The marketing team of 2023 and the marketing team of 2026 might have the same headcount, the same titles on the door, and the same desks, but the work happening inside those titles has changed so completely that the old org chart no longer describes the real organization. Roles that used to be about production are now about curation. Roles that used to be about strategy are now about systems. New roles have appeared that did not exist three years ago and senior marketers are scrambling to figure out which of their people are still in the right seat. This is a clear-eyed look at how a small marketing team should be structured for the era when generation is cheap and judgment is expensive.

The thesis is simple. In a world where producing the next ad takes minutes instead of weeks, the team's bottleneck moves upstream from production to direction, and downstream from craft to evaluation. The hours that used to go into making an asset go instead into deciding what should be made, briefing the system that will make it, and judging whether the output is worth shipping. The person who is good at all three of those things is a different person from the one who was good at the original production craft. Sometimes the same human can grow into the new role; often they cannot. Getting honest about which is which is the hardest part of restructuring.

The 2023 three-person team

The reference team for this piece is a fictional but representative direct-to-consumer brand with a three-person in-house marketing function. In 2023 the org chart looked like this:

• Marketing Manager (Maya). Owns the calendar, the budget, the campaign strategy, the agency relationships, and the reporting to the founder. Spends about half of each week in meetings and the other half writing briefs and reviewing work.
• Designer (Devon). Produces the static and motion creative. Works mostly in Figma, Photoshop, and After Effects. Cycles through two to three campaign briefs per month, producing perhaps fifteen to twenty finished assets per month with some help from a contract motion designer at peak times.
• Performance Marketer (Priya). Manages the paid social accounts, sets up campaigns, monitors performance, requests new variants when fatigue sets in, and produces weekly performance reports. Works mostly in Meta Ads Manager, TikTok Ads Manager, and a creative analytics tool.

The chain of work is linear: Maya briefs Devon, Devon produces, Priya deploys and reports. The cycle time from brief to in-market asset is about two weeks. Total monthly output across the team is perhaps fifteen to twenty unique creatives.

The 2026 same-three-people team

Now picture the same three humans in 2026 with a generative platform in the workflow. The same people, the same desks, dramatically different work. The titles have shifted to reflect what they actually do:

• Creative Strategist (Maya). Still owns the calendar and the budget, but the brief-writing has gotten much more granular and is now the primary deliverable rather than a side activity. She spends most of her week translating audience research into prompt templates, defining the variant matrices each campaign should test, and curating which generated outputs are worth shipping. Meetings are shorter because production capacity is no longer the constraint that has to be negotiated in every conversation.
• Brand-System Architect (Devon). The designer-who-produced has become the designer-who-systematizes. Devon spends his week maintaining the brand kit inside the generative platform, building component libraries of approved visual elements, evaluating outputs for brand fidelity, and producing the small percentage of hero work that still requires hand-crafted execution. He produces less individual finished work, but the work he does produce is more leveraged: each component he builds gets used across hundreds of generated variants.
• Performance Operator and Prompt Librarian (Priya). Still manages the paid accounts, but the role has expanded into building and maintaining the team's library of working prompts. She tags every shipped variant with its prompt provenance, tracks performance by prompt template, and weekly reports the prompt patterns that are working and the ones that are fatiguing. The analytics work has gotten more interesting because there are now hundreds of variants in flight at any time rather than dozens.

The cycle time from brief to in-market asset has compressed from two weeks to about three days. Total monthly output across the team is now between one hundred and two hundred unique creatives. The team's bottleneck is no longer production — it is the rate at which Maya can generate well-thought-out briefs and the rate at which Devon can keep the brand system current with the volume of generation. Both of those bottlenecks are addressable with different role definitions and different hiring than the 2023 team would have considered.

The role descriptions in detail

Here is what each of these new roles looks like at the level of detail you would need to actually hire for them or write the job description.

Creative Strategist

The Creative Strategist is the person who turns "we need a campaign for the spring SKU" into a structured brief that a generative system can act on. The deliverable is not a deck describing the idea; the deliverable is a prompt template (or set of templates) with the variant matrix defined, the audience targeting hypothesis explicit, the success metric specified, and the curation criteria written down before the first generation runs.

