AI use among marketers surges: tech skills stall

Show summary Hide summary

• Big jump in AI adoption across brands and agencies
• Real-world uses that consumers encounter today
• Workforce readiness is lagging behind tool deployment
  • Upskilling versus tool training
  • From comfort to mastery
• Infrastructure costs and the ROI question
• Most teams choose out-of-the-box tools over custom models
• Platform partnerships and a more open AI ecosystem
• Emerging tech stacks and cost-saving approaches
• Practical steps marketing leaders are weighing now

Marketing teams are racing to weave AI into their operations, yet adoption outpaces the skills and strategy needed to realize the full payoff. New research from Modern Retail+ shows fast uptake across brands and agencies, but also exposes gaps in training, economics and how companies deploy AI tools.

Big jump in AI adoption across brands and agencies

Year-over-year survey data shows a steady climb in companies investing in AI. The trend reflects both curiosity and urgency as firms try to keep pace with competitors.

• 2022: Roughly four in 10 firms began investing in AI.
• 2023: Investment increased notably.
• 2024–2025: Adoption surged, reaching the vast majority of respondents by 2025.

Alongside that growth, many organizations created executive roles focused on AI. New chief AI officers have appeared at major brands and agencies. That shift signals that AI is no longer experimental for many firms.

Real-world uses that consumers encounter today

Brands are moving beyond pilots and launching AI features for customers. These implementations include personalized promotions and guided product selection.

• Retailers using purchase histories to tailor offers for loyalty members.
• Specialty stores deploying chatbots to recommend gear and configurations.

Those consumer-facing deployments help normalize AI for users, but they also raise expectations for how smart and useful these tools should be.

Workforce readiness is lagging behind tool deployment

Many teams add AI tools quickly but struggle to develop employee capabilities to match. The gap is apparent at both individual and organizational levels.

Upskilling versus tool training

Some leaders say teaching people a new interface is not the same as increasing expertise. Designing better prompts or using a new app does not replace deeper craft skills.

Upskilling should multiply human judgment and domain knowledge, not simply swap in a new tool.

From comfort to mastery

While many users now understand baseline tools, far fewer push those tools to their limits. Familiarity with a chatbot does not equal mastery of automation across systems.

Infrastructure costs and the ROI question

Spending on cloud compute, GPUs and TPUs is creating a new line item for marketing budgets. That spending forces a hard look at return on investment.

• Deploying models at scale means paying API and compute fees.
• Leaders ask whether productivity gains justify ongoing infrastructure bills.

Brands that rushed to announce AI pilots sometimes delayed planning on cost models and monetization. That has made finance teams more cautious.

Most teams choose out-of-the-box tools over custom models

The survey found a clear preference for ready-made AI solutions rather than building from scratch. Reasons include cost, speed and talent constraints.

• Large majority use off-the-shelf AI tools in daily work.
• Fewer than half extend existing LLMs to build proprietary tools.
• A small share train and maintain their own LLMs.

For many organizations, buying a prebuilt model or platform is the fastest route to value. Custom work remains expensive and technically demanding.

Platform partnerships and a more open AI ecosystem

Major vendors are making it easier for marketers to mix and match models and creative tools. That shift reduces vendor lock-in and speeds experimentation.

• Cloud platforms now offer prebuilt models that can be customized.
• Creative suites are integrating third-party models for asset generation.
• Brands can combine tools from different providers in a single workflow.

Democratization of models means teams can start with a baseline and layer customizations, rather than building everything internally.

Emerging tech stacks and cost-saving approaches

Companies are experimenting with hybrid stacks that use multiple LLM providers and specialist APIs. That approach can keep costs down and increase flexibility.

1. Choose a base model from a major provider.
2. Stack additional models for niche tasks or higher accuracy.
3. Run experiments in a sandbox to avoid large upfront expenses.

Some teams treat products like Copilot, Claude or Firefly as foundations they can augment. This strategy often delivers faster returns than building proprietary models from day one.

Practical steps marketing leaders are weighing now

Executives tell researchers they are balancing investment, training and vendor choice as they scale AI. Common considerations include governance, measurable KPIs and team structure.

• Invest in training that focuses on craft, not just tools.
• Model the economics of adding seats or API usage before scaling.
• Favor modular tech stacks to prevent costly lock-in.

Thinking through people, process and platform together is emerging as the best way to turn AI adoption into sustained advantage.