Marketing Production And Campaign Flow

Marketing teams rarely describe themselves as having a delivery problem. They describe themselves as having a capacity problem, a stakeholder problem or a brief quality problem, and the remedies proposed are correspondingly a bigger studio, a stricter briefing template or an agency. Yet if you take a single piece of creative work — a landing page, a product video, an email sequence, a piece of long-form content — and reconstruct its journey from approved brief to live, you will find the same shape that appears in every other knowledge-work value stream. A small amount of making. A very large amount of waiting.

The making is what everyone discusses. The waiting is where the calendar went.

This matters more in marketing than in most functions, because marketing's value is unusually time-sensitive. A campaign that lands three weeks after the moment it was designed for has not merely arrived late; it has arrived at a different market, often against a competitor who moved first. The cost of delay in creative production is not a smooth line. It is frequently a cliff.

Map the asset, not the campaign

Start at the wrong altitude and you will learn nothing. Campaigns are too large and too heterogeneous to reveal a pattern; individual assets are the right unit.

Take ten recent completed assets of a comparable type. For each, record the moment the brief was agreed, every subsequent handoff with its timestamp, and the moment it went live. Then mark each interval as touch or wait, exactly as you would in any other value stream.

The composition is consistent enough to predict. Creation — writing, designing, filming, building — is a small share of elapsed time. Waiting for a reviewer is a large share. Waiting for the next scheduled review forum is another. Waiting for legal, compliance or a regulated-claims check is another again. And revision cycles, which feel like work, are largely queue: a round trip through a reviewer's inbox costs days and contains minutes of reading.

Two derived numbers are worth computing while you are there. Review round trips per asset — how many times the work went back for comment. And reviewers per asset — how many distinct people had an opinion to resolve. Both correlate with elapsed time far more strongly than complexity does, and both are policy choices rather than facts about the work.

Review committees are the dominant wait state

Approval is where creative work goes to wait, and the mechanisms are structural rather than personal.

Serial review multiplies queues. When an asset goes to brand, then to product, then to legal, then to the regional lead, each step is a separate queue with its own waiting period. Four sequential reviewers do not cost four times one reviewer's response time; they cost the sum of four full queue waits, each of which is governed by that reviewer's loading. Parallel review — everyone sees it at once, comments are collected on a deadline, conflicts are resolved by a named owner — collapses four queues into one.

Cadence-based forums impose batch delay. A weekly creative review adds an average of half a week to every asset, and a fortnightly one adds an average of a week, regardless of whether the asset needed the forum. For a body of work with a short shelf life, that is a large tax collected indiscriminately.

Undefined authority causes re-review. When it is unclear who decides, work is shown to more people than necessary, decisions get reopened by whoever was not consulted, and assets cycle. An asset that has been approved twice was not approved the first time; it was previewed.

Late review is expensive review. A reviewer who sees the work for the first time when it is nearly finished has to choose between accepting something off-brief and triggering an expensive rework. Both outcomes are bad, and both are caused by the timing of the review rather than by its content. Showing direction early — a route, a concept, a wireframe — is cheap to change and consumes little of the reviewer's time.

The reframing that helps most: review is not an interruption to production. It is a step in production, with capacity, variability and a queue, and it deserves to be designed with the same seriousness as the making.

Brand review is a constraint worth designing

Marketing teams under delivery pressure often argue for weakening brand governance, and this is usually the wrong fight. Brand consistency is an asset with real compounding value, and a function whose job is to protect it will correctly refuse to trade it for a fortnight.

Treat brand review the way you would treat any constraint: increase its effective capacity, reduce the demand placed on it, and reduce the variability of what arrives.

Reduce demand with a real system. Every question a reviewer answers repeatedly is a candidate for a rule. Clear guidelines, a component library, templated layouts and pre-cleared messaging for recurring claims remove work from the queue entirely rather than moving it through faster. This is the most effective intervention available and the slowest to build, which is why it is usually deferred and should not be.

Tier the review. Not every asset requires the same scrutiny. A reprint of an approved asset in a new size, a social post assembled from approved components, and a brand-defining hero film are different risks and should travel different paths. Define the tiers on observable criteria, agree them with brand, and let the lowest tier proceed without review at all. This mirrors the risk tiering used in legal and procurement, for the same reason and with the same effect.

Move the review earlier and make it cheaper. A ten-minute conversation at concept stage prevents a two-week rework at delivery stage. Early alignment is not extra process; it is the removal of the most expensive failure mode.

Reduce variability in what arrives. Incomplete briefs, unclear audiences and unagreed claims cause most review round trips. An enforced brief standard converts a highly variable arrival stream into a predictable one, and by Kingman's formula, reducing variability shortens the queue as surely as adding capacity does.

Protect reviewer time explicitly. A brand lead who reviews in the gaps between meetings is a server with wildly variable availability. Recurring protected blocks convert that into a predictable service time, which shortens queues even if total hours are unchanged.

