EP28: Automate, Analyse, Accelerate – AI Marketing That Delivers with Push Group
From Digital Marketing to AI Marketing
Ricky co-founded Push in 2007 after recognising how the internet was changing consumer behaviour.
The business initially focused on websites, search engine optimisation and Google advertising, helping companies understand how customers were increasingly using search engines to find products and services. As digital marketing matured, Push became a specialist performance marketing agency, working across Google, Meta, TikTok, Microsoft and other platforms.
The emergence of generative AI created another major shift.
Although Push had already been experimenting with AI internally, the launch of ChatGPT made the technology accessible to a much wider audience. Ricky and Steve began to see parallels with the arrival of Google: a new technology was not simply creating another marketing channel, but potentially changing how people found and interacted with information altogether.
A conference in the US in 2023 reinforced that view. An event originally planned around marketing traffic was almost completely repositioned around AI shortly before it took place.
For Push, it was a signal that the industry was changing quickly.
The company subsequently repositioned itself around AI, setting a deadline for the transition and restructuring its offering around consultancy and training alongside its existing marketing services.
From Service Provider to AI Partner
The repositioning changed not only what Push offered but how it worked with clients. Alongside traditional campaign management, the company began helping organisations understand what AI could mean for their businesses, where it could generate a return and how it could be introduced safely.
This created three connected areas of work: marketing delivery, consultancy and training.
For some existing clients, AI is largely invisible because Push incorporates its own tools and processes into campaign delivery. For newer clients, however, AI is often part of the reason they engage with the company in the first place.
The conversation has also expanded beyond marketing. AI makes it easier to connect marketing activity with sales and customer experience. A campaign can generate a lead, for example, but AI can then help nurture that lead automatically rather than leaving the business dependent on someone following up manually.
This creates a broader opportunity: using AI not simply to produce more marketing, but to connect the different stages involved in turning marketing activity into revenue.
Building AI Agents Around Marketing Workflows
One of Push's approaches is to create specialised AI agents around different parts of its marketing process.
The company builds a knowledge base containing information about a client's business, competitors and strategic objectives. Different agents can then draw on this information for specific tasks. There might be a research agent, a persona-creation agent, an advertising agent or a creative agent, each operating according to the processes Push has developed.
This allows marketers to interact with the relevant agent rather than starting from scratch each time. For example, someone working on creative could ask the relevant agent to produce several versions of advertising copy for a particular persona, using the information already contained within the client's knowledge base.
That output can then feed into other tools, including AI creative platforms, before the resulting assets are used in campaigns. The aim is not simply to automate individual tasks, but to connect them into a wider workflow.
From Individual Agents to End-to-End Automation
Push initially developed its own technology because, when it began experimenting with agents, there were few tools capable of doing exactly what the business needed.
Its platform, Dial Agents, became the framework for its agents and processes, with clients paying a technology fee for access.
As the market has developed, however, Push has increasingly been able to use external platforms that offer capabilities beyond its original system. This does not make the original work redundant. The underlying assistants and processes can potentially be connected to newer platforms that provide greater workflow automation.
Ricky describes a recent experiment where changing a lead to an opportunity in Push's CRM triggered a sequence of different AI assistants. Within minutes, the system had completed research on the prospective client and sent the resulting information to the team through Slack.
The significance was not just the time saved. Seeing the process work encouraged the team to consider what other workflows could be connected in the same way.
The Biggest Challenge Is Adoption
Having the technology is only part of AI transformation.
Ricky identifies adoption as one of the biggest challenges facing both Push and its clients. People need to change established habits and incorporate AI into their everyday working practices; that requires continuous training and reinforcement rather than a one-off technology rollout.
Push itself periodically runs internal “reset” sessions to review its AI tools, discuss new technologies and reconsider how employees should be using AI across the business.
It takes a similar approach with clients, running workshops that combine an overview of the changing AI landscape with practical exercises. Rather than simply telling employees about the latest tools, the workshops encourage them to identify a real bottleneck in their own organisation and consider whether an AI workflow or agent could address it.
The emphasis is therefore on practical experimentation.
AI Adoption and the Fear of Job Displacement
Marketing teams are often early adopters of new technology, but Ricky has found that AI adoption is not uniform even within marketing departments. Senior marketing leaders can be particularly receptive because they are thinking strategically about whether AI can increase output, improve efficiency and enable their teams to achieve more.
However, employees closer to day-to-day execution may have different concerns.
One reason is uncertainty about what AI adoption means for their own roles. Employees may wonder whether introducing AI will ultimately affect their careers or make their work less valuable. For Ricky, this highlights the importance of having more open conversations about why organisations are implementing AI and how employees will benefit.
Push also conducts AI audits to understand how different parts of a business are currently using AI. These audits can identify whether employees are already using tools without leadership awareness, whether the tools being used are appropriate and whether people are handling data safely.
The findings can then inform a roadmap for organisational adoption and cultural change.
The Emergence of the Hybrid Marketing Team
The future of marketing may involve teams in which humans and AI agents work alongside each other. Ricky describes a model in which AI performs a large proportion of the initial work, with humans providing quality assurance, judgement and refinement.
This changes the nature of the human contribution. Rather than spending most of their time carrying out repetitive production tasks, marketers can focus more heavily on strategic thinking, creativity and decision-making.
