12 minOctober 29, 2026
Our tools monitor millions of native, push, pop, and TikTok advertising campaigns.
Get StartedThe conversation about artificial intelligence in marketing often revolves around a familiar question: should marketers become AI builders or simply learn how to use the tools?
It sounds like a reasonable distinction. Some professionals develop custom GPTs, agents, and automated workflows, while others rely on existing AI products to improve their daily work.
But this binary overlooks a more important change taking place in media buying.
As AI becomes more involved in campaign analysis, planning, optimization, and execution, organizations need more than people who know how to build tools. They also need specialists who can maintain those systems, monitor their decisions, and interpret the competitive signals that influence advertising strategy.
A more useful organizational model has three distinct roles:
These roles work together.
Builders make intelligence accessible. Maintainers keep the systems reliable and accountable. Spies interpret what is happening in the market and identify opportunities worth testing.
For performance marketers, this creates a different way to think about AI adoption. The goal is no longer simply to have more people using AI or building custom tools. It is to create an organization that can consistently transform advertising intelligence into better decisions.
Marketing's AI conversation has become too focused on the idea that every professional needs to become a builder.
The pressure is understandable. Businesses are experimenting with custom GPTs, AI agents, CRM automations, and internal workflows. Being able to say, "I built this," can make a project appear more innovative and valuable.
However, not every marketer needs to become a software developer.
A discussion published by MarTech on why not everyone in marketing should be an AI builder explores this distinction. Many professionals will create more value by becoming highly effective users of systems built by specialists.
The same principle applies to media buying.
A media buyer does not necessarily need to build an advertising intelligence platform. They need to understand how to use the available information to make better decisions about audiences, creatives, placements, and budgets.
At the same time, someone must build the infrastructure that makes those insights available, and someone must ensure the systems continue to operate safely.
AI is moving beyond basic copy generation and automated reporting.
Modern advertising platforms can analyze large volumes of campaign data, identify patterns, and adjust certain campaign settings automatically. Depending on the platform, these systems may support bidding, audience optimization, creative testing, and budget allocation.
An overview of AI in advertising technology from illumin discusses how AI can influence campaign management, optimization, and programmatic buying.
This creates new opportunities, but it also introduces new responsibilities.
A system that optimizes toward a particular metric may produce results that look attractive in a dashboard while failing to support the company's broader business objectives.
For example, an AI system focused heavily on short-term acquisition cost might favor inexpensive leads without considering whether those leads become paying customers.
A system that automatically adjusts creative combinations might produce more variations without ensuring that every variation follows the brand's guidelines.
And an automated budget recommendation may be based on incomplete or outdated competitive information.
These are not simply technical problems. They are organizational problems.
AI Builders
Builders create the internal systems that connect competitive intelligence, advertising platforms, analytics, and business data. Their responsibility is to turn fragmented information into useful workflows and decision-support tools.
AI Maintainers
Maintainers oversee the reliability, performance, and governance of those systems. They monitor data quality, review automated decisions, maintain business rules, and make sure AI recommendations remain aligned with company objectives.
AI Spies
Spies are competitive intelligence specialists. They investigate advertising activity, monitor changes in competitor strategies, and translate market observations into creative, bidding, and funnel experiments.
The distinction is important because each role solves a different problem.
A builder can create an excellent intelligence dashboard, but the dashboard cannot automatically determine which competitor signal deserves attention. A spy can identify an interesting market shift, but that insight becomes more valuable when the builder's systems can distribute it to the right people. A maintainer ensures that the resulting automation does not quietly drift away from the organization's goals.
The strongest media-buying organizations will treat these roles as complementary capabilities rather than competing job titles.
Builders are not simply people who write prompts.
Their job is to create the infrastructure that allows a media-buying organization to collect, organize, interpret, and distribute advertising intelligence.
This does not mean every company needs to develop its own demand-side platform or build a complex machine-learning system from scratch.
The more practical objective is to connect existing tools and data sources into a reliable operating environment.
A useful starting point is to separate information into two categories.
The competitive intelligence pipeline
This collects information about the advertising environment outside the organization.
Depending on the available tools and data access, sources may include:
For Anstrex users, the research may include available advertisements, creative variations, destination pages, and campaign activity across supported advertising formats.
The objective is to understand what competitors are promoting, how they present their offers, and which advertising approaches deserve closer investigation.
The internal performance pipeline
This contains information about the organization's own campaigns and business results.
Possible sources include:
The internal pipeline is essential because competitive observations need to be compared against actual business performance.
A competitor may appear to be using a particular creative angle extensively, but that does not establish whether the same approach will work for your audience.
