For anyone who owns content and cannot explain why it is not working. Here is the normal way. We open the accounts. We agree on three posts a week. Someone designs a template. We post the office photos, the quote cards, the product announcements, a greeting for every occasion on the calendar. Three months later we sit in a review and ask why nothing moved.
Nothing moved because nobody wrote down what was supposed to move.
That is not a guess. In the 2026 Content Marketing Institute and MarketingProfs research, 97% of B2B marketers said they have a content strategy. Only 12% said they exceeded their goals. The most-named challenge, at 40%, was creating content that prompts a desired action. Having a strategy and having a strategy that works are two different claims, and a survey only ever measures the first one.
The gap is not talent, and it is not budget. It is that a piece of content moves through nine stages, and most teams run four.
Everything below is attributed. Where a claim comes from a book, the book is named. Where a number comes from a study, the study, its year and its sample are named. Where the popular version of a framework is wrong, the correction is in the same paragraph, because repeating a misquote teaches the misquote. This is the short version of a 47-page playbook. Prefer it as a document?
The Content System.
47 pages: every framework, every checklist, and the format specs for the eight platforms, in one PDF.
Contents
Most teams run four stages out of nine
Idea, make, post, check the numbers. That is stages 3, 4, 6 and 8, and it is what almost every content operation actually runs. The four that get skipped are the brief, the review gate, repurposing, and the learning loop. Those four are the ones that compound, which is why year three of a content programme so often looks like year one with a different template.
| What 1,015 B2B marketers said in 2026 | Share |
|---|---|
| Have a content strategy | 97% |
| Name "content that prompts a desired action" as their top challenge | 40% |
| Exceeded their content goals | 12% |
Read the first row against the third. A strategy is a claim anyone can make on a survey. A working system is a claim you have to show, and the distance between the two is exactly the four stages nobody is auditing.
The four skipped stages are the four that compound.
The nine stages, and what each one owns
Content operations is a real discipline with real published models. The Content Marketing Institute frames it as three pillars, people, processes and technology, with seven components: intake, analysis, create, manage, distribute, repurpose, measure. AIIM, the standards body, has defined the enterprise content lifecycle for decades in five verbs: capture, manage, store, preserve, deliver.
Split create from review, because in practice that is where quality is either enforced or quietly abandoned, and you get nine stages. Each one owns a question, and each one is finished when it has produced an artifact rather than a feeling.
| # | Stage | The question it answers | Output artifact |
|---|---|---|---|
| 1 | Intake | What is being asked for, and why now? | A logged request with a business reason |
| 2 | Brief | One idea, one audience, one action | A signed brief |
| 3 | Ideation | Which angle, and is it the obvious one? | 5 to 10 angles, one chosen |
| 4 | Production | Who makes it, in what format, by when? | A first draft with visuals specified |
| 5 | Review and QA | Does it pass voice, hook, clarity, spec? | An approved file plus a QA report |
| 6 | Publish | Right platform, right time, right link | A live post |
| 7 | Repurpose | What else does this one asset become? | 5 to 10 derivatives, staggered |
| 8 | Measure | Did it move the metric this stage owns? | A number against a benchmark |
| 9 | Learn and archive | What do we keep, kill, and reuse? | A decision, and a filed asset |
The lesson worth taking from AIIM is stage 9. Content teams almost never preserve or retire anything, so year two rebuilds what year one already made.
One gate, stage five, and nothing ships around it.
Two definitions of content, announced as two
People argue about content without defining it, so the argument never ends. Here are two definitions, and the point of announcing them as two is that they are for different rooms.
For the non-marketer: content is anything you publish that earns attention you did not pay for. If you paid a platform to put it in front of someone, that is advertising. Content borrows attention on the strength of the thing itself.
For the practitioner: content is any published asset that moves a person one step along the funnel, from not knowing you exist, to knowing, to believing, to buying, to coming back. Every piece should be able to name the step it owns. A piece that cannot name its step is decoration.
Three terms get used through the rest of this, so they are defined once here. A pillar is a recurring topic the brand owns. A content type is one of the six purposes a post can serve: entertainment, inspiration, education, conversation, connection, promotion. A format is the physical shape of the post: a reel, a carousel, a thread.
