How Large D2C Brands Should Structure Their Marketing Org (2026)

Why is org structure the real growth ceiling?

When a scaled account stalls, everyone audits the media first: bids, audiences, pixels, creative. The audit usually finds only symptoms, because the disease sits one level up — in who decides what, how fast learnings travel, and how many jobs one talented person is quietly holding. We described the mechanics in what breaks past ₹50 lakh a month: the decision volume, creative burn and measurement complexity of a scaled account outgrow the two-person structure that built it, and the failure arrives gradually enough to be misdiagnosed as a performance problem for two quarters.

This guide is the structural answer, drawn from six years across 160+ D2C brands and the enterprise budgets we currently run at ₹25–60 lakh a month per brand: which functions must be separated, who owns the number, what the operating cadence is, and how the structure should change by spend band.

What changed about the work in 2026?

As of 2026, the platforms have restructured the job themselves. Meta’s Advantage+ suite and Google’s Performance Max — per the platforms’ own product documentation and years of trade coverage — have automated much of the bid-and-audience labour that media teams were originally built around, shifting the human leverage to what feeds the machine: creative volume, offer architecture and measurement design. Directionally, that inverts the classic org: the media buyer stops being the hero role, and the constraint moves to creative supply — our own published working norm is 8 to 12 genuinely new concepts a month for a scaled account, rising with spend per the fatigue arithmetic — and to measurement, where automation’s self-reported success needs independent verification. An org chart designed for 2019, with three buyers and a designer, is structurally wrong for this decade regardless of how good the three buyers are.

The org chart is the media plan. Every structural gap shows up in the account within a quarter, wearing a CAC costume.

Which four functions must be separated?

The same four we flagged in the scaling guide, now as job descriptions. Media trading: daily budgets, pacing, exclusions, platform hygiene — the operational cockpit. Creative strategy: deciding what gets made next and why: hooks, angles, offers, the testing agenda — the highest-leverage seat in the modern account and the one most often left vacant. Creative production: shipping the volume the strategy orders, on a contracted monthly quota sized to spend. Measurement: MER, contribution, the credit and curve sheets, and the experiment calendar — deliberately separated from media trading, because the person spending the budget should not be the only person grading it.

The separation matters more than the staffing model. Any function can sit with a partner — production most commonly does, trading often, measurement design never — but each needs a named owner, and no owner should hold more than two. The standard failure, one good buyer holding all four at half capacity, is not a talent problem; it is an org design that guarantees the account lives off savings.

What structure fits each spend band?

Two notes on the middle band, where most readers of this page live. First, “four names” does not mean four employees — it means four accountabilities, each with exactly one name, some internal, some at partners; the hybrid split that works is mapped in our in-housing guide. Second, the first hire that changes everything is almost never another buyer; it is the creative strategist, because that seat converts media spend from a bidding problem into a learning system. Third, the bands are about decision volume, not prestige — a brand at ₹40 lakh a month with clean structure will outperform a ₹1 crore account with a hero buyer and a vacant strategist seat, every quarter, and the gap widens with spend.

Who owns MER, and what are their decision rights?

One person. Internal head of growth or lead-partner principal — but one, named, and answerable for the blended number, because shared ownership becomes rotating blame at exactly the spend level where blame is expensive. Around that owner, write the decision rights down — a one-page RACI outperforms a hundred alignment meetings:

• Budget moves: the MER owner reallocates freely inside the month up to an agreed threshold; beyond it, founder or CMO sign-off. Nobody else touches budgets, including the founder mid-week.
• Kill rights: media trading can pause any ad on performance; only creative strategy can kill a concept, because concept-level verdicts need the testing context.
• Test authority: measurement owns the experiment calendar and its pre-committed decisions; a test result executes without relitigating unless the owner formally overrides — and logs it.
• Escalation: defined guardrails — MER floor, spend ceiling, delivery anomalies — trigger a same-day conversation, not a month-end surprise.

The cadence that holds it together is the weekly trading meeting, forty-five minutes, fixed agenda: spend versus plan, MER versus guardrail, creative pipeline versus quota, test readouts, decisions logged. Boring on purpose; the boredom is the point.

How does the creative engine fit into the structure?

Because creative is where scaled org charts actually fail, the seam between the strategy seat and the production quota deserves its own page in the design. The working contract: creative strategy owns the question list — which angles, offers and audiences the account needs to learn about next quarter, drawn from the testing framework and the fatigue data, written as briefs that specify the hypothesis rather than the storyboard. Production owns the shipping rate — a monthly quota in genuinely distinct concepts, not resizes, contracted like a supply agreement whether the producer is internal, a partner, or our own studio model, and reviewed in the weekly meeting as pipeline-versus-quota, one line above the media numbers it feeds. Media trading owns the verdicts at ad level, measurement owns them at concept level — a distinction that sounds pedantic until you watch an account kill a strong concept because its first execution shipped into a bad week. Two structural details protect the seam. First, the learning log: every tested concept gets a one-line autopsy — hypothesis, result, what it changes about the next brief — so the org accumulates creative knowledge instead of re-testing last year’s failures in new fonts. Second, the strategist attends the trading meeting, always; an org where creative decisions happen in a separate room from budget decisions is two orgs, and the account will perform like it. Size the whole engine from spend, not from team capacity: the quota is a function of burn rate, and if the org cannot ship it, that is the constraint to fix before any media optimisation will hold, because no bidding strategy compensates for an empty pipeline.

What are the common failure patterns?

Five recur across every scaled account we audit. The hero buyer: all four functions in one head; works brilliantly until roughly ₹30 lakh a month, then becomes the bottleneck nobody wants to name. The vacant strategist seat: production ships variants of last quarter’s winner because nobody owns deciding what to learn next — the account tests fonts while competitors test offers. The graded-homework problem: measurement reporting into media trading, which is how a quarter of over-credited retargeting survives four QBRs; keep the sheets with someone whose bonus does not depend on their contents. The founder as random reallocator: mid-week budget moves from the top that reset every learning in flight — solved by decision rights, not by better founders. The partner black box: an agency holding trading, strategy and measurement with no internal truth layer — fine until it is not, and unfixable in a hurry; the antidote is the internal baseline from month one, whoever runs the media. If you are choosing that partner now, our evaluation checklist tests for exactly these structures, and our enterprise engagements are deliberately built with the four functions separated and the truth layer on your side of the table.