Support for regulating AI development and use reached 83.7%, up from 70.0%. A further 6.0% opposed regulation and 10.3% were unsure. The report discusses this against a background of "federal regulators taking a deliberately hands-off approach toward AI." This reflects a US-focused view, although the report does not state which countries the respondents come from.

Sentiment: positive, and more skeptical

Respondents were more positive about AI but also more doubtful about its claims. According to the report, 79.3% believed AI would have a positive effect on marketing and advertising, up from 75.7%. Another 90.2% expected AI to radically change digital advertising within three to five years. Yet 53.3% said AI was overhyped, up from 40.7% in 2025 and close to the 53.2% recorded in 2024.

When asked about their personal attitude, 25.5% described themselves as excited, up from 16.4%, while 52.7% said they were cautiously optimistic, down from 59.7%. Leaders were more positive than other employees: 87.7% expected AI to have a positive effect on the industry, compared with 74.7% of respondents at director level or below.

Source discrepancies

Some numbers in the report conflict. The text says 22.5% of respondents consider first-party data somewhat useful but not yet integrated, while the chart on the same page gives 25.5% for the "somewhat" category. The text puts environmental concern in 2025 at 63.6%, but the chart shows 63.3%. The report's statement that concern rose by "nearly 18 percentage points" matches the chart's figure. The report also says that, "for the first time, a majority" consider the brand safety threat significant, but the chart shows exactly 50.0%, which is half, not a majority. Several charts label the junior group as "Director-level and above," even though the text and leadership charts describe it as director level and below. PPC Land uses the chart figures where the text and charts disagree.

What Basis says

"Marketers have moved well past the question of whether or not AI belongs in their workflow. Yet, more work is needed to justify the investment," said Katie McAdams, chief marketing and commercial officer at Basis, in the news post. She added that she believes "the next stage in our industry's AI evolution will come from unified media platforms that are building data foundations to show investment accountability, thus allowing organizations to scale agentic AI with confidence."

The report ends by explaining three problems caused by the measurement gap. The first is effectiveness: organizations "cannot separate a genuinely effective deployment from an expensive one that feels productive." The second is credibility: "AI budgets are growing faster than almost any other line item in marketing, and they are currently the least defensible ones." The third is the foundation needed for AI agents: "autonomy applied to an unmeasured process simply produces more of that process, faster."

Why this matters for the marketing community

For agencies and in-house marketing teams, the report captures an important change. AI has moved beyond writing draft copy to producing reports, creating campaign briefs and adjusting creative work with limited supervision. Half of the organizations surveyed now use AI agents. Spending on these tools has increased every year. Yet nearly two-thirds of organizations still cannot show whether AI improves results or simply makes work faster.

The findings about AI tools matter for another reason. Marketing teams that chose one provider two years ago now spread their work across three or four providers and pay for plans from several companies. This affects purchasing decisions, data management and the reliability of internal comparisons between tools. Measuring results also becomes harder when several AI models contribute to the same campaign output.

The figures have important limits. The overall sample is small, some groups are much smaller, the recruitment process is not disclosed, and every figure is based on respondents' own reports. The study also comes from a company that sells the integrated platform its conclusions recommend. Still, the four-year series from the same publisher provides a consistent view of how professionals describe their use of AI. Their use has grown faster than their ability to measure its results.

Timeline

Summary

Who: Basis, a Chicago-based media software company, surveyed 184 marketing and advertising professionals working at agencies, brands and publishers. Katie McAdams, the company's chief marketing and commercial officer, commented on the findings.

What: The fourth annual "AI and the Future of Marketing" report found that 61.4% of respondents use AI daily. ChatGPT use fell to 74.5%, while Claude use rose to 72.3%. A total of 51.1% of organizations use AI agents, 47.8% have replaced human tasks with AI, and only 6.5% measure AI returns against defined benchmarks.

When: The survey ran from July 16 to August 14, 2026. Basis published the report and its news post on October 19, 2026.

Where: Basis released the findings from Chicago. The report does not identify the respondents' countries, although its discussion of regulation focuses on the United States.

Why: The findings show that AI use and spending continue to grow, but very few organizations can prove whether their investments improve results. This gap becomes more important as AI agents take on reporting, campaign briefs and creative tasks with limited human supervision.