Pillar is what. Type is why. Format is how.
Stage 1: every request gets logged before it gets made
The most expensive content is the piece that gets made because someone asked in a corridor. Intake is a small discipline that kills that. Every request enters through one door, gets logged, and answers four questions before anyone opens a design file.
What is the business reason, in one sentence? If the answer is "we have not posted in a while," that is a cadence problem, not a content request.
Which funnel stage does it serve, and therefore which metric does it own?
Is there an existing asset that already does this? Check stage 9 before building anything.
What is the real deadline, and what is driving it?
This is where workflow friction shows up in the data. In the Content Marketing Institute's 2025-outlook research, 33% of B2B marketers named managing workflow and content approval processes as a challenge, down from 41% the year before. It is a solvable problem, and the solution sits upstream: an intake question answered on day one prevents a rewrite on day nine.
Two intake rules worth enforcing. A request with no named owner does not proceed. And a request that arrives inside another team's review cycle joins the next cycle, because a queue that anyone can jump is not a queue.
One door in. Everything else is a favour.
Stage 2: the brief is where the thinking happens
The brief is not paperwork, and the industry has known this for sixty years. Stephen King introduced the T-Plan at J. Walter Thompson in June 1964, and with Stanley Pollitt at BMP founded account planning in 1968. The modern creative brief is what that discipline produced.
The nearest thing to an industry standard is The Client Brief, published in 2003 by the IPA, ISBA, MCCA and PRCA. Its eight headings still beat most templates sold today, and headings two and three are the whole point: where are we now, and where do we want to be.
For a single post that is too much. Use the shorter tool instead. Saatchi and Saatchi's internal device is the OIIC, named in public by Richard Huntington when he was their chief strategy officer in London: objective, issue, insight, creative challenge. Four boxes. Expanded for a content team, it becomes a seven-field page that fits on one screen.
| Field | The test it has to pass |
|---|---|
| Objective | One sentence, one metric, a number if there is one |
| Audience | A person, not a segment. Named behaviour, not demographics |
| The one idea | One sentence. If it takes two, it is two pieces |
| The insight | Something true that the audience has not heard said out loud |
| Type, stage, format | One of six types, one of four stages, format chosen from the job |
| The one action | What exactly should they do, and where does it lead |
| Mandatories, approver | Anything that cannot move, and the one name who signs off |
The insight box is the one that decides everything. And there is a fast test for whether it is real: if the insight box contains a restatement of the objective, it is empty. An insight is an observation about what the audience believes or does, not about what you want to say.
"Customers want better quality" is not an insight. "The buyer asks for a third quote so they can justify, to their boss, a decision they made two weeks ago" is an insight. One of those changes the work. The other is a sentence.
An empty insight box means the brief is not done.
Stage 3: find the angle nobody else would take
Ideas do not arrive, they get mined. Eight veins produce almost everything worth publishing, and the first three are the ones teams underuse: customer language and the questions sales gets asked every week; search, meaning what people actually type into TikTok, YouTube, Pinterest and Google about this category; and wins and proof, the results and the before-and-after with permission.
The other five: behind the scenes, the process and the mistakes and the room. Contrarian takes, the thing your industry agrees on that you do not. Trends and formats, a current shape filled with your own substance. Personal stories, real ones, from real notes. And the audience's own words back to them, from polls, replies and comment questions.
Then push past the first angle, because the obvious lens on a topic is the one twenty other accounts already used. Four moves produce a sharper one: invert the villain, because the obvious villain is rarely the real one. Find the asset hiding in the failure. Jump to the second-order effect nobody is connecting. Switch the point of view to the customer, the loser, or the thing itself.
Two documented processes are worth borrowing here, and both are commonly misattributed. IDEO's own three phases, published in its Field Guide to Human-Centered Design in 2015, are inspiration, ideation, implementation. The five-stage empathise, define, ideate, prototype, test sequence is Stanford d.school's, and the d.school explicitly calls them modes rather than steps, noting they can be used in various orders. For a time-boxed version, the design sprint developed by Jake Knapp at Google in 2010 and brought to GV in 2012 runs five named days, and a content team can run the same week for a new format or a new series.
