Social media marketing with a reason to post

A channel with no stated job will drift

Accounts that go nowhere tend to share a single root cause, and it is not the content. It is that nobody ever wrote down what the channel was supposed to change. “Build brand awareness” is not a job; it is a category of jobs, and each one inside that category implies a different platform, a different format, a different cadence and a different definition of a good month.

Settle it and the rest becomes tractable. If the job is pipeline from a specific buyer group, a founder posting in their own name on LinkedIn will usually outperform a company page publishing three times a week, and the measure is conversations started. If the job is credibility ahead of a long, considered purchase, the work is depth published consistently, and it will look slow for a while before it looks obvious. If the job is recruitment — which, for a surprising number of businesses, turns out to be the honest answer once you ask what social has ever actually produced — the content is about how you work rather than what you sell, and the measure is applications.

Our social media marketing services start at that question rather than at a calendar, because platform choice, content pillars and reporting are all downstream of it. Answer it properly and they largely answer themselves.

Choosing platforms by subtraction

The most valuable output of a channel strategy is the list of platforms you stop using. Every additional channel consumes native content, community management, reporting and a share of everyone’s attention, and those costs come directly out of the channels that were working.

Audience data you can check for yourself

Pew Research Center publishes one of the few platform-usage datasets that is public, methodologically transparent and regularly refreshed. Its 2025 survey of US adults found YouTube (84%) and Facebook (71%) reaching the broadest adult audience, Instagram at 50%, TikTok at 32% and LinkedIn at 25%. The age distribution matters more than the headline: 80% of 18-to-29-year-olds use Instagram against 19% of over-65s, and TikTok splits 63% against 12%.

That is US data, and it is directional rather than decisive for a UK or European audience. But it settles several arguments cheaply. If your buyer is a facilities director in their fifties, a TikTok-first plan needs a better justification than the observation that everybody is on TikTok. If your customer is twenty-four, a Facebook-first plan needs exactly the same scrutiny.

What each platform is actually good at

Platforms are not interchangeable surfaces for the same content. LinkedIn works because professional identity is declared, which makes both targeting and relevance possible; it rewards specificity and punishes press-release tone. Instagram rewards craft and consistency, and its saves and shares tell you far more than its likes do. TikTok and YouTube Shorts distribute heavily to people who do not follow you, which makes them the best available test of whether a message stands up without the benefit of an existing relationship. YouTube proper is the only mainstream platform where someone will voluntarily give you twelve minutes, which makes it the natural home for explanation. Pinterest behaves more like a visual search engine than a social network, and rewards evergreen content that has been titled and described properly. X is now useful mainly for real-time and industry conversation rather than reach. Reddit and specialist forums are where your category gets discussed candidly, and they are usually more valuable to read than to post in.

A presence should follow from one of those properties matching your commercial job. Anything else is an account you maintain because closing it feels like an admission.

Listening is the cheapest research in marketing

Before publishing anything, it is worth spending a period simply reading: your own comments and direct messages, competitor accounts and the replies underneath them, review sites, category forums, and the questions your sales team answers repeatedly on calls.

This produces two things that are otherwise expensive to obtain. The first is language — the actual words customers use for the problem, which are almost never the words in your marketing. The second is a supply of content ideas that are answers to questions somebody genuinely asked, which is a far better filter than a brainstorm. Listening also tends to surface the objection your sales team handles every week and your website has never addressed, and that finding alone often justifies the exercise.

Structured listening then continues after launch, because it is the earliest warning you get that a message has stopped working or that a complaint pattern is forming.

Pillars before calendars

A content calendar assembled from ideas is a treadmill. A calendar assembled from pillars is a system.

Pillars are three to five recurring themes, each attached to a commercial job rather than to a topic. “Answering the questions buyers ask before they buy” is a pillar. “Behind the scenes” is not, unless somebody can say what it is for. Each pillar carries example posts, the formats it suits, and a note on the evidence it needs, so that a writer who joins in six months produces something recognisably yours.

Pillars also make the awkward conversation easier. When a stakeholder asks for a post about an internal award, the question is not whether it is interesting — it is which pillar it serves. Usually none, and the calendar survives intact.

Where a social programme sits alongside a broader publishing effort, the pillars should be the same ones. Social and content marketing pulling in different directions produces twice the work and half the coherence.

Production is the constraint nobody budgets for

Strategy documents rarely fail. Production schedules fail constantly, and almost always for the same reason: the calendar was designed against ambition rather than against capacity.