Once the information is collected, builders need to make it usable.
Different tools often organize information in different ways. One may identify an advertisement by creative ID, another by campaign name, and another by destination URL.
Without a shared structure, analysts may spend more time reconciling exports than interpreting the information.
A practical data model could include:
| Data category | Information to organize |
|---|---|
| Competitor | Advertiser or brand |
| Creative | Headline, image, video, and format |
| Offer | Product, promotion, or conversion objective |
| Destination | Landing page or website |
| Market | Geography and relevant audience context |
| Platform | Advertising network or placement |
| Observation date | When the information was collected |
| Internal response | Test, decision, or follow-up action |
Where available, add internal campaign performance data such as spend, clicks, conversions, revenue, and contribution margin.
Keep observed information separate from estimates and interpretations. A visible advertisement is an observation; an assumption about its profitability is a hypothesis.
After establishing a shared data structure, builders can create narrow AI workflows.
The most useful applications often solve recurring operational problems rather than attempting to automate the entire media-buying process.
Examples include:
Competitive monitoring agent
Collects selected competitor advertisements, landing pages, and public announcements, then summarizes what has changed.
Creative research assistant
Groups competitor creatives by angle, format, headline structure, and offer type. It can help identify repeated patterns and generate briefs for new tests.
Campaign analysis assistant
Combines internal performance information with relevant competitive observations to help analysts investigate changes in campaign results.
Weekly intelligence digest
Summarizes important developments, identifies unanswered questions, and highlights potential opportunities for the team to investigate.
A practical workflow follows three stages:
This is more useful than building an agent that simply produces a large collection of screenshots or generic marketing summaries.
Builders should not expect every media buyer, strategist, or creative professional to understand the underlying technical system.
The interface should present information in a format that supports each person's responsibilities.
For example:
The best internal tools reduce the time between discovering an important signal and deciding what to do about it.
Builders create the infrastructure. Maintainers make sure it continues to work as intended.
This role is easy to underestimate because maintenance is less visible than launching a new AI agent or building an automation.
However, an advertising system that makes decisions using outdated data, incorrect assumptions, or poorly defined objectives can create serious problems.
Maintainers need to understand the relationship between system performance and business performance.
Their responsibilities generally fall into four areas: calibration, governance, data quality, and organizational learning.
An AI system is only as useful as the objective it is trying to achieve.
A campaign optimized for low cost per lead may generate large volumes of inexpensive inquiries. But if those leads rarely become customers, the business may be spending money on activity that does not create meaningful value.
Maintainers should periodically review whether the system's optimization objectives still reflect the company's priorities.
Examples include:
They should also examine whether automated recommendations are producing sustained improvements.
One approach is to compare AI-generated recommendations with human decisions on a sample of campaigns. This creates a reference point for evaluating the system's judgment.
Where practical, controlled experiments and holdout tests can help determine whether an automated change is producing incremental improvement rather than simply taking credit for results that would have happened anyway.
AI systems can process large amounts of information and identify patterns that would be difficult for a person to inspect manually.
But optimization efficiency does not automatically mean a decision is appropriate.
An automated system might favor placements that are inexpensive but unsuitable for the brand. It might generate creative variations that overstate product benefits or use audience signals that the organization considers inappropriate.
Maintainers should establish clear rules covering:
These controls should be part of the system design rather than added after a problem occurs.
AI systems need accurate context.
A marketing agent may have access to advertising data but still produce poor recommendations if it does not understand the company's current product information, pricing, positioning, or commercial constraints.
Maintainers should oversee a shared knowledge base that includes:
This information needs regular updates.
A product launch, pricing change, new regulation, or revised brand guideline can make previous instructions inaccurate.
MarTech's discussion of AI agents in marketing workflows provides useful context for the role of shared information and structured workflows in making marketing agents more effective.
Maintainers should also help teams understand why an AI system made a particular recommendation.
A weekly or monthly review can examine:
This creates a feedback loop between the technology and the people using it.
Builders improve the infrastructure. Spies contribute new competitive observations. Maintainers ensure those insights are incorporated without compromising the organization's objectives.
A well-maintained AI system should not simply become more automated. It should become more reliable, explainable, and useful to the business.
If builders create the systems and maintainers keep them reliable, spies turn competitive information into actionable market intelligence.
The term "spy" describes a specialized research function, not simply someone who occasionally browses competitor advertisements.
These professionals use advertising intelligence tools to understand what other advertisers are doing, identify meaningful changes, and develop experiments based on what they observe.
