An editorial line is not a list of topics. It is a list of things you have decided not to publish, and that is what makes it usable by anyone other than the person who wrote it.

A list of topics tells a writer what is permitted, which is everything adjacent to your product. A list of exclusions tells them what to do with the eleven ideas they had this morning, which is the actual decision.

The reason this matters more than it should is that the evidence normally used to make format and channel decisions turns out to be weaker than its circulation suggests. Most of it surveys the wrong population.

The format research surveys marketers, not readers

This is the structural problem with almost every published statement about what works.

The most cited annual survey publishes its numbers honestly. Its latest edition states that the survey “was fielded between June 24 and August 14, 2025” and “drew 1,229 global responses”, of which the analysis covers “1,015 B2B marketers, mostly from North America.”

What it does not publish. Recruitment method, sampling frame, response rate. It is a self-selected sample from the publisher’s own distribution lists.

And the population is the wrong one for the claim. It asks marketers what they do and what they think works. Any figure from it describing a top-performing format is a marketer’s belief, and marketers are not a sample of buyers.

Which does not make it useless. It is a reliable picture of what the profession is currently doing, and that is worth knowing, particularly if your aim is to do something else.

A second survey supplies the consumption figures, and its bands overlap. One edition reports 1 to 3 at 28 percent, 3 to 5 at 42 percent, 5 to 7 at 19 percent. The values 3 and 5 sit in two bands each, so any aggregate built from them cannot be reproduced.

And its most recent edition contains no methodology at all. No sample size, no field dates, no demographics. It is sponsored by a content agency.

One further absence covers the format most B2B teams invest in most heavily. No published study with a stated methodology measures what proportion of a downloaded document is actually read.

Which produces an uncomfortable position, and a freeing one. There is no external evidence telling you which format works for your audience. That means the decision is yours, and it also means nobody can overrule it with a benchmark.

The population surveyed by content format research and the consequences for what may be claimed from itThe population surveyed by the research most commonly used to justify content format decisions, and the consequences of that population for the claims that may legitimately be drawn from it. The most frequently cited annual survey publishes its field window and response count transparently, stating that the survey was fielded between the twenty fourth of June and the fourteenth of August 2025, that it drew one thousand two hundred and twenty nine global responses, and that the reported analysis covers one thousand and fifteen business to business marketers located mostly in North America. It does not publish its recruitment method, its sampling frame or its response rate, making the sample self selected from the publisher’s own distribution lists rather than drawn from a defined population. More consequentially, the survey asks marketers what they do and what they believe performs well, so that any figure derived from it describing a top performing content format constitutes a belief held by content producers rather than a measurement of the behaviour of content consumers, and marketers do not constitute a sample of buyers. This does not render the survey useless, since it provides a reliable picture of what the profession is currently doing, which is valuable information particularly for an organisation intending to do something different. A second survey supplies the figures describing how many pieces of content a buyer consumes before purchasing, and its published response bands overlap, reporting one to three at twenty eight percent, three to five at forty two percent and five to seven at nineteen percent, so that the values three and five each appear within two bands and no aggregate constructed from them can be reproduced. That survey’s most recent edition contains no methodological disclosure whatsoever and is sponsored by a content production agency. A further absence covers the format attracting the heaviest investment, since no published study with a stated methodology measures what proportion of a downloaded document is actually read.Who is being asked, and what that permitsWhat the survey does establishWhat 1,015 B2B marketers report doing,from fieldwork in mid-2025.A reliable picture of what the professionis currently doing.What it cannot establishWhich format an audience prefers, becausethe audience was not asked.”Top-performing format” from it is amarketer’s belief.And the consumption figures have a construction problemBands of 1-3, 3-5 and 5-7 overlap on 3 and 5. Any aggregate from them cannot be reproduced.Which is uncomfortable and also freeingThere is no external evidence telling you which format works for your audience. So the decision is yours,and nobody can overrule it with a benchmark, because the benchmark does not exist.
They measure what marketers do and believe. That is a real thing to know, and it is not evidence about readers. Source : Content Marketing Institute annual B2B research; Demand Gen Report content preferences surveys (2026)

The engagement techniques the platforms suppress by name

Before deciding what to publish, it is worth knowing what one distribution system has published about what it actively demotes.

The mechanism is stated. “Our Content Distribution Guidelines describe the types of content we think may either be problematic or low quality, so we reduce its distribution in Feed for everyone. These reductions, also called demotions.”

Two named categories are exactly what engagement advice recommends. Clickbait links, and engagement bait.

And borderline material is reduced without violating anything. “We use personalization to reduce the distribution of content that doesn’t violate our policies but may get close.”

Which creates a specific trap. Techniques taught as engagement optimisation, provoking a reply, withholding the point, asking a question with an obvious answer, are named categories of demotion on the system they are designed to exploit.

The platform that matters most for B2B publishes almost none of this. Its help page says only that “our AI systems and algorithms consider hundreds of signals”, with no stages, no named signals and no demotion categories.

One sourcing warning worth carrying. Pages on that platform’s own domain under a content aggregation path look official and are aggregated member posts. They contain precisely the unsourced engagement multipliers this section is about. Domain is not provenance.

What follows for an editorial line. Optimising for reactions is optimising against a documented demotion list on one platform and against an undocumented system on the other. Neither is a stable strategy.