Original short-form video is the clearest example. It is currently the most heavily distributed format on nearly every platform, which makes it tempting, and it is also the most expensive thing on the calendar once you count filming, editing, captioning, versioning per platform and approvals. A plan that assumes four original videos a week from a team with no in-house editor is not a plan; it is a deficit that will be discovered in month two.

The workable approach is a mixed calendar: a small number of genuinely produced pieces, a larger volume of derivative formats cut from the same source material, and a repeatable template set for the rest. One filmed conversation can yield a long-form upload, several vertical clips, a carousel and a set of quotation graphics — provided that was the plan when it was filmed, rather than a salvage operation afterwards.

Templates are what stop the derivative work from looking derivative. Building them properly is a design system and brand asset job, not a favour asked of whoever happens to own the brand file.

Video that works with the sound off

Most social video is watched muted, at least at first, which makes captions a performance requirement before it is an accessibility requirement. Conveniently, the two point in the same direction: burned-in or platform captions, adequate contrast, no critical information carried by audio alone, and a transcript where the content deserves one. The W3C’s guidance on accessible audio and video sets out what adequate actually means, and it is a lower bar than most teams assume.

The first two seconds do the work. A hook that states a specific problem — not a teaser, not a countdown, not an instruction to wait for it — will hold the right people and lose the wrong ones, which is the correct outcome for a business account. Retention graphs are the most useful diagnostic any social platform gives you, because they show precisely where interest died.

Captions and copy

Write per platform. The same paragraph pasted across LinkedIn, Instagram and X reads as native to none of them. Front-load the point, because every platform truncates: assume the first line is the whole message and the rest is for people already interested. Hashtags are a discovery mechanism with sharply diminishing returns, and a small number of specific ones beats a wall of generic ones. Links in captions are throttled on some platforms and inert on others, so the route back to your site should be designed deliberately rather than by pasting a URL and hoping.

Publishing, approvals and the unglamorous machinery

The difference between a programme that ships and one that stalls is usually the approval route, not the creative.

Name the approver. One person, with a deputy, and a stated turnaround. Where legal or compliance review is genuinely required, define what triggers it, so routine posts are not queued behind the exceptional ones. Keep a documented tone-and-claims boundary so that most content never needs escalation at all: the approvals that slow teams down are mostly re-litigating decisions nobody ever wrote down.

Scheduling tools — Meta Business Suite, Buffer, Hootsuite, Sprout Social — are close to commodities, and the right one is whichever your team will actually open. What matters far more is that somebody can pause the entire queue within minutes, from a phone, without asking permission. That capability costs nothing and prevents the single most common brand-safety failure: a cheerful scheduled post publishing into the middle of an incident.

Community management is a commitment, not a vibe

An unanswered comment under a good post is a worse outcome than not posting. The reply section is where a social presence is actually judged, and it is the part most often left to nobody in particular.

Community management done seriously means a stated response window that you meet, an escalation route for anything that turns into a complaint or a legal question, an approved answer bank for the questions that recur, and a written policy on what gets hidden. Hiding abuse and spam is normal and correct. Deleting legitimate criticism is not, and it is almost always screenshotted before it disappears.

The commercial argument is simpler than the reputational one: a meaningful share of direct messages to a business account are sales enquiries in an unfamiliar wrapper. If nobody is reading them within a working day, the channel is losing revenue quietly enough that it will never appear in a report.

Paid amplification, and when boosting is the wrong tool

Organic performance is a cheap test of a message. Paid distribution is how a message that has already proved itself reaches people who have never heard of you. Doing those two things in the wrong order is how budget disappears without a trace.

The boost button is a narrow tool: extra distribution for an existing post, with minimal control over objective, placement, audience or optimisation event. It is defensible for a genuinely time-sensitive announcement. It is not defensible as an entire paid strategy, because you are paying advertising rates for publicity outcomes and receiving reporting you cannot act on.

Structured campaigns in Meta Ads Manager or LinkedIn Campaign Manager give you conversion objectives, audience construction, creative testing and attribution you can interrogate. They also depend on measurement being set up properly, which on Meta now means the Pixel and the Conversions API running together with events deduplicated correctly — browser-only signal has been degrading for years, and pretending otherwise produces campaigns that optimise against a partial picture. Once budget becomes material this is a discipline in its own right, and paid media management is where it belongs.

Creators, employees and founders

Three sources of distribution sit outside your own accounts, and all three are more credible than the brand page.

Creators and influencers work when they are briefed on the problem rather than handed the copy. A brief that specifies the audience, the claim that must survive, the claims that must not be made, and the disclosure requirement — then leaves execution alone — outperforms a script every time. Selection matters more than follower count: a small, specific audience that trusts someone beats a large one that scrolls past them.

