Most brand content fails quietly: it is not disliked, simply not watched, saved or passed on, and in a recommendation-led feed, as you saw in Module 1, content that fails to hold attention is shown to fewer people. So before planning what to post, ask why people share anything at all.

The most cited evidence comes from Berger and Milkman (2012), who analysed every New York Times article published over a three-month period and which ones made the paper's most-emailed list. Positive content was shared more than negative content, but valence was not the main story. Sharing was driven partly by physiological arousal: content evoking high-arousal emotions, whether positive (awe) or negative (anger, anxiety), was more likely to be shared, while content evoking low-arousal emotions such as sadness was shared less. Practically useful content was also shared more. The study concerns news shared by email, not branded TikToks, so treat its findings as a hypothesis for social content, not a rule.

Why people share is as important as what they share. Berger's (2014) review of word-of-mouth research identifies five functions it serves for the sharer: impression management (what sharing says about me), emotion regulation, information acquisition, social bonding and persuading others. In his popular synthesis, Berger (2013) groups the drivers as social currency, triggers, emotion, public visibility, practical value and stories. For a manager, the useful question is which function a piece of content performs for the person who passes it on. A saved routine is practical value; a shared sustainability post may be identity signalling, saying "I am someone who cares about waste".

Spread is not the same as value. Akpinar and Berger (2017) compared emotional and informative ad appeals using sharing data and experiments: emotional appeals were shared more, but informative appeals did more for brand evaluations and purchase intent, and the best performers were emotional content in which the brand was integral to the story rather than bolted on at the end. Their phrase "valuable virality" is the test this module applies. A post that travels while nobody recalls whose it was has bought attention for someone else.

That is why "going viral" is a poor objective. Virality is rare, unpredictable and largely outside a brand's control. What a brand can plan for is content that reliably performs one of the sharing functions for its audience, with the brand embedded in it, published often enough for the occasional outlier to happen. For Moorfield Botanics, the listening insight from Module 2 — "People with reactive skin trust other users' routines more than brand claims" — already suggests which function matters most: practical value delivered by peers, not announcements delivered by the brand.

Without structure, social output drifts towards what is easiest to make: product shots, offers and announcements. A pillar is a recurring theme that a brand's content is organised around, each tied to an audience need and a brand objective. Pillars keep a plan coherent over months, focus idea generation and make performance comparable, because posts in the same pillar can be judged against each other.

Good pillars are derived, not brainstormed. Each should trace back to an insight about what the audience needs (from listening, research or first-party data) and to an objective the brand has set (from the cascade in Module 1). A pillar that answers an audience need but serves no objective is a hobby; one that serves an objective but meets no audience need is advertising that people will scroll past. Three to five pillars is a practical range: fewer gives too little variety, more spreads production thin.

The pillars should also be balanced across jobs. Some content builds the brand (memorable, distinctive, reaching people who are not yet customers), some serves the community (giving existing customers reasons to participate) and some converts (moving interested people to buy). The research suggests the balance matters. Ashley and Tuten (2015), in a content analysis of top brands' social content, found that most of it was functional, while experiential, image and exclusivity appeals were the ones associated with higher engagement; the design was correlational, so it shows association rather than cause. Lee, Hosanagar and Nair (2018), coding over 100,000 Facebook posts from 782 companies, found that content expressing brand personality, such as humour and emotion, was associated with more engagement, while directly informative content such as prices and deals reduced engagement when used alone but increased it when combined with brand-personality content — and deals did drive click-throughs. The lesson is not to avoid conversion content but to stop it crowding out everything else.

Each pillar needs a success measure that matches its job, and none of them should be a follower count. A community pillar might be judged on saves, shares and replies; a brand-building pillar on reach among non-followers; a conversion pillar on tracked visits to product pages. Measures should be set against a baseline, as the strategy module argued, and reviewed: a pillar that consistently under-performs its measure is a candidate for rethinking or retirement.

For Moorfield Botanics, the key insight points directly at one pillar. If people with reactive skin trust other users' routines more than brand claims, then a pillar built on real customers' routines will do more than one built on Moorfield's own product claims — and it connects naturally to the Skin Circle, where members already share those routines.

Platform-native creative is made in the format, length, tone and grammar of a specific platform rather than produced once and cross-posted. Each platform rewards different behaviour: a TikTok is watched full-screen with sound, often by someone who has never heard of the brand; an Instagram carousel is swiped through and saved by someone who wants to come back to it; a YouTube video may be found months later through search. One asset resized for all three is usually second-best on each; an idea can travel, but its execution should be rebuilt for each platform.

On video platforms the first moments carry most weight. The hook is the opening of a piece of content — typically the first one to three seconds of a video — designed to stop the scroll. Teixeira, Wedel and Pieters (2012), tracking viewers' facial expressions, eye movements and decisions to skip internet video ads, found that surprise concentrated attention and joy retained viewers — the level of surprise mattered most for attention, while rising joy kept viewers from skipping — so an emotional payoff saved for the end reaches only those who are still watching. It was a small laboratory study of conventional ads, so take the principle, not a benchmark: earn attention at the start, then give people a reason to stay. Hook devices include showing the result first, naming the viewer's problem in their words, or posing a question the video answers.