The obvious angle is taken. Take the second one.
Stage 4: batch the mode, separate the roles
Production fails for two reasons: the team makes one piece at a time, and the same person writes, designs and approves. Both are structural, so fix them structurally.
Batch by mode, not by post. Write all the week's scripts in one session, film all the talking-head pieces in one session, edit in another, design in another. Switching between writing and designing five times a week costs more than any tool saves.
Then separate the roles, even when one person holds several of them, because the roles have different jobs.
| Role | Owns | Hands off to |
|---|---|---|
| Strategist | Objective, platforms, pillars, mix, cadence | Editor |
| Editor | Intake, brief, angle selection, the calendar | Maker |
| Maker | Script, copy, design, edit, visual direction | Reviewer |
| Reviewer | The QA gate at stage 5 | Publisher |
| Publisher | Platform mechanics, timing, link placement, first-hour replies | Strategist, with the numbers |
One rule on visuals keeps briefs honest. Nobody hands over a generated image and calls it a design. Every visual leaves production as two things: a generation prompt, and a human brief with exact dimensions, safe zones, copy to set, typefaces, palette and logo placement. Brand colours and fonts are never invented or approximated. If a value is unknown, it stays a labelled placeholder until someone fills it in.
One person, four hats, never two at once.
Why anything travels, and the study everyone misquotes
Jonah Berger's Contagious (Simon and Schuster, 2013) is the standard reference on why things spread. His framework is STEPPS: social currency, because people talk about things that make them look good. Triggers, meaning top of mind, tip of tongue, so tie the idea to something the audience meets on a schedule. Emotion, because when we care, we share. Public, because when we can see other people doing something, we imitate it. Practical value, news you can use. And stories, because information travels inside narrative.
Two working rules. Never engineer all six, because a piece optimised for everything spreads for nothing. Pick the two or three the topic naturally owns and build them into the beats rather than into extra length.
The academic paper under the book is Berger and Milkman, "What Makes Online Content Viral," Journal of Marketing Research, April 2012, 49(2), pages 192 to 205. Its finding is that content evoking high-arousal emotion, awe or amusement on one side, anger or anxiety on the other, is more viral, and that low-arousal emotion such as sadness is less viral. The mechanism is arousal, not positivity. The popular summary "positive content goes viral" drops the actual finding.
Now the limit, which matters if you are going to cite it in a client deck. The dataset was 6,956 New York Times articles that appeared on the home page between August and November 2008, with making the most-emailed list as the outcome measured. That is email forwarding of news over three months in 2008. It is not a study of TikTok, Instagram, or any feed algorithm, and citing it as one is a stretch the paper does not support.
The 2025 research updates the picture. In PNAS Nexus, Cosme and colleagues ran four studies with 2,559 participants and 18,780 observations across the United States and the Netherlands. Two things predicted sharing, both at p below 0.001: self-relevance, meaning "this says something about me," and social relevance, meaning "this is useful to someone I know." The practical translation is two questions before you publish. Does this let the reader say something about who they are? And can the reader name one specific person who needs it? The second one is where most content fails. "Useful to marketers" is not a person. "Useful to the person on your team who keeps rewriting the same brief" is.
Aim for energy, not for mood.
Why anything is remembered, and why experts are worst at it
Chip and Dan Heath published Made to Stick with Random House in 2007. Where Berger explains transmission, the Heaths explain retention. Their framework is SUCCESs: simple, one core idea said in one line. Unexpected, breaking a pattern the audience assumed. Concrete, sensory and specific. Credible, real checkable detail. Emotional, making them feel the stake for one person. And stories again, because people remember what they can retell.
Two accuracy notes, since this framework is misquoted constantly. The mnemonic the Heaths actually wrote is "a Simple Unexpected Concrete Credentialed Emotional Story," using Credentialed to make the acronym work, while the principle itself is credibility. And the spelling is SUCCESs, with the trailing lowercase s. Six principles, not seven.
