Marketing Systems, Not Campaigns

Campaign-led marketing has a recognisable shape. Budget goes in, attention spikes, the spike decays over a few weeks, and the team starts planning the next one from roughly where they started. Draw twelve months of it and you get a sawtooth: a lot of vertical effort, a flat trend line.

It isn't that campaigns are wrong. Launches need moments, and moments need concentration. The problem is when campaigns are the only unit of work, because then nothing survives between them, and every quarter begins by rebuilding attention from scratch.

The diagnostic is uncomfortably simple. You can probably name your last four campaigns. Can you name your baseline — the demand that arrives when you're not doing anything? If the answer is no, or if pausing spend causes pipeline to fall away within a few weeks, you don't have a marketing function. You have a series of events.

What actually compounds

An asset compounds if two things are true: it keeps producing after the spend stops, and it makes the next thing cheaper or better. That second clause is the one people forget, and it's the one that changes the arithmetic.

Things that compound:

• A body of writing that answers the questions buyers genuinely ask — not keyword pages, but the material your sales team already sends by hand. It earns links, gets cited, gets forwarded, and quietly shortens conversations you're having anyway.
• An audience you own. An email list, a community, a customer base you can reach without paying an intermediary for permission each time.
• A design system and a set of templates. Unglamorous and enormously leveraged: every future landing page, ad set, and deck gets faster and more consistent because the decisions were made once.
• Measurement infrastructure. Once you can see which activity produced revenue, every subsequent decision is better. This one compounds harder than almost anything else and gets funded last.
• Distinctive brand assets. A colour, a shape, a phrase people recognise. Every impression contributes to something that already exists rather than starting over.
• Customer stories and proof. Reusable across sales, site, ads, and pitch — and they appreciate as the customers get more successful.

Things that don't: paid clicks, sponsorships, a conference stand, a burst of influencer posts. All of those are rented attention. Rented attention is a legitimate purchase — it just isn't an asset, and confusing the two is how a marketing budget can grow for three years while the business gets no easier to sell.

One question sorts almost everything: if we stopped paying for this tomorrow, what would still be here in six months?

Brand and performance are one budget arguing with itself

Performance marketing harvests demand that already exists. Brand marketing creates the demand performance will later harvest. They are sequential, not competing, and every organisation that treats them as rival line items eventually discovers this the hard way.

The failure runs like this. Brand spend is cut because it can't be attributed. Performance efficiency improves for a quarter or two, because there's a stock of existing awareness to convert. Then the pool of people who already know who you are stops being replenished, acquisition costs climb, and the response is to spend more on performance — bidding harder for the same finite set of in-market buyers with nothing to distinguish you from anyone else bidding.

Performance marketing does not create customers. It finds the ones your brand already made.

The trap underneath is a measurement asymmetry, not a strategic disagreement. Performance is legible inside an attribution window. Brand mostly isn't. So budget flows towards what can be reported rather than what works, and it does so under the entirely sincere banner of being data-driven.

The honest way through is to stop pretending the two can be judged on the same evidence. Hold performance to short-horizon efficiency metrics. Hold brand to longer-horizon ones — branded search volume, direct traffic, unprompted recall inside your actual segment, win rate against named competitors, how long the sales cycle takes. Then say out loud that the brand allocation is a judgement call. A stated judgement is manageable. A judgement disguised as a measurement is not.

What to instrument before you spend

This is the least interesting section and the one that decides whether anything above is knowable. Before meaningful money goes into media:

1. One written definition of a lead or conversion, agreed by sales and marketing. Two definitions produce two dashboards that can never be reconciled, and a permanent argument that consumes more time than the work.
2. A path from first touch to closed revenue, even a crude one. Knowing roughly which channel produced revenue beats knowing precisely which channel produced form fills. Optimising towards form fills is how teams double their lead volume and their sales team's frustration simultaneously.
3. A baseline. Measure the four to eight weeks before you start. Without it, every result afterwards is unfalsifiable — you can neither prove it worked nor accept that it didn't.
4. First-party, consented data capture. Assume third-party signal keeps degrading and build accordingly. Direct relationships with your audience are the hedge.
5. A plan for a holdout. Geographic splits, or deliberately pausing a channel for a defined period. It feels like leaving money on the table and it is the only way to distinguish incremental effect from correlation.
6. Naming conventions for campaigns and URL parameters, decided once and enforced. This sounds beneath discussion. It is the single most common reason a year of data turns out to be unanalysable.

Then match reporting cadence to mechanism. Paid search can be read weekly. Content, brand, and community cannot — judging a compounding asset on a campaign clock is exactly how good work gets killed in month three, right before the curve would have turned.

Making the shift without stalling

You can't stop harvesting while you build; the pipeline still has to fill this quarter. The practical move is to allocate by horizon rather than by channel, and to hold the ratio steady:

• Harvest — activity expected to produce pipeline this quarter.
• Build — assets expected to pay out over the next two to four quarters.
• Bet — deliberately small, deliberately unproven, cheap enough to lose.

The ratio matters less than the discipline of not raiding build to rescue a bad quarter — which is, without fail, the exact moment it gets raided. That single act is what keeps organisations on the sawtooth for years: every downturn cannibalises the thing that would have prevented the next one.

Alongside that, do fewer things and maintain them. One article revised every year outperforms twelve that quietly rot. One landing page tested properly beats six launched and abandoned. Systems reward maintenance in a way campaigns never do, and maintenance is unglamorous work that nobody puts in a quarterly review — which is precisely why it's available as an advantage.

The shift isn't really about tactics. It's about what you're willing to be measured on. Campaigns let you report activity. Systems make you report a trend line, and a trend line is harder to hide behind — which is the whole point.