Most social media guides are lists of tactics. This one is about the decisions underneath them, because the tactics change every eighteen months and the decisions do not.

Four things have genuinely shifted the way social media marketing works, and each changes what a sensible plan looks like.

1. Broad targeting now beats manual targeting

For most of the last decade, the skill in paid social was audience construction — interest stacks, lookalikes, layered exclusions. That skill has largely been automated away. The platforms now find the buyer faster than a media buyer can describe them, provided they are given a clean conversion signal to aim at.

What this means practically: stop spending time on audience definition and start spending it on creative volume and on the quality of the event you optimise toward. Feeding back a qualified-sale event instead of a form fill changes results more than any targeting change now does. That is the core of how paid social is run today.

2. Social platforms are search engines

A meaningful share of product research now begins inside TikTok, Instagram and YouTube rather than in Google. People type queries into social apps and expect answers, which makes the words in your captions, on-screen text and video titles function as ranking signals.

  • Write captions that contain the question someone would type, not a slogan.
  • Put the query on screen in the first frame.
  • Name your services the way buyers say them, not the way your industry does.
  • Treat each platform’s search results as a channel, not an afterthought.

3. Creator and customer content outperforms studio work

Content that looks like it came from a person consistently outperforms content that looks like it came from a brand. This is not a style preference; it is what the feed rewards, and the gap has widened rather than closed.

The practical implication is a production model built for volume and iteration rather than for polish. Fewer expensive assets, more variations, faster replacement of what stops working. That is easier to sustain with an in-house rhythm than with a quarterly agency shoot — one reason the freelance consulting model suits this shift.

4. First-party data is the durable asset

With third-party cookies gone and platform signal degraded, the data you own is worth more than the data you rent. Email lists, customer records, on-site behaviour and server-side event tracking now carry the accuracy that pixels used to.

Brands that invested in owned data have measurable campaigns. Brands that did not are optimising toward numbers they cannot verify. Building that layer sits inside lead generation and it is unglamorous, necessary work.

Choosing platforms without guessing

The honest answer for most businesses is two platforms, chosen from where the last ten customers actually came from. Everything else is optional. A strong presence on two beats a weak presence on five, because both the algorithm and the audience reward consistency.

If you genuinely do not know where customers come from, that is the first thing to fix, and it takes a week. Strategy work is documented on the social media marketing strategy page.

The content system that survives a busy month

Content fails on sustainability far more often than on quality. Build a system that works when you are busy: a running list of real customer questions, one publishing rhythm you can hold, and a format simple enough to produce on a phone.

  • One recurring series beats scattered one-off posts.
  • Repurpose across formats before creating anything new.
  • Publish on a cadence you can keep for six months, not six weeks.
  • Answer comments — distribution and conversations both improve.

Measuring what predicts revenue

Reach, impressions and follower counts do not predict revenue and never did. Four numbers do: cost per qualified enquiry, enquiry-to-customer rate, average customer value, and repeat rate.

Attribution has become harder, not easier, which makes incrementality thinking more useful than last-click reporting. The practical version for most businesses is simple: hold spend flat for a period, change one thing, and see whether the qualified enquiry count moves.

Where AI genuinely helps

AI has lowered the cost of production, variant creation and first drafts substantially. It has not lowered the cost of judgement. Deciding what to say, to whom, and on which platform still determines whether any of the output is worth publishing.

The businesses getting the most from it are using it to produce more variations of a strategy that was already right, not to generate a strategy.

A ninety-day plan that works

Weeks one and two: audit, and switch off what is not working. Weeks three to six: fix the offer, the site’s next step and the measurement layer. Weeks seven to twelve: scale the two channels that earned it. Doing these in sequence is what lets you tell what caused what.

If you would like this applied to your own numbers rather than in the abstract, book a free 30-minute call, or read how a consulting engagement is structured.