The reason all of this is hard has a name and a number. In 1990, Elizabeth Newton ran a study at Stanford for her psychology doctorate. She split people into tappers and listeners. Tappers picked a well-known song and tapped the rhythm on a table. Listeners guessed the song.
| The curse of knowledge, measured | Figure |
|---|---|
| What tappers predicted listeners would guess | 50% |
| What listeners actually guessed, 3 songs out of 120 | 2.5% |
The tappers heard the melody in their heads while they tapped. The listeners heard knocking. That gap is the curse of knowledge, and it is the single most reliable reason expert content underperforms. You cannot un-know your subject, so you cannot hear how your explanation lands.
The target to write against: a sharp 13-year-old follows it on the first pass. Hunt the load markers, which are mechanical and easy to spot. Words over three syllables that have a plain substitute. Sentences over about 25 words. More than one nested clause. Passive constructions that hide who acted. Jargon the audience does not already use. Two cautions, so this does not fight the rest of the craft: do not flatten every sentence to the same short length, because rhythm is what keeps people reading, and do not remove a technical term your audience knows and expects, because that reads as talking down.
Predicted 50, delivered 2.5. Assume you are the tapper.
The hook has one job and exactly four ways to fail
The hook does not sell the piece. It helps the right person decide to keep going, and it does that with two things only: topic clarity within a second or two, and on-target curiosity, meaning the viewer believes this is for them and a question is left open. Every weak hook fails in one of exactly four ways. Diagnose which, then apply the matching fix.
| Killer | What it looks like | The fix |
|---|---|---|
| Delay | The topic arrives after the greeting and the setup | Delete everything before the topic. The first noun phrase carries the subject |
| Confusion | The line needs a second pass to parse | Fewer words, simpler words, active voice, one idea per sentence |
| Irrelevance | Clear, but framed around you instead of the viewer | Swap me and my for you and your. Frame around a pain they already feel |
| Disinterest | Clear and relevant, and still skippable | Name the belief they hold, put your claim against it. The gap is the curiosity |
The deadline changes by platform, the disease does not. A reel has roughly a 2-second scroll cliff, and on TikTok closer to 1.5. A LinkedIn post has the "see more" fold at about 140 characters on mobile. A YouTube video has two hooks: the title and thumbnail earn the click, then the first 15 to 30 seconds confirm it. A carousel has slide one, at nine words or fewer. A newsletter has the subject line.
Banned openers, because the wrapper words never survive a rewrite: "let me explain," "let's talk about," "story time" with no story in the same breath, "nobody talks about this" without naming the thing, "wait till you see this," and any call to action used as an opener. A call to action closes. It never opens.
Clear in two seconds, or it is not a hook.
Stage 5: one gate, and it has no authority
The most useful review model in creative work is Pixar's Braintrust, described by Ed Catmull in Harvard Business Review in 2008. A group of directors watches a film in progress and says what is not working.
Two rules make it function. There are no mandatory notes, and the group has no authority. The director decides what to do with the advice.
That structure matters more than any checklist, because a review board with authority turns into an approval queue, and approval queues produce work that is safe rather than good. This is measurable friction, by the way: in the Content Marketing Institute's 2025-outlook research, 33% of B2B marketers named managing workflow and content approval processes as a challenge, down from 41% the year before.
Then run the fixed checks. They are deliberately mechanical, so judgement gets spent on the work instead of on remembering the list.
Voice. Right sound, right register for this channel.
Hook. Clears delay, confusion, irrelevance and disinterest, inside this platform's deadline.
Clarity. A sharp 13-year-old follows it on the first pass. Flag any sentence over 25 words.
Human check. Strip the machine tells: hollow reframes, inflated significance, category nouns where an instance belongs, adjectives with no number behind them.
Specificity. At least one real number, named case, or lived detail. Nothing invented.
Spec. Ratio, limits, hashtag count, link placement, captions, safe zones, no foreign watermark.
Type and funnel fit. Does it still serve the type and stage the brief assigned, or has it drifted into a sales post?
Conversation layer. Does it end with something a person can reply to?
One action, and it closes rather than opens.