The skills that matter for this role: strong written communication (because prompts are written instructions and ambiguous prompts produce ambiguous outputs), deep audience research instincts (because the prompts have to ground in something real rather than be invented at the desk), comfort with structured thinking and matrix design (because variant testing is fundamentally about isolating variables), and an aesthetic sense strong enough to curate outputs honestly (because the volume of generation means most outputs are wrong and the team has to be clear-eyed about which ones to ship). Notably, this role does not require traditional design or production skills. A great Creative Strategist may have come from a planning, copywriting, or product-marketing background and never have produced a finished asset themselves.

Brand-System Architect

This is the evolved designer role. The work is less about producing individual finished assets and more about codifying the brand into a form the generative system can use reliably. The deliverables include the brand kit configuration (colors, type, logo handling), the component library (approved photographic styles, illustration vocabulary, motion patterns), the do-not-use list, and the QA rubric the team uses to evaluate generated outputs for fidelity.

The skills that matter: deep brand instincts (the kind you develop from years of producing in a brand), comfort with systems and configuration (because the work is now defining rules rather than executing them), the ability to write clear documentation (because the brand system has to be legible to people who are not designers), and the strength to defend brand standards even under pressure to ship faster. The role suits a senior designer who is tired of being a production bottleneck and ready to move into a more leveraged seat. It does not suit a designer who derives their satisfaction from the craft of finishing individual pieces, because that part of the work has largely moved to the machine.

Performance Operator and Prompt Librarian

The performance marketing role has gained a librarian function. Because the team now ships an order of magnitude more variants than before, and because the value of those variants compounds when the team can tell which prompts produced which performance, somebody has to own the cataloguing of the prompt library and the linkage between prompts, variants, and outcomes.

The skills that matter: traditional performance marketing fluency (because the campaigns still need to be set up and optimized), analytical rigor (because the prompt library is only useful if its tagging is consistent), and a librarian's instinct for organization (because an undisciplined prompt library is worse than no prompt library, since people will trust it and act on bad signals). The role suits a performance marketer who has always wished the team had better feedback loops between creative and outcomes — they finally have the data to do that work.

The before-and-after comparison

The new roles that did not exist before

At slightly larger team sizes — somewhere between five and fifteen marketers — two genuinely new roles tend to emerge that no team had in 2023.

The Prompt Librarian (dedicated). At small team sizes, this lives inside the performance role. At larger sizes, it becomes a dedicated role. The Prompt Librarian owns the canonical store of working prompts, the tagging taxonomy, the access controls, the deprecation process for prompts that have stopped working, and the onboarding documentation that lets a new team member ramp up on the library quickly. Think of it as developer-relations for the prompt library: making it usable, discoverable, and trustworthy.

The Brand-System Architect (dedicated). At small team sizes, this lives inside the designer role. At larger sizes, it becomes a dedicated role distinct from the production designer. The Brand-System Architect owns the configuration of the brand inside every generative tool the team uses, the relationships with the platform vendors, the evaluation of new tools as they emerge, and the strategic decisions about which parts of the brand are codified into the system and which parts remain in human judgment.

The hiring market for both of these roles is thin. The candidate pool is mostly senior people who have been at this work for two to three years and have figured out what works through trial and error. Expect to pay for that experience and expect to lose candidates to competing offers if you negotiate slowly.

The hiring questions that filter for the right thinkers

Hiring for the generative era requires different interview questions than hiring for the production era. A craft portfolio still matters for some roles, but it does not predict success in the new work the way it used to. The questions below filter for the thinking patterns that actually correlate with strong performance in these roles.

For the Creative Strategist seat:

• "Walk me through how you would brief a generative system to produce twenty variants of a back-to-school ad. What axes would you vary and why?" The strong answer talks about explicit variables (testimonial vs. demo, indoor vs. outdoor, parent voice vs. kid voice) and the rationale tied to audience hypotheses. The weak answer talks about general aesthetic preferences.
• "Show me a piece of creative you shipped that did not perform and tell me what you learned." The strong answer attributes the failure to a specific misread of the audience or the platform, not to the design execution. The weak answer blames production quality.
• "How do you decide what not to test?" The strong answer talks about cost of testing, prior evidence, and the limits of attention. The weak answer says "we test everything."

