Survey: marketers say AI saves time, yet 25% report more burnout
Quick answer
In a survey of 252 AI-using marketing creatives, 46% reported saving at least five hours a week, while 25% reported more burnout. Magic Hour fielded the survey September 16–18, 2026. The unweighted responses describe participants’ perceptions, not measured productivity, a causal effect of AI, or all marketing creatives.
Key findings
Finding | Survey result | How to interpret it |
|---|---|---|
Saved at least five hours per week | 46.0% | Self-reported; combines 5–9, 10–14 and 15+ hour categories |
AI touched at least half of deliverables | 26% | Share of surveyed AI users, not all creative professionals |
AI-assisted work outperformed human-only work | 38% | Respondent perception; no common KPI or controlled comparison |
AI-assisted work performed on par | 36% | Respondent perception |
Burnout decreased after adopting AI | 48% | Self-reported change |
Burnout increased after adopting AI | 25% | Self-reported change |
Methodology and limits
Centiment Audience conducted the survey for Magic Hour from September 16 through September 18, 2026. The sample contained 252 creative professionals in marketing departments who reported using AI in their workflows. Data was unweighted. The stated approximate margin of error for the full sample is ±6 percentage points at a 95% confidence level.
- Population boundary: the sample was screened for current AI use, so the results cannot estimate adoption among all marketers or creatives.
- Self-reporting: hours saved, KPI effects and burnout were reported by respondents and were not observed in logs or a controlled study.
- Subgroups: role, industry and workplace percentages have smaller bases and wider uncertainty than the full sample; subgroup counts are not published on this page.
- Questionnaire availability: the exact questionnaire, response-level data and crosstabs are not embedded here, which limits independent recomputation.
- Rounding: displayed percentages may not total 100 because of rounding, omitted response categories or multi-select questions.
- Causality: associations in this survey cannot show that AI caused time, KPI or burnout changes.
46% reported saving at least five hours per week
Self-reported weekly time saved | Respondents |
|---|---|
0 hours | 1.2% |
1–2 hours | 12.3% |
3–4 hours | 40.5% |
5–9 hours | 26.2% |
10–14 hours | 12.7% |
15+ hours | 7.1% |
The “at least five hours” figure is 46.0%, the sum of the 5–9, 10–14 and 15+ hour response categories. The survey does not verify timesheets or output volume, so this is best read as perceived time savings among respondents.
Reported savings were larger in several subgroups: 65% of respondents in technology said they saved at least five hours; 49% at agencies or consultancies said the same. Designers reported the largest 10+ hour share at 26.7%, while 28.6% of production-studio respondents reported 10+ hours. Because subgroup counts are not shown, these comparisons should be treated as descriptive.
Respondents associated AI with both better and worse outcomes
Thirty-eight percent said AI-assisted assets outperformed human-only work, 36% said performance was on par and 6% said it was worse. The survey did not apply one shared KPI definition or randomize work to AI and non-AI conditions, so these results describe respondents’ judgments.
Area respondents associated with improvement | Respondents selecting it |
|---|---|
Time to first draft | 40% |
Engagement | 39% |
Cross-channel consistency | 33% |
Click-through rate | 27% |
These improvement areas were reported by respondents and should not be interpreted as independently measured lifts. A team can turn them into testable hypotheses by defining one KPI, retaining the non-AI baseline and holding audience, offer and distribution constant.
26% said AI touched at least half of their deliverables
Across the full sample, 26% said at least half of their deliverables were AI-assisted. Role-level responses varied: the share reporting at least 50% AI-touched output was 30% in visual design, 21% in writing and content, 19% in video and photography and 29% in strategy and leadership.
“AI-assisted” can include brainstorming, research, drafting, editing, generation or accessibility work. It does not mean that a model created half of every finished asset, and the survey does not provide a common intensity measure across roles.
Saved time most often went back into exploration
Deeper exploration of work in progress was the most commonly reported destination for saved time at about 32%. Other reported uses included taking on more deliverables (15.9%), strategic planning (15.5%), rest and recovery (10.4%) and more stakeholder or client communication (5.6%). The chart also includes learning and experimentation.
This suggests a practical operating question: whether saved time is deliberately reassigned. Without an explicit choice, faster first drafts can become more variants, more review work or a higher expected workload rather than lower cost or better output.
Burnout moved in both directions
Forty-eight percent reported less burnout after adopting AI, while 25% reported more. Writing and content respondents had the largest reported decrease at 68%. Decreases were 44% for visual design, 46% for video and photography, 46% for strategy and leadership and 43% for multidisciplinary roles.
Production-studio respondents were more divided: 38% reported less burnout and 31% reported more. The survey cannot identify the cause. Workload expectations, learning effort, review burden, job uncertainty and team process are plausible explanations that require separate measurement.
Brainstorming was the most common time-saving task
Task | Respondents identifying it as a major time saver |
|---|---|
Brainstorming | 49% |
Accessibility work | 40% |
First drafts | 40% |
Research and synthesis | 37% |
These tasks share a pattern: they create candidate material or remove repetitive setup. Teams still need a named owner and acceptance criteria for factual accuracy, brand fit, accessibility and final approval.
How a creative team can use these findings
- Pick one repeated task. For example: first storyboard, caption draft, product-image variant or transcript cleanup.
- Record the baseline. Measure elapsed time, accepted-output rate, revision rounds and the business metric the asset serves.
- Keep the quality bar fixed. Do not call a faster draft a gain if review time, errors or rejection increase.
- Assign saved time. Decide whether it goes to more exploration, stakeholder review, additional output or recovery.
- Measure burnout separately. Ask about workload and control before and after adoption; throughput alone cannot describe well-being.
- Publish the method. If sharing results, include sample, denominator, question wording, dates and limits alongside the headline.
A red ceramic mug sits on a wooden café table beside a window. Steam rises slowly while the camera makes a gentle push-in. Soft morning light, realistic materials, one continuous shot, no people or readable text.
Test one creative workflow
Choose one repeated creative task, record the current time and acceptance criteria, then test an AI-assisted version without changing the quality bar.
Explore AI Video ToolsFrequently asked questions
252 creative professionals in marketing departments who reported using AI. The fieldwork ran September 16–18, 2026 through Centiment Audience.
46.0%: 26.2% reported 5–9 hours, 12.7% reported 10–14 hours and 7.1% reported 15 or more hours. The measure was self-reported.
48% reported lower burnout, while 25% reported higher burnout. The survey was observational and cannot show that AI caused either change.
38% said AI-assisted assets outperformed human-only work, 36% said they performed on par and 6% said they performed worse. Those are perceptions across respondents using different KPIs, not one controlled performance test.
Choose a repeated, reviewable task and hold its acceptance criteria fixed. Compare current and AI-assisted time, rejected outputs and revisions. Explore current AI marketing tools or test image generation, video generation and subtitle generation only where they match that task.