The output of the gate is two things: the corrected file, and a short report marking each check as pass, fixed, or flagged. Fixed means the reviewer changed it. Flagged means only the owner can decide. A reviewer who silently rewrites intent is not reviewing.
Fix what is broken. Flag what is a choice.
Stage 6: the first hour is part of the post
Publishing is not uploading. Three things decide how far a finished piece travels: when it goes out, where the link sits, and whether anyone is there to reply.
On LinkedIn the first 60 to 90 minutes largely set the total reach, because the platform tests the post on the most-engaged part of your network and expands on early meaningful comments. So the posting time is not "when the audience is online," it is "when the author can sit with it for an hour." LinkedIn's own engineering paper, published in February 2026, names the two behaviours its feed ranker optimises: long dwell, meaning time on a post past a threshold, and contribution, meaning a like, comment or share. Both are behaviours a real reader produces and an engagement pod does not.
| Platform | The rule that costs reach when broken |
|---|---|
| Post when you can reply for the first hour. Link in the first comment | |
| X | Link in the first reply, never mid-thread. Space posts out, since a per-author cap means a flood cannibalises itself |
| Design for sends. A direct-message share is the main path to non-followers | |
| TikTok | Keyword the caption, the on-screen text and the spoken words, because search is real there |
| Threads | Reply to your own post to keep it alive. Reach is largely follower-independent |
| Publish seasonal content 30 to 60 days early |
Publish when you can reply, not when you are free.
Stage 7: produce once, publish many
The most common production failure is treating eight platforms as eight separate jobs. Make one deep asset on a rhythm, then derive everything else from it.
Google's own YouTube Creator Playbook for Brands named three tiers. Hero is the push moment, the launch or the tentpole, and a brand may have only a few in a year. Hub is the regular content that gives a fresh angle on the audience's interests. Hygiene is the always-on pull content, the tutorials and answers people are actively searching for. The industry now says Help instead of Hygiene, and Google itself drifted to that name.
One correction worth carrying, because this error is everywhere: the 70/20/10 split does not appear in that playbook, which contains no percentage allocation at all. It is Coca-Cola's, from Content 2020, presented by Jonathan Mildenhall: 70% low-risk proven content, 20% innovating on something that worked, 10% brand-new ideas that could fail. Two useful frameworks, two different companies. "Google's 70-20-10 Hero-Hub-Help model" fuses them.
The operational version is Gary Vaynerchuk's Content Model, published in 2018 with the subtitle "how I make 30+ pieces of content from a single keynote." Pillar content, micro content, distribute both, harvest community insight, make community-driven micro content, distribute again. Treat the 30-plus as his own claim rather than independent research. The shape of the model holds either way. Start by pulling out the atoms.
| Atom | Becomes | Platform |
|---|---|---|
| Best moments | Vertical clips | Reels, TikTok, Shorts |
| Key points or steps | A carousel | Instagram, LinkedIn |
| Core argument | A thread and a question | X, Threads |
| A sharp line | A short post | X, LinkedIn, Threads |
| Quotes, stills, numbers | Pins | |
| The whole piece | An article or newsletter | LinkedIn, blog, email |
And reshape the tone, not only the size. Low-fi and sound-on for TikTok. Polished and captioned for Reels. Professional framing for LinkedIn. A conversational question to open on Threads. Strip any source watermark and rebuild natively, because watermarked cross-posts get suppressed on every platform that detects them. Then stagger the release across days, because dumping ten derivatives in one day competes with itself.
One asset, ten shapes, zero copy-paste.
