Artificial intelligence has become routine in marketing, but widespread access has not yet produced equally widespread organizational change. The central challenge in Q4 2026 is connecting AI tools with company data, applications and working practices so that adoption translates into business value.
MarTech receives a steady stream of research on AI, and the findings do not always align. Depending on the report, the approaching holiday shopping period will be either the first or second season shaped by AI. Forecasts also disagree on whether agents will handle some shopping activities or the entire purchase process.
There is more agreement about marketing itself. Almost all marketers use AI, most commonly to produce content. Some also build applications, support personalization and improve workflows. Yet research offers differing assessments of investment returns, workforce effects and hiring intentions. The observations below reflect the MarTech author's conversations with marketers and marketing technology leaders, both public and off the record, rather than a single survey's verdict.
Connecting AI to everyday work remains difficult
Across the research, more than 90% of marketers report using AI. Many have settled on one large language model, or LLM, and some use several. Access to basic assistance for individual tasks is no longer the main obstacle.
The harder step is making those tools part of a connected team process. Integrating an LLM with other applications can demand technical expertise and additional approvals. What a team can build depends on IT policies, available technical support and whether colleagues have the knowledge to use the resulting solution.
These constraints help explain why so many marketers still copy model responses and paste them into another application. Manual transfer is straightforward, but it falls short of an integrated workflow. The gap is often organizational or technical, rather than a simple lack of interest among marketers.
Separate AI features can reinforce separate systems
AI now appears in nearly every application used by marketing teams. Implementations range from well-designed assistants and integrations to additions that feel awkward. The same pattern extends beyond specialist marketing software to Excel, Google Sheets, PowerPoint and other productivity tools.
An assistant usually helps inside its own application. Its presence does not establish a connection to the next tool in the process. A CRM, a marketing automation platform and office applications can therefore remain disconnected even when each offers AI capabilities.
This creates a concern for vendors as well as users. Building AI into software costs money, while marketers have limited capacity to engage with competing features. A supplier's investment only matters to users if they choose to use it. Agents may help address the connection problem, but almost every vendor now offers agents too.
Weak data undermines confidence in AI results
Many marketers are unsure about the quality of their data. That uncertainty carries over to AI because a model can amplify problems already present in its inputs. Inaccurate information can then reach customers or colleagues.
The article uses Salesforce as an example. Its ecosystem places Data 360 beneath Agentforce and the CRM as a data layer. The author supports the emphasis on data management and governance while acknowledging concerns about expensive contracts and being sold a broader solution than a company intended to buy.
Salesforce has made AI central to its direction, and poor data can jeopardize those initiatives. The same dependency matters to marketing teams. They can expand AI use with the information they already hold, but expanding usage does not correct weaknesses in that information.
Increasing output does not ensure stronger marketing
AI makes it possible for one person to generate large quantities of copy, designs, drafts and emails. That capacity encourages an existing marketing habit: filling more channels with more messages.
Several marketing leaders interviewed by the author believe teams are placing too much emphasis on volume. Producing additional material is easy, but that alone says little about whether it resonates with its audience or has a meaningful effect. From a customer's perspective, a larger stream of messages is not automatically an improvement.
The value of the work therefore cannot be inferred from the number of assets produced. Marketers still need to consider what the material communicates and how people receive it.
Review and revision can absorb the expected time savings
The burden becomes particularly visible in content production. Many marketers report that checking substantial AI output creates additional work and prevents them from achieving the efficiency they expected.
Tools and guardrails can help maintain a brand's voice and style. However, many teams still work primarily through an LLM interface and rely on prompts to steer the result. Prompts can support idea development and a first draft, but they do not remove the need for further work.
A piece of content being generated by AI is therefore insufficient evidence of a more efficient process. The effort required to review and improve it also belongs in the assessment.
Customer needs should determine where AI is useful
Greater production is one example of teams focusing on what AI lets them do without fully considering the recipient. Customers do not necessarily want additional content. AI chat and assistance functions also do not improve every interaction.
Sometimes a person simply needs a fast answer, which AI can provide effectively. Other situations call for empathy, understanding and human involvement. A useful customer experience depends on recognizing the difference.
Marketing teams should examine the reason behind an increase in output. Is it responding to demand from customers, or mainly reflecting how easy generation has become? Technical convenience alone does not establish customer value.
Proving returns requires a broader view of costs and outcomes
Marketers already face pressure to justify brand campaigns, event sponsorships and advertising. AI investment now brings the same obligation. In 2026, fewer marketers are confident that they can demonstrate ROI than in 2025.
Marketing attribution remains complicated, especially in B2B. Nevertheless, retaining a budget requires evidence that spending delivers a return. Counting content items or other outputs provides only part of the picture.
An assessment of AI investment must also account for rework, integration expenses, customer outcomes and business impact. Without these elements, higher production can give an incomplete impression of the value created.
Model changes can disrupt established workflows
A team may successfully build an LLM-based solution and incorporate it into individual or shared work, only to find that its behavior later changes. Models evolve rapidly, and updates can introduce uncertainty into processes that previously worked reliably.
Marketers then have to investigate why yesterday's dependable tool behaves differently today. How disruptive this becomes depends on how deeply AI has been embedded in the workflow.
Further progress requires a strategy for the whole organization
Together, these eight observations show that tool adoption has moved faster than organizational transformation. Marketing teams have embraced AI, but their systems and working practices often have not changed enough to realize its potential.
Readily available LLMs made it easy for individuals to get started. Connecting AI platforms, existing applications and data requires decisions across the company. Marketing cannot solve that challenge alone; technology and business leaders need to develop the next stage of the strategy together.
The marketing questions themselves remain familiar: the balance between quality and quantity, workflow efficiency, customer experience and business value. AI changes the tools available to address them, but teams should not expect the technology to resolve those questions on its own.





Комментарии