Does content marketing still pay back

The question arrives in a specific way. Somebody has pulled up the analytics, the sessions line is flat or gently down, and the person who has been paying for four articles a month for two years asks whether this is still worth doing. What follows is usually a conversation at cross purposes: the marketing side answers with traffic and rankings, the finance side wanted an answer about money, and the meeting ends with a decision to "refresh the content strategy", which means the same four articles a month with different titles.

Before arguing that the channel has narrowed, it is worth stating the case against that view at full strength, because it is better than the sceptics think. Content is the only acquisition spend most small companies make that leaves something behind. A page that ranks keeps producing while nobody attends to it, which is the property that separates a channel from a campaign and the reason it sits at the top of the list in our piece on acquisition channels that survive scrutiny. The companies that cut content hardest during the last two downturns did not discover a cheaper route to customers. They discovered that they now buy every single visitor, forever, at a price that goes up each year, and that the pages they wrote in 2016 are still bringing in enquiries that cost nothing at the margin. That asymmetry is real and it has not gone away.

What has changed is that the asymmetry no longer applies to the whole category. It applies to a shrinking part of it, and the part it applies to can be identified in advance. Our position is that content still pays back where the buyer has to be taught something before a purchase is possible, and has stopped paying back — not slowed, stopped — for informational topics that a search result page now answers on the spot.

The test is whether the buyer has to learn something before they can buy

Most content planning starts from keywords: what do people search, what has volume, what can we plausibly rank for. That produces a list of topics that are related to your business and mostly useless to it, because search volume measures curiosity rather than intent, and curiosity does not convert at any rate worth modelling.

The more useful question is whether there is something a buyer must understand before they are capable of buying from you. Some purchases have that property structurally. Anyone switching payroll providers has to understand what happens to the mid-year tax records, and until they understand it they cannot approve the switch — so the page explaining it is not marketing adjacent to the sale, it is a step inside the sale. Anyone specifying an industrial component has to know which tolerance actually matters for their application. Anyone buying a compliance product has to grasp what the regulation requires of them specifically before a proposal means anything.

Where that requirement exists, editorial content does work no salesperson can do at scale and no ad can do at all. The buyer arrives at the page mid-decision, teaches themselves something they needed, and forms an opinion about who explained it. That opinion is most of what a shortlist is made of.

Where the requirement does not exist, content is decoration. If someone can buy your thing without learning anything — because the category is understood, the choice is obvious, the purchase is small — then articles about the category do not shorten the path. They fill a calendar. A large amount of B2B publishing is in this second position and the companies doing it cannot tell, because traffic arrives either way.

The distinction is not about industry or company size. It is a property of the specific purchase, and it is worth writing down before commissioning anything:

Topic type	Does the buyer need it to proceed	Now answered on the results page	Worth publishing
What a term means	No, they can look it up anywhere	Yes, almost always	No
Best X for Y, generic	Rarely, and they distrust your answer	Increasingly	Rarely
How the decision affects their existing setup	Yes, and only you can say	No, it depends on specifics	Yes
What goes wrong and how you handle it	Yes, it is the objection	No	Yes
Industry news and commentary	No	Not relevant, nobody searches it	Only for an owned audience

The traffic that disappeared was never the traffic that paid

The visible symptom of the change is that informational queries no longer produce clicks the way they did. A question with a short factual answer gets that answer above the results, and a meaningful share of people never scroll. If your archive is built on definitional and explanatory pieces aimed at the top of the funnel, the decline you are seeing in sessions is structural and will not be fixed by rewriting them, adding schema, or publishing more.

The reason to be calm about this is that the lost traffic was mostly never worth anything. A glossary page ranking for a definition brought students, competitors, people writing their own articles, and a thin sliver of prospects who would have found you another way. It looked like the top of a funnel because it sat at the top of a chart. The pages that produce enquiries tend to be the specific ones — the awkward comparison, the piece on what the migration actually costs, the one that names the situation where your product is the wrong choice. Those hold up, because the answer is contingent on the reader's circumstances, and a summary above the results cannot resolve a contingency.