Stage 8: compare like with like, or do not compare
The most common measurement error in social reporting is comparing two engagement rates that use different denominators. Some vendors divide engagement by followers. Others divide by reach. Both are correct. They answer different questions, per-reach numbers run several times higher, and there is no conversion between them, because the two sets come from different vendors and different customer bases.
| Denominator | What it tells you | Typical 2025 medians |
|---|---|---|
| Per follower | Audience health, and whether your own people respond | TikTok 2.01%, Instagram 0.30%, YouTube 0.21%, X 0.03% · Rival IQ and Quid, 2026 edition, about 2,700 brands |
| Per reach | Content quality, and whether the people served responded | LinkedIn 6.2%, Facebook 5.6%, Instagram 5.46%, TikTok 4.6%, Pinterest 4.0%, Threads 3.6%, X 2.5% · Buffer, 2025 data, 52M+ posts |
State the limit whenever you quote either. Every one of these benchmark sources is a social media tool vendor reporting on its own customer base, not a random sample of any platform. Rival IQ and Quid skew to large brands. Buffer skews to solo creators and small teams. And watch the labels: a report published in 2026 almost always contains 2025 data. Write the data year, not the edition year.
Measure against the stage's one metric first, then against your own last 30 posts, then against an industry benchmark, in that order. Your own trailing median is the only benchmark that controls for your audience.
Your own median beats any industry benchmark.
Stage 9: the reason year two costs the same as year one
Two rituals close the loop, and both are the first thing dropped when a team gets busy.
The first is the postmortem, in Pixar's format: each group lists the top five things they would do again and the top five they would not. The forced count matters. Five and five stops the meeting from becoming either a celebration or a blame session, and it produces a written artifact instead of a feeling.
The second is the archive. Content teams almost never retire or catalogue anything, so the same explainer gets rebuilt every eighteen months by someone who did not know it existed. Three shelves is enough: evergreen assets that can be re-cut with a refresh, reference assets that inform new work without being republished, and retired assets that are wrong now and should stop circulating.
| The monthly review, in four questions | What you do with the answer |
|---|---|
| Which pillar and which content type over-performed? | Shift next month's mix toward it, by a few points, not a rewrite |
| Which type sat at zero? | Schedule two of them next month |
| Which hero asset produced the most derivatives that worked? | Make more of that hero shape |
| What did we rebuild that already existed? | It goes in the archive index today |
One caution against over-engineering this. Netflix's own engineering team describes its productions as snowflake workflows, noting that no two productions share the same workflow, and organises around functions rather than one rigid pipeline. A lifecycle is a checklist for what must happen, not a claim that every piece takes the same path.
Five to repeat, five to never. Written down.
Nearly everyone uses AI. Almost nobody reports better performance
In the Content Marketing Institute's 2026 research, 95% of B2B marketers said their organisations use AI-powered applications and 89% use content-creation tools. Among those using content-creation AI, 87% reported better productivity and 80% better operational efficiency. Then the number that matters.
| Among marketers using content-creation AI | Share |
|---|---|
| Say it improved productivity | 87% |
| Say it improved content performance | 39% |
| Saw no change in performance | 34% |
| Say content quality decreased | 12% |
The discovery side agrees. Graphite classified 65,000 article URLs sampled from Common Crawl and found that as of May 2025 the sample was about 52% human-written and 48% AI-generated, yet among top-ranking Google articles 86% were human-authored, and among ChatGPT and Perplexity citations 82% were. Ahrefs, looking at 900,000 newly created pages in April 2025, found the honest version of the saturation story: pure AI pages were 2.5%, pure human 25.8%, and mixed 71.7%. AI assistance is near-universal. AI authorship is rare, and it does not rank.
Both studies rest on imperfect AI detectors and show correlation rather than cause, so hold them loosely. The working rule they support is not loose at all. Use AI where the lifecycle is mechanical: intake logging, transcription, first-pass repurposing, variant generation, spec checking, tagging the archive. Keep the human where the value is: the insight in the brief, the lens, the real number, the story only this company can tell, and the final read.
Automate the pipeline. Never automate the point.
What this is actually worth
The same CMI research that produced the 97% and the 12% also asked teams whose results improved what drove it. 74% pointed to refining the strategy, ahead of 51% who pointed to new technology and 40% who pointed to restructuring the team.
| What teams with improved results credited | Share |
|---|---|
| Refining the strategy | 74% |
| New technology | 51% |
| Restructuring the team | 40% |
The honest limit on that: it is a self-reported survey of 1,015 marketers, sponsored by a software vendor, and it shows correlation rather than proof. I might be wrong about how large the effect is. I am not wrong about the direction, because the four skipped stages cost nothing to add and every one of them removes rework later.