The next stage could take this even further by making agents directly visible to clients. Instead of an agency using AI behind the scenes and simply presenting the final output, clients could potentially interact with the same systems, review work and provide feedback within a shared environment.
This would turn AI from a hidden productivity tool into part of the collaborative relationship between agency and client.
The Brief Could Become a Living Process
One particularly interesting implication is how AI could change something as fundamental as the creative brief.
Traditionally, a brief is often treated as a relatively static document: information gathered during meetings is turned into a formal reference point for the creative team.
AI could make this much more dynamic. A specialised agent could draw on previous client conversations and projects, identify gaps in a brief and ask additional questions. As the project develops, it could continue prompting for clarification or identifying changes in requirements.
The brief therefore becomes less of a document and more of an ongoing conversation.
For marketing teams, that could make AI valuable not just at the production stage, but at the point where the problem itself is being defined.
AI Is Changing Creative Production
The pace of development in AI-generated creative content has been particularly striking.
Ricky notes that AI video tools were much less convincing a year earlier, whereas current systems can generate increasingly sophisticated video from relatively simple inputs.
Marketers can specify a script, persona, actor, voice, pace and style before combining the generated material with human-created footage and other assets. The result is dramatically faster production. Push has seen situations where AI-supported creative production can operate at several times the speed of traditional production, while significantly reducing costs.
More importantly, the performance data can then determine which assets are worth scaling.
In one example discussed on the podcast, an AI-generated creative contributed to a 20% reduction in customer acquisition cost in a split test. The client subsequently began considering moving a substantial proportion of its creative production towards AI.
Yet Ricky does not see this as a future where everything becomes AI-generated. Human-produced assets still have an important role, particularly in brand building.
Performance and Creativity Need Different Roles
Ricky makes an important distinction between different parts of the marketing funnel. AI-generated creative can be particularly effective when the goal is performance optimisation: producing variations quickly, testing them and identifying what drives conversions.
But the upper part of the funnel has a different purpose.
Brand campaigns need creativity, identity, values and cultural understanding. They establish what a business stands for rather than simply optimising a particular conversion. That means human creative teams remain important.
The future may therefore be less about replacing traditional creative work and more about combining human-created brand assets with AI-generated performance content.
For Ricky, that balance matters because brands risk losing something if every piece of communication is generated solely according to what an algorithm predicts will perform.
The Risk of Losing Control of the Brand
There is a broader question about what happens if platforms themselves begin generating increasingly personalised advertising. Ricky points to the possibility of platforms such as Meta eventually creating advertising in real time based on individual interests and behaviours.
A single product could theoretically be presented differently to two people depending on their activities and preferences.
From a performance perspective, this could be extremely powerful; but it could also reduce the control brands have over how they communicate. If platforms generate the creative themselves, brands may have less visibility over exactly what different audiences are seeing. That could lead to a more efficient advertising ecosystem, but potentially a less distinctive one.
The challenge will be ensuring that personalisation does not eliminate the creativity and cultural insight that make brands memorable.
Using AI to Expand What Small Teams Can Achieve
For businesses considering AI adoption, Ricky's advice is to start experimenting.
Tools such as ChatGPT provide a relatively accessible starting point for understanding what generative AI can do. Businesses can experiment with prompts, context and different types of outputs before moving into more sophisticated applications.
From there, the focus should shift towards identifying where AI could create tangible value. For marketing teams, this might mean asking where the brand could be present but currently is not, and then considering how AI could make it possible to create and distribute content across additional channels.
A single piece of content can potentially become a blog post, social media content, video clips and other formats much more quickly than before. For smaller businesses in particular, this creates an opportunity to compete with organisations that have much larger marketing teams.
AI can allow a small team to increase its output without necessarily increasing its size at the same rate.
Turning Marketing Data into Recommendations
Looking ahead, Ricky is particularly excited about the development of AI-powered data analysis.
Marketing organisations already generate huge amounts of data across advertising platforms, CRM systems and other channels. The challenge has traditionally been turning that information into reliable, actionable insight.
AI is beginning to make that process more accessible. Push's platforms aggregate data from different advertising channels and can identify high-performing assets before generating recommendations for improving them.
The next step is to connect those recommendations directly to creative production: identify what is working, understand why, generate a new variation and return it to the testing cycle.
The more effectively those stages can be connected, the closer marketing comes to becoming a continuously learning system.
Marketing's AI Future Is About More Than Efficiency
For Ricky, AI's potential ultimately goes beyond automating individual marketing tasks. It is about changing how businesses think about the relationship between data, creativity, technology and people.
The immediate opportunity is to help teams work faster and do more with limited resources. But the longer-term transformation could involve entirely new relationships between marketers, clients, platforms and AI agents.
The challenge will be retaining the human qualities that make marketing effective while taking advantage of AI's ability to analyse, produce, test and iterate at unprecedented speed.
For businesses beginning that journey, Ricky's message is straightforward: start experimenting, identify practical opportunities and build from there. The technology is moving too quickly for organisations to wait for a perfect roadmap.
The businesses that benefit most may be those willing to learn by doing — while keeping humans firmly involved in the parts of marketing where judgement, creativity and brand identity matter most.