Disclosure is not optional and it is not a formality. In the UK, paid creator content falls under the CAP Code, and the ASA and CAP influencers’ guide is explicit that the label must be upfront and unmissable. The FTC’s guidance for US audiences says the same thing in different words: disclose the relationship, place it with the endorsement rather than behind a “more” link or in a video description, and do not rely on a platform’s built-in tool alone. Vague abbreviations do not qualify. We treat this as a hard requirement, partly because it is the law and partly because an undisclosed ad, once identified, costs the brand more than the campaign ever earned it.

Employee advocacy works when it is genuinely optional and supported with material people are willing to attach their own name to. It fails the moment it becomes a rota. Give people assets, a reason and permission; do not measure them on shares.

Founder-led content is frequently the highest-return social activity available to a smaller business, because a person can say things an organisation cannot. It also requires a real, recurring time commitment from someone whose time is expensive, so it is worth being honest about whether that commitment exists before building a strategy that depends on it.

Measuring social without flattering it

Reach, impressions and follower count describe activity. They are worth watching as diagnostics and worthless as objectives, because all three can be bought and none of them has ever paid an invoice.

The measurable chain starts with tagging. Every outbound link carries a consistent UTM convention, documented so it survives staff changes, because Analytics classifies traffic into Organic Social and Paid Social from source and medium values — get the convention wrong and your best-performing channel reports as direct traffic indefinitely. From there the questions are all about what happened after the click: enquiries, qualified leads, assisted conversions, and whether branded search demand is moving.

Two honest limitations belong in every report. First, a large share of social sharing is invisible: a link pasted into a private message or a group chat arrives untagged and lands in direct traffic. A “how did you hear about us” field on your enquiry form is imprecise, but it is better evidence than a model that cannot see the referral at all. Second, last-click attribution systematically undercredits social, because social is usually an early touch in a journey that ends on a branded search. Looking at attribution paths rather than a single model is the minimum honest treatment, and where the stakes justify it, proper analytics and data engineering can join platform data to what your CRM eventually recorded as revenue.

One quieter benefit rarely makes it into reports: consistent, active profiles referenced through the sameAs property in your site’s Organization structured data help search engines connect those accounts to your brand entity. It is not a ranking trick. It is disambiguation, and for a business with a common name it is worth doing properly.

The bad day, planned for in advance

Every account eventually has one: a product failure, a badly judged post, a news event that makes the scheduled queue look grotesque, or a coordinated pile-on.

The plan is short, and it needs to exist before it is needed. Pause the queue immediately — one person, from a phone, no permission required. Name the single decision-maker for the day and the person who drafts. Keep one holding statement pre-approved that acknowledges without conceding, and use it once. Move detail into a private channel. Do not argue in replies, ever; nobody has won that, and it manufactures the screenshots. Log what happened and what was said, because the review afterwards is where the actual improvement comes from.

Most reputational damage on social is caused by the response rather than by the incident. Silence and defensiveness are the two reliable ways to turn a small problem into a large one.

What actually drives the cost

We do not publish prices, because production volume — not strategy — sets the number, and volume is unknowable before scope is agreed. The variables are not a secret, though, and knowing them lets you brief better and compare quotes honestly.

How many platforms, and whether content is made natively for each. Cross-posting one asset everywhere is cheap and performs accordingly. Native production per platform is the single largest cost line in any social programme.

How much original video. Filming, editing, captioning and versioning are the expensive part of modern social. Repurposing existing footage is dramatically cheaper and often sufficient.

Cadence. Frequency multiplies everything above it. Doubling the calendar roughly doubles production, review and reporting.

How many approvers. This is a cost even though it never appears as one. Every additional reviewer adds cycle time and rework, and in regulated sectors it can dominate the schedule entirely.

Community management cover. Business hours is one thing; evenings, weekends and multiple languages are another. The response-time commitment is what makes this expensive, so it should be set deliberately rather than inherited.

Creator fees and media budget. These belong in the plan explicitly, so that nobody discovers halfway through that amplification was assumed to be free.

How this connects to the rest of the work

Social rarely stands alone. The pillars should come from the same strategy driving your wider digital marketing programme, the assets need a template system that survives volume, the paid layer becomes a specialism as soon as budget is material, and the measurement only means anything if the analytics underneath it are trustworthy.

If you are not sure which of those you actually need, that is a reasonable thing to put to us directly. We would rather tell you that two of your six platforms should be closed than sell you content for all of them.