None of the nine stages requires budget. The brief is one page. The review gate is a checklist and a person who is allowed to say what is not working. Repurposing is a table. The archive is a folder with three shelves in it.
If you want the same logic run on your marketing function rather than your content lifecycle, the assessment below takes about five minutes and returns a maturity stage and the three things to fix first.
None of the nine stages requires budget.
Stand the system up in four weeks
Do not try to run all nine stages from day one. Stand them up in this order, because each week produces the input the next week needs.
| Week | What you do | What exists at the end of it |
|---|---|---|
| 1 | Rank the goal. Assign one metric per funnel stage. Cut platforms without a role. Name 3 to 5 pillars | A one-page strategy anyone on the team can recite |
| 2 | Choose the hero asset and its rhythm. Write the one-page brief template. Mine 20 to 30 angles from the eight veins | A brief template and an idea bank |
| 3 | Build the calendar to your real capacity, with standard columns and a five-status ladder. Produce the first hero and batch its derivatives | Four weeks scheduled, one hero made |
| 4 | Run the QA gate on everything before it ships. Publish with the reply window covered. Set the monthly review date | A live cadence and a review in the diary |
Three things to resist in month one. Do not add a platform, because the system has not been tested on the ones you have. Do not raise the cadence, because week 4 is when the real capacity shows itself. And do not skip the QA gate to hit a posting date, because the first time it gets skipped is the moment it stops being a gate.
By day 30 you should be able to answer four questions without opening a file: what are we here for, what do we talk about, how often do we publish, and who says yes. If any of the four takes more than a sentence, the strategy page is not finished yet.
Four weeks to a system. Then hold it.
Frequently asked questions
What are the nine stages of a content process?
Intake, brief, ideation, production, review and QA, publish, repurpose, measure, and learn and archive. The Content Marketing Institute frames content operations as three pillars, people, processes and technology, with seven components. This version splits create from review, because the review gate is where quality is either enforced or quietly abandoned. Most teams run only four of the nine: ideation, production, publish, and measure.
Why do most content teams fail even when they have a content strategy?
Because having a strategy and having a strategy that works are two different claims. In the 2026 Content Marketing Institute and MarketingProfs research, 97% of B2B marketers said they have a content strategy and only 12% said they exceeded their goals, with 40% naming content that prompts a desired action as their hardest challenge. The gap is process: the four stages teams skip are the brief, the review gate, repurposing, and the learning loop.
What should a one-page content brief contain?
Seven fields: objective in one sentence with one metric, an audience described as a person rather than a segment, the one idea, the insight, the content type and funnel stage and format, the one action, and the mandatories plus the single approver. The short form is Saatchi and Saatchi's OIIC, named in public by Richard Huntington: objective, issue, insight, creative challenge. If the insight box restates the objective, it is empty.
Is the 70/20/10 rule part of Google's Hero-Hub-Help model?
No. The YouTube Creator Playbook for Brands names Hero, Hub and Hygiene, now commonly called Help, and it contains no percentage allocation at all. The 70/20/10 split is Coca-Cola's, from Content 2020, presented by Jonathan Mildenhall: 70% low-risk proven content, 20% innovating on something that worked, 10% brand-new ideas that could fail. Two useful frameworks from two different companies, routinely fused into one.
Does AI improve content performance?
It improves throughput more than results. In the Content Marketing Institute's 2026 research, 95% of B2B marketers said their organisations use AI applications and 87% of those using content-creation AI reported better productivity, but only 39% said content performance improved. Graphite's classification of 65,000 article URLs found that among top-ranking Google articles 86% were human-authored. Use AI where the lifecycle is mechanical and keep the human on the insight, the lens, and the final read.
Sources, method and limits
Method. Every framework above is a published one, cited to its own source and edition rather than to a secondary summary. Where the popular version of a framework differs from the original, both are printed and the difference is named. Platform specs and benchmark figures were verified in July 2026.