Batch size in campaigns

Marketing has an unusually strong cultural pull toward large batches. The campaign launch, the annual brand refresh, the full-funnel programme that goes live at once — these are how the function has historically organised itself, how agencies price, and how success has traditionally been presented internally.

Large batches carry the costs the theory predicts. Feedback arrives only at the end, so a mistaken positioning decision is discovered after everything built on it is finished. All the risk sits on a single date. Nothing can be learned and applied within the campaign, only after it. And because the batch is large, its transaction cost — approvals, coordination, launch machinery — is large, which appears to justify the batching in a loop that sustains itself.

Continuous publishing is the alternative. The argument is not that campaigns are wrong, but that the same volume of work released in smaller increments generates feedback while there is still time to act on it.

Approach	Feedback arrives	What a wrong assumption costs	Where risk sits
Single campaign launch	After the launch date	The whole campaign	Concentrated on one date
Phased release	After each phase	One phase	Spread across phases
Continuous publishing	Within days, continuously	One asset	Distributed and small

The practical path is rarely to abandon campaigns. It is to reduce the size of the smallest thing you can put in front of an audience — a route to market that does not require the full launch apparatus. Once something small can go live in days rather than weeks, propositions can be tested before production spend is committed, and the large batch stops being the only available shape.

Work in progress across simultaneous campaigns

The most common structural failure in marketing operations is running too many campaigns at once, and it is almost always the result of an accumulation of individually reasonable decisions.

Each campaign has an internal sponsor, a deadline and a rationale. None of them individually looks unreasonable. Collectively they exceed the capacity of the studio, the review chain and the channel calendar — and, by Little's Law, cycle time equals work in progress divided by throughput, so every additional live campaign slows every other one proportionally. The organisation experiences this as marketing being slow, and responds by asking for more campaigns to be run in parallel, because the ones in flight appear stalled.

The symptoms are diagnostic. Designers and writers splitting across several campaigns and losing time to context switching. Nearly finished assets sitting for weeks because attention moved to the next priority. Deadlines met by reducing quality rather than scope. A studio that is fully utilised and yet visibly failing to finish things.

The intervention is a cap on concurrent campaigns, agreed with the commercial stakeholders who request them, with an explicit sequence for what follows. The conversation is difficult once and then becomes easier, because the alternative currently running is an implicit, invisible, worse prioritisation in which everything is started and the finishing order is set by whoever escalates hardest.

A useful reframing for stakeholders: their campaign is not being delayed by the cap. It is being delayed by the seven other campaigns that were already in flight when they arrived. The cap simply makes the delay visible, finite and schedulable instead of unbounded and unexplained.

Measuring throughput without gaming it

Marketing measurement has a well-known failure mode, and it is Goodhart's Law in its clearest form. Measure assets produced and you will get more, smaller, less valuable assets. Measure campaigns launched and you will get campaigns that launch. In each case the number improves and the business does not, which is why so many marketing leaders have become sceptical of production metrics altogether.

The resolution is to keep flow measures strictly as diagnostics of the production system, held alongside outcome measures, and never to report one without the other.

Flow measures describe the machine. Cycle time by asset type as a distribution, work in progress, flow efficiency, review round trips per asset, and the proportion of assets requiring full review. These tell you whether the production system is healthy. They say nothing whatever about whether the work was worth doing.

Outcome measures describe the result. Whatever your function is genuinely accountable for — pipeline, awareness in a defined segment, retention, conversion. These are the reason the machine exists, and the subject of outcomes over output.

Report them together, as a pair, always. Rising throughput with flat outcomes means you have built a more efficient way of producing things that do not matter. Flat throughput with rising outcomes means you have found leverage and should look for more of it. Either number alone is an invitation to optimise the wrong thing.

Two further protections are worth building in. Never set a production volume target, because a volume target converts a diagnostic into a quota within a single planning cycle. And measure cycle time from brief agreement rather than from studio pickup, so that the queue in front of the studio — usually the largest one — remains visible rather than being defined out of the measurement.

What to do on Monday

Pick ten completed assets of a comparable type from the last quarter. Reconstruct each timeline from agreed brief to live, marking touch and wait. Compute flow efficiency and count review round trips and distinct reviewers per asset.

Take the worst timeline to your next marketing leadership meeting and walk it date by date. Do not editorialise; the dates do that work.

Count live campaigns and divide by campaigns completed per month. That is your expected campaign cycle time by Little's Law. Compare it to what stakeholders are being told.

Convert one serial review chain to parallel this week: everyone sees the work simultaneously, comments close on a stated deadline, a single named person resolves conflicts. Measure the elapsed review time before and after. This change costs nothing and typically produces the largest single reduction available.

Then agree a cap on concurrent campaigns with your commercial stakeholders, with an explicit queue for what comes next. Expect the conversation to be uncomfortable, and expect the first month afterwards to be the fastest the function has moved in a year.