Content Production Workflow: Stages, Gates and Metrics

A content production workflow is the defined sequence through which a piece of content passes from approved idea to published, measured asset, together with the criteria that must be met before it advances. The most durable designs share three properties: a small number of stages, an explicit exit condition for each stage, and a record of every state change. This article sets out those stages, examines the handoffs where quality is most often lost, and shows how production can be treated as a system that is measured rather than merely managed. The reasoning applies to written content and to video production alike, since both fail at the same junctures.

TL;DR

• A workable pipeline has seven stages (intake, briefing, drafting, editing, build, publishing, measurement), and each has an exit gate that must be satisfied before the asset moves on (Gatilab).
• Smaller teams may combine roles, but combining gates is the shortcut that produces rework.
• Handoffs, including those to subject-matter experts, designers and AI tools, are the principal failure point and require explicit human review.
• Latency, traffic, errors and saturation, borrowed from site reliability engineering, give a content team measurable indicators of throughput and capacity.
• In regulated settings, approvals must be logged natively at the point of action, not reconstructed from email.

Table of Contents

• What the workflow is, and what it is not
• The seven stages
• Gates and roles are different things
• Handoffs: where quality is lost
• Audit trails for regulated environments
• Observability for content operations
• Applying the workflow to video production
• Our perspective
• Where PROJECT-JTH fits
• FAQ
• Sources

What the workflow is, and what it is not

Search results for this topic tend to describe the workflow as a list of tasks. A more useful definition treats it as a set of commitments between stakeholders: who must supply what, in what condition, by when, before the next party begins. Under that definition, a workflow is not a project plan (which is specific to one deliverable) nor a content calendar (which specifies what appears when). It is the reusable structure beneath both.

The distinction matters because most organizations that report a workflow problem have, in fact, a handoff problem. Tasks are identified, owners are named, and yet assets stall in review or arrive at publication in a condition that requires correction. The remedy is rarely a longer task list. It is a clearer statement of what “done” means at each boundary.

The seven stages

The Gatilab guide to content operations describes a pipeline of seven stages, with one owner for each: intake and scoring, briefing, creation, editing, build and formatting, publishing and distribution, and measurement and refresh. The following table adapts that structure and adds an illustrative exit gate for each stage. The gates are examples of the kind of criterion that works, not a prescribed standard.

Stage 1: intake and scoring

The first gate is the one most frequently skipped. Guidance from Canva on content workflows reports that marketers cite difficulty with meeting search intent (70.6%) and getting content to rank (77.6%), and the common thread is production that begins before the topic has been checked against evidence. A brief intake form that requires a stated audience, a demand signal and a business purpose will remove a substantial share of redundant work before any hours are spent.

For video, the equivalent is a concept review: whether the film has a defined viewer, a defined action the viewer should take afterward, and a realistic scope. Reshoots are expensive, and a large proportion of them originate in an unexamined concept rather than in poor camera work.

Stage 2: briefing

The brief is the contract between the requester and the producer. It should state the audience, the question the content answers, the format, the required sources, the approvers and the due date. For video, the brief extends to locations, talent, permissions, deliverable formats and, where the subject is a regulated workplace, the safety constraints on filming. PROJECT-JTH has written on how industrial video briefs can reduce reshoots, which illustrates the point: a precise brief is cheaper than a corrected edit.

Stage 3: creation

Creation is the stage on which attention is usually concentrated and the one at which the least structural benefit is found. The relevant design question is whether drafting must be strictly sequential. Research on collaborative publishing describes the “unitary author” model as increasingly obsolete, noting that concurrent editorial processes, in which drafting, review and design overlap under a facilitator, can produce a finished work in a compressed period. The qualification is important: concurrency requires someone to maintain focus and quality, and without that role overlap becomes confusion.

Stage 4: editing

Editing should be understood as several distinct checks that are sometimes performed by one person: structural review, factual verification, voice and style, and compliance. Naming them separately allows a small team to assign them to a single editor without losing sight of the fact that each is a different test.

Stage 5: build and formatting

Build covers the work of putting the asset into its final form: layout, metadata, captions, transcripts, alt text and tagging. Current industry expectations place weight on accessibility (for example conformance to WCAG 2.2 AA) and on structured metadata that makes content legible to automated systems, as reflected in recent senior production-standards roles at large educational publishers. Treating these as a gate, rather than as a late correction, is considerably less costly.