Platforms such as Anstrex can support this process by helping marketers investigate available advertisements, creative approaches, and landing page destinations across supported advertising channels.
The important skill is not collecting more advertisements.
It is knowing what to do with the information.
A single competitor advertisement may reveal very little.
It could be an experiment, a seasonal promotion, or a creative that has not received enough traffic to produce meaningful results.
A stronger research process looks for recurring patterns.
For example:
These observations help identify areas worth investigating.
However, visible creative activity should not be confused with verified media spend, conversion rates, or profitability. Unless a tool provides a reliable and clearly defined measurement, those figures should remain hypotheses rather than facts.
The spy's most important responsibility is turning a signal into a specific experiment.
Consider three examples.
Creative research
Observation: Several competitors appear to be using customer-focused testimonials in their advertising.
Hypothesis: A testimonial-led creative may communicate credibility more effectively than the current product-focused message.
Experiment: Develop three variations using customer stories, expert explanations, and documented product outcomes. Test them against the existing creative under comparable conditions.
Geographic research
Observation: Competitor advertisements appear repeatedly in a particular market.
Hypothesis: The market may offer an audience opportunity worth investigating.
Experiment: Review your own customer data and campaign economics, then test a controlled geographic segment if it fits your targeting strategy.
Funnel research
Observation: Multiple competitors are using an educational pre-sell page before directing visitors to an offer.
Hypothesis: Additional explanation may help address objections among cold traffic.
Experiment: Compare a direct-to-offer landing page with an educational bridge page while keeping the audience and other important campaign conditions as consistent as possible.
The point is to make every research finding lead to a decision or a test.
Anstrex can be part of the spy's daily workflow.
A structured research routine might look like this:
The research should also be documented.
For each observation, record the source, date, market, creative, destination page, and reason it may matter.
This makes it easier to distinguish a recurring market pattern from an isolated example.
A competitive intelligence specialist should not be evaluated simply by the number of advertisements collected or reports produced.
More useful measures include:
Not every hypothesis will succeed. The value comes from consistently identifying relevant opportunities and learning from the results.
Creative intuition remains important in advertising.
Experienced strategists understand audience motivations, brand positioning, and the emotional elements that make a message compelling.
But intuition becomes more powerful when it is supported by evidence.
Modern advertising markets change continuously. Competitors introduce new creatives, adjust offers, enter additional geographies, and experiment with different placements.
A team that relies entirely on internal brainstorming may overlook changes that are already visible in the market.
Competitive intelligence gives those teams another source of information.
Public case studies, press releases, and earnings announcements often describe advertising strategies after they have already been implemented.
Competitive research can sometimes reveal observable activity earlier.
For example, a marketer may notice:
These signals do not prove that a competitor has discovered a profitable strategy. But they can indicate a change worth investigating.
This is the broader idea explored in AdExchanger's discussion of competitive signals inside social ad auctions.
The value is in using those observations to ask better questions.
Instead of asking, "What should we test next?" without any market context, teams can ask:
The answers can inform a creative brief.
For example, if competitor research reveals several advertisers emphasizing product demonstrations, your team could test a demonstration-led creative against its existing lifestyle or testimonial approach.
The goal is not to imitate the market. It is to understand the market well enough to develop a differentiated response.
Competitive signals can also help identify areas for further campaign analysis.
If your internal reporting shows declining performance in a particular geography, and competitor research suggests increased activity in that market, the combined information may justify a closer investigation.
Possible responses could include:
Do not assume that competitor activity alone justifies a budget change. Your own campaign results and commercial constraints should determine whether a change is appropriate.
AI can help process large volumes of advertising information more efficiently.
It can group creatives by themes, summarize landing page structures, flag changes, and organize observations into research briefs.
Advertising platforms can also automate certain bidding and optimization decisions.
But faster processing does not remove the need for human interpretation.
A system can identify a change in advertising activity. A specialist still needs to determine whether that change is relevant, whether the data supports the conclusion, and what experiment should follow.
The advantage comes from combining machine-assisted analysis with human judgment.
The builder role deserves particular attention because it connects the other two specialties.
Without the right infrastructure, competitive research remains scattered across screenshots, spreadsheets, browser tabs, and individual notes.
The builder's responsibility is to turn that fragmented information into a reusable internal intelligence system.
Think of it as an internal advertising brain: a structured environment that helps the organization understand what is happening in its market and connect those observations to its own campaign decisions.
Start by centralizing relevant information from available research tools and internal systems.
Possible inputs include:
The goal is not to collect everything.
Focus on the sources that answer important questions for your business.