There is a real loss buried in this, and it is not traffic. It is the discovery path. Companies used to be found by people who did not yet know they had a problem, and that route is now much narrower. The compensating move most publishers are making is to build a direct audience rather than rent one from search, which is a different economic argument with a different failure mode — we have gone through it separately in what a newsletter is worth to a B2B company.

Twice a month buys you a long wait and then, sometimes, an annuity

Assume a B2B company publishing two substantial pieces a month on a domain without much history. The honest shape of the return, from what we see rather than from anyone's case study, is that nothing happens for two to three quarters, the first attributable enquiries appear somewhere after that, and the cumulative return crosses the cumulative cost noticeably later still — often in the second year, and for genuinely competitive categories later than that.

Two things make that number worse than it looks in a plan. The first is that the clock starts at the first invoice rather than the first result: the hiring, the false start with an agency that did not understand the product, the six pieces published before anyone worked out which topics mattered. The second is that latency is not the same as delay. A page does not switch on at month nine; it drifts up over months, which means the early data is genuinely uninformative and the temptation to change course lands exactly when there is nothing to read. The mechanics of that ramp are worth understanding before signing anything annual, and we have set them out in how long before SEO shows up in the numbers.

What makes the wait rational is the far end. Once the pages are ranked, the marginal cost of the next enquiry is close to zero and stays there while you do other work. No paid channel has that property. The decision is therefore not about whether content works. It is about whether you can fund eighteen months of something producing nothing, which is a question about the balance sheet and not about marketing.

The expensive hour is the one nobody invoices

Content budgets are built from the wrong line. The writer's fee, the freelance rate, the agency retainer — that is the number in the spreadsheet, and it is the smaller number.

The larger one is the internal expert. Anything worth reading in a specialist field contains something only your people know: how the implementation actually goes, why the obvious approach fails, what the number in the proposal is made of. Extracting that takes an interview, a draft that gets most of it wrong, a correction pass, and usually a second conversation because the correction revealed a better piece. That is a few hours of someone senior per article, and those hours come out of delivery, sales or engineering, where they already had a use.

Two pieces a month at that standard is a meaningful ongoing claim on your best people, and it is the reason most content programmes quietly degrade rather than stop. Nobody cancels them. The expert becomes unavailable, the writer fills the gap with material assembled from what is already online, and the output continues at the same cadence with the one ingredient that made it worth anything removed. The cost line stays flat, the asset stops being built, and the analytics take a year to show it.

If you are not willing to commit the expert's hours, that is a legitimate decision. It is also a decision not to do this channel, and it is cheaper to make it openly than to discover it in the archive eighteen months later.

Fewer, longer, and with something at stake

The cadence habit comes from a period when publishing volume itself produced results. Under current conditions, twenty-four competent pieces a year lose to five that a salesperson would actually send to a prospect.

The reason is not aesthetic. A piece that takes a position can be disagreed with, which means it can be right about something, which means it can change what a reader does. A piece assembled from the consensus of the first page of results cannot — the reader has already absorbed that consensus and gets nothing from meeting it again under your logo. Positions also survive the summarisers better, because a judgement with reasoning behind it does not compress into a two-line answer without losing the part that mattered.

The practical version: fund the pieces where your company knows something the internet does not, write them long enough to be complete, and let the calendar be whatever it turns out to be. Then keep a short list of the pages that produce enquiries and update those rather than adding new ones. Maintenance of a working page is cheaper than a new piece and returns more, and almost nobody does it because it does not feel like output.

The test we would apply before commissioning anything: would someone in your sales team send this to a live prospect, unprompted, because it answers the thing that is currently holding up the decision. If nobody would, the piece has no job. Most content programmes that stopped paying back are full of pieces that would fail that question, and the archive is where you can check.