Brand prominence needs handling with care. Viewers skip content that looks like an advert early, but content in which the brand never appears does nothing for the brand, as the "valuable virality" argument showed (Akpinar and Berger, 2017). The resolution is to integrate the brand into the story — the product in use within the routine — rather than opening with a logo or closing on a pack shot nobody stays for.

Several other choices make creative native. Sound is central on TikTok and Reels but many people watch without it, so on-screen text and captions carry the meaning either way and make the content accessible. Format follows purpose: short vertical video for discovery, carousels for step-by-step information people save, live sessions for questions and community, and long-form video for depth and search. A lo-fi, creator-style look often reads as more authentic among ordinary users' posts, but polish suits content that needs precision. Finally, social platforms increasingly act as places to search, so the words people actually use (from listening) belong in captions, on-screen text and titles. Format specifications such as video lengths and carousel limits change often; check each platform's current guidance.

These principles apply equally to creators' content, which Module 4 takes up.

User-generated content (UGC) is content about a brand made by its customers: reviews, routines, unboxings. For an audience that trusts peers more than claims, it is the most credible content available. Goh, Heng and Lin (2013), linking a Facebook fan-page community's posts to members' purchases, found that engagement in the community was associated with higher purchase spending, and that both user- and marketer-generated content influenced purchasing through the information and persuasion they carried, in different ways.

UGC has to be encouraged, curated and cleared. Encouragement means giving people a reason and a format, such as a routine-sharing prompt; curation means choosing what to amplify against the pillars. Clearance is the step brands most often skip. A customer owns the copyright in what they post; platform terms license the platform, not the brand. Reusing it therefore needs the person's explicit, recorded permission, covering where, for how long and with what credit; use in paid advertising needs wider permission than an organic repost. And once a brand shares a customer's words, it adopts them: a customer's claim that a product "cured" their eczema becomes a medicinal claim made by the brand. Claims rules for cosmetics and the evidence needed to support them are covered in Module 5.

Generative AI is genuinely useful for ideation, caption variants, versioning one idea into several formats, transcription and alt text. It harms where it substitutes for what the audience values — real people and real experience — or introduces errors no one checks. An AI-generated "customer" with flawless skin is not just inauthentic; it misleads about what a product does.

The UK advertising codes are media-neutral: ads made with AI face the same rules on misleadingness, harm and social responsibility as any other. CAP's guidance on disclosing AI in advertising, published on 29 May 2025, sets no blanket requirement to label AI use; instead it asks whether the audience would be misled without disclosure, and warns that a label cannot cure an ad that is misleading in itself (CAP, 2025). The ASA has ruled against beauty ads whose filters exaggerated a product's effect (influencer posts for tanning and cosmetics brands, 2021); AI imagery raises the same risk. Platforms also have their own labelling rules: TikTok requires creators to label realistic AI-generated content, YouTube has required disclosure of realistic altered or synthetic content since March 2024, and Meta applies an "AI info" label (introduced in July 2024) — all as at the time of writing and changing often, so check current rules. A sound policy states the principle in-house now: AI may help make content, but it may not fake the people, results or evidence the content depends on.

A content plan turns pillars and creative principles into a schedule a team can deliver week after week. Cadence is a resourcing decision before it is a creative one. Every post on the plan needs an idea, production, editing, captions, approval, publishing and someone to reply to the comments it draws. A cadence the team cannot sustain produces rushed, under-performing posts — exactly what recommendation-led feeds punish — so the right question is not "how often should we post?" but "how often can we post content that meets our standard, with the people and budget we have?"

A plan also has to integrate two rhythms. Always-on content runs every week across the pillars; campaign content is concentrated around a launch or moment. A useful lens is the hero, hub and hygiene model popularised by Google's YouTube creator guidance for brands in the mid-2010s (hygiene is sometimes called "help"): hygiene content answers what the audience is searching for and is always available, hub content is the regular programming that brings people back, and hero content is the occasional big moment designed for reach. Most of a calendar is hygiene and hub; heroes are planned around them, not instead of them. Paid amplification of the best organic content belongs to Module 6; the content plan should simply flag which pieces are candidates.

Workflow is what makes the cadence real. A simple workflow names each stage (idea, brief, production, edit, approval, scheduling, community management), who owns it and how long it takes, and works backwards from publishing dates. In a regulated category the approval stage carries real weight: in skincare, any statement about what a product does needs evidence, so the plan must leave time for claims to be checked by the people who hold that evidence, and the creative team needs a list of what can and cannot be said so that most content passes first time.

Test-and-learn replaces opinion with evidence. Choose one variable at a time — the hook style, the format, the length, the posting time — hold the rest as steady as you can, decide in advance what result would change your practice, and run the test long enough to see past the noise of individual posts. Organic results vary a great deal from post to post, so treat a single comparison as a clue rather than proof; the rigorous testing methods in Module 7 show what stronger evidence looks like. Finally, schedule a review: each quarter, compare each pillar against its success measure, and retire or rework what consistently fails. A plan that never drops anything is not being managed.