Stage 6: publishing and distribution

Publication should be a checkable event. The gate is simple: the live page or file is opened and verified, the links resolve, and the distribution tasks (newsletter, social, internal notification) are recorded as complete or deliberately omitted.

Stage 7: measurement and refresh

The final stage returns information to the first. A piece is compared with the expectation recorded at intake, and a review date is assigned. Without this stage the workflow is a line; with it, the workflow becomes a cycle in which the intake criteria improve over time.

Gates and roles are different things

A frequent misreading of staged workflows holds that each stage requires a separate person. The Gatilab analysis distinguishes the two cleanly: collapsing roles is an acceptable shortcut for smaller teams, whereas collapsing gates leads to quality degradation and costly rework. A freelancer who is simultaneously strategist, writer, editor and publisher still benefits from performing the stages in order and checking the exit criteria, even if the checks are conducted on separate days to create some distance from the work.

This has a practical implication for team design. When capacity is constrained, the question to ask is not which stage can be eliminated but which roles can be combined while the gates remain intact. A common arrangement for a small agency assigns intake and briefing to one person, creation and build to another, and editing and publishing to a third, with the gates documented in a shared checklist.

Handoffs: where quality is lost

Each transition between stages, and each transition between people or tools, is an opportunity for information to be dropped. Analysis of AI-assisted production is explicit on the point: the critical failure is the handoff, and an effective workflow must enforce human expertise injection and formal QA after generation, including fact-checking and plagiarism detection, to prevent generic or unreliable output.

Three controls address most handoff failures:

1. A written acceptance condition. The receiving party confirms that the incoming asset meets the stated gate, or returns it with a reason.
2. A single named owner per stage. Shared ownership of a stage produces delays that no one is positioned to resolve.
3. Comments held in one place. When feedback is dispersed across chat, email and documents, reconciliation becomes a task in its own right. Distributed teams that coordinate mainly through chat commonly report precisely this difficulty.

For video, the same controls apply to the transition from production to post-production (a shot log and clean media transfer), from rough cut to picture lock (a single consolidated round of feedback), and from picture lock to delivery (a checklist of formats, captions and rights clearances).

Audit trails for regulated environments

Where content touches healthcare, government or contractual obligations, the workflow acquires a further requirement: the ability to demonstrate what happened. Guidance on content-management governance states that reconstructing approvals from email threads or file metadata is insufficient for regulatory audits, and recommends that every state change (create, edit, approve, reject, publish, archive) be logged automatically with a timestamped user identity at the point of action.

Two practical consequences follow. First, the system of record should be chosen for its logging capability before it is chosen for its interface. Second, organizations subject to HIPAA or records-retention obligations should confirm the specific retention periods and logging requirements with their compliance counsel, since the content workflow will inherit them. A workflow that is merely convenient will not satisfy an auditor.

The same logic applies to production records that carry commercial weight: releases, licenses, usage terms and client approvals. PROJECT-JTH has addressed this in its material on designing a contract management workflow as a pilot and on ten clauses that belong in video production contracts.

Observability for content operations

Readers who arrive from operations and engineering will recognize the underlying problem: a system that cannot be observed cannot be improved with confidence. The monitoring guidance in Google’s Site Reliability Engineering book identifies four golden signals: latency, traffic, errors and saturation. Translated to content production, they yield a compact set of indicators.

The analogy is a clarification, not a claim of equivalence: content quality is not binary in the way a failed request is. Nevertheless, the discipline transfers. Recording the date at which each asset enters and exits a stage is sufficient to compute latency per gate, and the gate with the longest queue is, in most cases, the constraint to address first. Return rates at editing indicate whether briefs are adequate. Work in progress per person indicates whether the apparent delay is a capacity problem or a prioritization problem.

A minimal observability roadmap

A team need not instrument everything at once. A sequence that has proven manageable proceeds as follows.

1. Record state changes. Establish a single status field with the seven stage names, and require that it be updated at each gate.
2. Timestamp transitions. Capture the date of each change automatically where possible, manually where not.
3. Review weekly. Examine the oldest items at each stage and the number returned from editing.
4. Set thresholds. Once a baseline exists, define an acceptable wait time per gate and flag exceptions.
5. Retain the log. Keep the history for audit and for retrospective analysis of why particular assets stalled.