Three limits, stated plainly. One, every social benchmark quoted here comes from a tool vendor reporting on its own customers, not a random sample of any platform, so treat all of them as directional. Two, a report published in 2026 almost always contains 2025 data; the data year is what matters, not the edition year. Three, the AI-content studies rely on imperfect detectors and show correlation rather than cause. Platform specs will move again.
Books and research. Jonah Berger, Contagious, Simon and Schuster, 2013. Berger and Milkman, "What Makes Online Content Viral," Journal of Marketing Research, 2012, 49(2), 192 to 205. Chip and Dan Heath, Made to Stick, Random House, 2007, including Elizabeth Newton's 1990 Stanford doctoral study. Cosme and colleagues, PNAS Nexus, 2025, 4(2) pgaf019. Marcus Sheridan, They Ask, You Answer, Wiley, 2017. Donald Miller, Building a StoryBrand, 2017. Nir Eyal, Hooked, 2014. Seth Godin, Purple Cow, 2003. Robert Cialdini, Influence, 1984, expanded 2021. Malcolm Gladwell, The Tipping Point, 2000. Joe Pulizzi, Content Inc., 2015 and 2021.
Process and agency sources. The Client Brief, IPA, ISBA, MCCA and PRCA, 2003. Richard Huntington on the Saatchi and Saatchi OIIC, 2008. The 4A's record of Stephen King's T-Plan at J. Walter Thompson, 1964. IDEO.org, The Field Guide to Human-Centered Design, 2015. Stanford d.school design thinking process guide. GV, The Design Sprint. Ed Catmull, "How Pixar Fosters Collective Creativity," Harvard Business Review, September 2008. BuzzFeed on Pound, 2015. The Washington Post on Bandito, 2016. Netflix Technology Blog on production media management. Content Marketing Institute on content operations, 2026. AIIM on the content lifecycle.
Strategy frameworks and platform data. Avinash Kaushik, See Think Do and See Think Do Care, Occam's Razor, 2013 onward. HubSpot Research, Mimi An, "Topic Clusters," 2017. Smart Insights, Chaffey and Bosomworth, The Content Marketing Matrix, 2012. Ehrenberg-Bass Institute, John Dawes, the 95-5 rule. Google, The YouTube Creator Playbook for Brands, on Hero, Hub and Hygiene. The Coca-Cola Company, Content 2020, 2011, on the 70/20/10 split. Gary Vaynerchuk, The GaryVee Content Model, 2018. Content Marketing Institute and MarketingProfs B2B research, 2024, 2025 and 2026 editions, with the 2026 edition fielded June to August 2025, n=1,015, and the 2025 edition n=980. Rival IQ and Quid, 2026 Social Media Industry Benchmark Report. Buffer, The State of Social Media Engagement 2026, 2025 data, 52M+ posts. Socialinsider LinkedIn benchmarks. Metricool 2026 studies of Instagram, LinkedIn, and social media overall. Adam Mosseri on Instagram ranking signals, January 2025. LinkedIn engineering, arXiv 2602.12354, February 2026. YouTube Help, TikTok Transparency Center, and Meta on labelling AI content. Graphite via Axios, 2025. Ahrefs, 2025.
Want the playbook with every checklist?
47 pages: the strategy layer, the craft layer, all nine stages, and the specs for eight platforms, in one PDF.
About BMD
Building Your Marketing Department (BMD) helps executives in Saudi Arabia and the Gulf build marketing departments that produce measurable return, using the BUILD framework. Redha Alayesh has built 3 marketing departments and worked with more than 40. Clients and collaborations have included Alyaseen Agri, Ithra, IBM and the Saudi Esports Federation. Delivered in Arabic and English, founder-led.
Redha Alayesh
A marketer with a software engineer's discipline and a scientist's mindset. Across 20+ organizations in the GCC, he built the BUILD framework to solve the problem he kept finding: capable marketers trapped inside companies that never decided what marketing was for.
Disclosure, since this argues for a system that a consultancy sells: every framework here is public, attributed, and usable without hiring anyone. The lifecycle in this article comes from the agencies and media houses that published it, and from watching which stage a marketing team quietly drops first when it gets busy. It is almost always the brief, and then the review gate.