For the Brand-System Architect seat:

• "How would you document the photographic style of a brand so that a system could reproduce it reliably?" The strong answer talks about specificity — lighting characteristics, framing tendencies, color palette, post-processing — and acknowledges the parts that resist documentation. The weak answer says "I would make a mood board."
• "Walk me through a brand standard you have defended under pressure to deviate." The strong answer has a specific story about why the standard mattered and what the cost of deviation would have been. The weak answer is vague.
• "How do you feel about producing less individual finished work and more system?" Listen carefully. If the candidate visibly deflates at the question, they are wrong for the role even if their portfolio is excellent.

For the Performance Operator and Prompt Librarian seat:

• "How would you organize a library of two hundred prompts so that another team member could find the right one in under a minute?" The strong answer talks about taxonomy, tagging, search, and discoverability. The weak answer says "I would use a spreadsheet."
• "Tell me about a time you noticed a pattern in performance data that nobody else had spotted." The strong answer is specific about the data move and the resulting decision. The weak answer is generic.
• "What do you do when an analytics tool gives you a result that contradicts your prior belief?" The strong answer talks about checking the data, looking for confounds, and updating beliefs. The weak answer is defensive.

Who from the old org chart fits which new seat

The hardest part of any restructure is figuring out which existing people grow into which new role. The honest pattern, from the teams we have watched through this transition:

• The marketing manager from the old structure usually grows well into the Creative Strategist role because the core skill — translating business goals into actionable creative direction — is unchanged, just expressed at finer granularity. The growth curve is real but achievable.
• The designer from the old structure splits roughly in half between people who thrive in the Brand-System Architect role and people who become frustrated by it. The deciding factor is whether the person derives their identity from production craft or from brand stewardship. The latter group thrives; the former group is happier moving to a more traditional production role at a different organization where the craft work still exists.
• The performance marketer from the old structure usually grows well into the expanded role because the analytical work has gotten more interesting, not less. The librarian dimension is new for most performance marketers but is learnable.

The restructure that fails is the one that assumes everyone in the old roles will be happy in the new roles. They will not all be happy. Making the difficult call about who is in the right seat is the leadership work that determines whether the transition succeeds.

The leadership shift

At the level above the three-person team, the marketing leader's job has changed too. The traditional marketing leader was a coach who helped the team produce better individual pieces of work. The new marketing leader is more like an editor-in-chief: the person who sets the standard for what gets shipped, the person who maintains the brand system across all the team's outputs, the person who decides which experiments are worth running and which prompts are worth investing in. The role is more curatorial and less directive.

This is a meaningful identity shift for marketing leaders who came up in the production era, and not everyone makes it gracefully. The leaders who succeed are usually the ones who were always more interested in the strategic and editorial dimensions of the work than in the executional dimensions. The ones who are most attached to executional craft sometimes struggle, because the new role asks them to delegate exactly the parts of the work they enjoyed most.

What the org chart looks like in 2028

Predicting two years out is risky in this space, but the direction of travel is clear. The team gets a little smaller in headcount but each person operates at a much higher leverage. The role of the production designer continues to shrink in pure production volume and continues to grow in systems leverage. The role of the prompt librarian becomes a standard part of any marketing team above ten people. New roles will emerge that we cannot yet name — probably around model evaluation, around cross-tool orchestration, around the increasingly thorny work of distinguishing real audience signal from algorithmic noise. The teams that are building thoughtful org charts now will adapt to those new roles more gracefully than the teams that are still pretending the 2023 chart still works.

The marketing team for the generative era is not a smaller version of the old team and not a bigger version either. It is a structurally different team doing structurally different work with the same job titles on the wall. Getting the structure right is more important than any individual tool choice. The tools change; the discipline of clear roles, well-designed seats, and honest hiring is what compounds over time.