The log produced by step one is also the foundation of a logging strategy in the regulatory sense described above; a single well-designed record serves both operational and compliance purposes.

Applying the workflow to video production

The seven stages map directly onto video work, with differences of vocabulary and cost. Intake corresponds to concept approval; briefing to treatment, script and shot list; creation to principal photography; editing to the rough cut and revisions; build to color, audio mix, captions and export; publishing to delivery and hosting; measurement to viewer retention and the achievement of the stated objective.

The principal difference is the cost of late changes. A sentence can be rewritten at negligible expense, whereas a location cannot be re-entered without rebooking crews, talent and permissions. For this reason the earlier gates (concept, brief, shot list) deserve proportionally more rigor in video than in text, and feedback after rough cut should be consolidated into a single round with one designated approver. Where the setting is an industrial or regulated site, safety requirements form an additional gate before the camera is turned on.

Our perspective

We build and use production workflows across three kinds of work: technical consulting, video and photography, and software for production records. Our consistent observation is that teams rarely lack effort; they lack legibility. When an engagement, a shoot or a contract stalls, the cause is almost always that no one can see which gate an item is waiting at, or what the gate requires.

This is why we regard the workflow as an operational design problem first and a tooling problem second. A clear stage definition, a named owner and a consistent record will outperform an elaborate platform that no one maintains. It is also why we approach leadership under pressure as a matter of structure: calm decisions are easier when the system displays its own state, and difficult when the state exists only in a few people’s memory.

We would add one caution. Observability metrics are instruments, not objectives. A team that optimizes latency at the expense of the editing gate will have shorter cycle times and worse content. The signals should prompt questions; they should not replace editorial judgment.

— Jesse Hart

Where PROJECT-JTH fits

For teams whose production output carries contracts, licenses and records, Project Relay is built to manage production records and contract management in one place, at $49 per month. Independent consultants who deliver assessments and reports can look at Project Praxis, priced at $99 per month. Organizations that need an outside review of the systems beneath their workflow, including risk review and infrastructure automation, can read about our technical consulting engagements, and those who need video or photography produced to a defined brief can start from the main site.

FAQ

What are the steps in content production?

A standard sequence has seven steps: intake and scoring, briefing, creation, editing, build and formatting, publishing and distribution, and measurement and refresh. Other guides compress this into four, five or six steps, and the differences are largely a matter of grouping. What stays constant is the presence of an exit criterion at each boundary.

What is the difference between a workflow and a process?

In practice the terms overlap, but a useful distinction is that the process describes what is done, whereas the workflow also specifies who hands what to whom, under what conditions and in what order. The workflow is therefore the more operational of the two, and it is the one that can be measured.

How should a small team or freelancer adapt it?

Combine roles, not gates. One person can perform several stages, but each exit check should still occur, ideally after an interval that allows a fresh reading. Research on staged pipelines treats this as an acceptable compromise, provided the gates remain.

What are the 5 C’s of content?

The framework lists clarity, consistency, creativity, credibility and customer-centricity. Each can be assigned to a stage: clarity and customer-centricity at briefing, credibility at editing, and consistency at build. A summary of the framework appears in the Powerful Marketers workflow guide.

Do AI tools change the workflow?

They change the creation stage but not the need for gates. Output generated by AI should pass through human expertise injection and a QA gate for fact-checking and originality, as set out in analysis of AI-assisted production.

Sources

• Content Operations: Stages, Gates & the Minimal Stack, Gatilab
• The Content Strategist’s AI-Assisted Production Workflow, Phrasly
• How to Eliminate Content Bottlenecks Without Sacrificing Security, dotCMS
• Monitoring Distributed Systems, Google SRE Book
• A Book Sprint as a concurrent editorial process, Knowledge Futures
• Content workflows, Canva
• Content Creation Workflow Guide 2025, Powerful Marketers
• Production, Publishing and Content Standards Leader, Taskium

Recommended

• Cut Reshoots in Industrial Video Production with OSHA-Aligned Advice
• Teach One Behavior in 5 Minutes: Training Video Production for L&D
• Publish a Standard Operating Procedure in One Afternoon for Teams