A collection of advertisements is difficult to use if every item has to be reviewed individually.
Builders should organize the information into categories that the rest of the organization can understand.
Examples include:
Each decision object should include its supporting observations and the date it was collected.
Where the system generates an interpretation, label it as an AI-generated hypothesis rather than a verified fact.
A useful next step is a conversational research interface connected to the shared intelligence layer.
Media buyers and strategists could ask questions such as:
The system should return concise answers with links to the underlying advertisements, landing pages, and source records whenever available.
This makes the output easier to verify.
The final step is creating a bridge between research and execution.
For example:
Observed pattern: Several relevant competitors are using product comparison pages.
Research question: Are visitors in this market responding to comparison-based messaging?
Proposed experiment: Test a comparison-focused landing page against the current product-focused page.
Measurement: Compare qualified conversions, revenue per visitor, and other relevant business metrics.
Review: Decide whether the result supports expanding the approach.
Over time, these documented experiments create a valuable internal record of what the organization has tested and learned.
The system should make information accessible to the people who need it.
Media buyers should not need to query a database manually to discover a relevant competitor pattern. Creative strategists should not have to interpret raw campaign exports. Managers should not need to read lengthy technical reports to understand what changed.
The builder's work is successful when the rest of the organization can use the intelligence without becoming system developers themselves.
The Builders, Maintainers, and Spies model becomes most useful when the responsibilities are connected through a shared operating process.
Consider a hypothetical example.
A spy researching the home improvement vertical notices that several competitors are using comparison-focused native advertisements and sending visitors to educational landing pages.
The spy documents the advertisements, destination pages, and recurring messages. They do not assume the competitors are profitable; they identify a pattern worth testing.
The builder's intelligence system groups the examples, identifies the common themes, and creates a research brief for the marketing team.
The media buyer and creative strategist review the evidence and develop a controlled experiment using a comparison-led creative and landing page.
The maintainer checks that the experiment follows brand guidelines, uses approved claims, tracks the correct conversion events, and remains within its budget and targeting constraints.
After the campaign collects sufficient data, the team evaluates the results.
If the test improves qualified conversions and business performance, the insight can become part of the organization's creative and media-buying playbook.
If it fails, the team records what was learned and moves on.
The process can be summarized as:
Competitive observation → AI-assisted analysis → Human review → Controlled experiment → Performance evaluation → Updated playbook
Each role contributes something different:
| Role | Main contribution | Key output |
|---|---|---|
| Builder | Connects data and automates research workflows | Internal intelligence systems |
| Maintainer | Protects system reliability and business alignment | Guardrails, audits, and calibrated workflows |
| Spy | Interprets competitor activity and market changes | Research briefs and testable hypotheses |
| Media buyer | Evaluates and executes campaign decisions | Controlled campaign experiments |
| Creative strategist | Turns insights into relevant messaging | Creative concepts and landing page directions |
This model does not require every company to hire five separate specialists. In smaller teams, one person may cover multiple responsibilities.
What matters is that the responsibilities are clearly understood and consistently performed.
The future of media buying is not simply about having more AI tools or building more autonomous agents.
It is about creating an organization that can collect information, interpret market changes, make disciplined decisions, and learn from the results.
The Builders, Maintainers, and Spies framework provides a practical way to organize that work.
Anstrex and other advertising intelligence tools can support the spy function by helping teams investigate competitor creatives, campaigns, and landing page strategies.
AI can help organize and interpret that research. Internal campaign data can help validate new ideas. Human specialists can decide which opportunities deserve attention and which decisions require additional oversight.
The advantage does not come from collecting the most advertisements or automating the most tasks.
It comes from shortening the distance between seeing something important in the market and learning whether it can improve your own advertising performance.
Build the systems. Maintain the standards. Study the competition. Then turn intelligence into action.
Receive top converting landing pages in your inbox every week from us.
How-To
AI is changing media buying, but building more tools is not the only answer. Explore a three-role organizational model that combines AI infrastructure, system oversight, and competitive ad intelligence to help media-buying teams make better decisions.
Rachel Thompson
7 minOct 29, 2026
Most Read
Industry averages offer a useful starting point, but they rarely reflect the competition in your exact market. Learn how to build more meaningful conversion benchmarks using competitor research, funnel analysis, and live advertising signals.
Elena Morales
7 minOct 29, 2026
Most Read
Creative production is only one part of performance marketing. Discover how to use competitive intelligence to map funnel patterns, turn research into AI-assisted ad concepts, test landing pages continuously, and connect campaign decisions to measurable profitability.
Priya Kapoor
7 minOct 28, 2026