Published content demotion categories and the asymmetry in platform disclosureThe content demotion categories published by one major platform, and the absence of any equivalent disclosure by the platform most relevant to business to business publishing. The disclosing platform states that its content distribution guidelines describe the types of content it considers may be either problematic or low quality, whose distribution it reduces in the feed for all users, and that these reductions are also called demotions. Named categories among those demoted include clickbait links and engagement bait. The platform separately states that it uses personalisation to reduce the distribution of content which does not violate its policies but which may come close to doing so, described as borderline content. This creates a specific difficulty for content strategies built around engagement optimisation, because techniques commonly taught for that purpose, including provoking a reply, withholding the substantive point of a post, and posing a question with an obvious answer, correspond to named categories of demotion on the very system those techniques are designed to exploit. The platform most relevant to business to business publishing discloses almost none of this, its feed ranking help documentation stating only that its systems and algorithms consider hundreds of signals in determining what content appears in each member’s feed, without describing stages, naming signals or identifying demotion categories. A sourcing warning applies to that platform, since pages hosted on its own domain under a content aggregation path present with an official appearance while consisting of aggregated posts written by individual members, and those pages carry precisely the unsourced engagement multipliers that a content strategy should be sceptical of, meaning that the domain on which material appears does not establish its provenance. The consequence for an editorial position is that optimising for reactions means optimising against a documented demotion list on one platform and against an undocumented system on the other, neither of which constitutes a stable basis for a strategy.Published demotions, and published silenceNamed categories it demotesClickbait linksEngagement baitContent likely violating standardsAnd “borderline” content that violates nothingThe B2B platform’s disclosure”Our AI systems and algorithms considerhundreds of signals.”No stages. No named signals. No demotioncategories.The trap that createsProvoking a reply, withholding the point, asking a question with an obvious answer. These are the demotion list.Optimising for reactions means optimising against a documented demotion list on one platform, and against anundocumented system on the other. Neither is stable.
Engagement bait is a published demotion category. The techniques taught to increase reactions target a system that names them. Source : Meta Transparency Center, types of content we demote; LinkedIn Help, feed ranking (2025)

What an editorial line actually contains

Six decisions, most of them negative, all of them portable to somebody who did not attend the meeting.

The subjects you will not cover. Not because they are uninteresting but because covering them would put you in a category you are not trying to win. This is the hardest one and it does the most work.

The claims you will not make. Numbers you cannot substantiate, comparisons you cannot support, and predictions you have no basis for. Writing this down converts a hundred future editorial arguments into one policy.

The formats you will not produce. Deciding in advance not to make short video, or not to make gated documents, removes a recurring debate whose evidence base does not exist anyway.

The reader you are writing for, named specifically. Not a persona document. One sentence identifying who the piece assumes the reader is and what they already know, because that determines everything about the register.

The standard of proof you hold yourself to. Whether an assertion needs a source, whether a figure needs a method attached, whether an example needs to be real. Publishing that standard is a differentiator in a field where most published statistics do not survive tracing.

And what a piece must do to be worth publishing. Not a word count. A test: does it say something the reader could not have assumed, and could you defend it to somebody who disagrees.

The six decisions constituting a usable editorial line, of which four are exclusionsThe six decisions that together constitute an editorial line capable of being applied by a person who did not participate in the discussion that produced it, of which four take the form of exclusions. The first decision identifies the subjects the organisation will not cover, chosen not because those subjects lack interest but because covering them would place the organisation within a category it is not attempting to win, this being described as the most difficult decision and the one performing the greatest quantity of work. The second identifies the claims the organisation will not make, comprising figures it cannot substantiate, comparisons it cannot support and predictions for which it has no basis, and recording this converts a large number of future editorial disputes into a single policy. The third identifies the formats the organisation will not produce, since deciding in advance against short form video or against gated documents removes a recurring debate whose supporting evidence base does not in any case exist. The fourth identifies the reader the material is written for, expressed not as a persona document but as a single sentence stating who the piece assumes the reader to be and what that reader already knows, since this determines the register throughout. The fifth identifies the standard of proof the organisation holds itself to, covering whether an assertion requires a source, whether a figure requires an accompanying method, and whether an example must be genuine, the publication of which constitutes a differentiator in a field where a substantial proportion of published statistics do not survive being traced to their origins. The sixth identifies what a piece must accomplish to justify publication, expressed not as a word count but as a test asking whether the piece states something the reader could not have assumed, and whether the writer could defend it to a person who disagreed. Four of these six are refusals, which is the property that makes the resulting document applicable by a writer rather than merely agreeable to a committee.Six decisions, four of them refusals1. Subjects you will not coverNot uninteresting. Just in a category youare not trying to win.2. Claims you will not makeUnsubstantiated numbers, unsupportablecomparisons, baseless predictions.3. Formats you will not produceRemoves a recurring debate whose evidencebase does not exist anyway.4. The reader, in one sentenceWho the piece assumes they are and whatthey already know.5. Your standard of proofDoes an assertion need a source. Does afigure need a method attached.6. What makes a piece publishableNot a word count. Does it say somethingthe reader could not have assumed.Why the refusals are the useful halfA list of permitted topics tells a writer that everything adjacent to your product is allowed. A list of refusals tellsthem what to do with the eleven ideas they had this morning.
A list of permitted topics helps nobody. A list of refusals is what a writer can actually apply. Source : Method (2026)
The derivation of the in market proportion and its appropriate use in editorial planningThe derivation of the widely cited proportion of business buyers who are in market at any given moment, and the appropriate and inappropriate uses of that proportion in editorial planning. The derivation is arithmetic rather than empirical. If companies change supplier approximately every five years, then approximately one fifth of them are in market during any given year, and consequently approximately one twentieth are in market during any given quarter. This is a deduction from an assumed purchase cycle rather than a count of buyers, and it should not be presented as a measurement or used as the basis of a forecast. Its legitimate use is as a reframing of the audience, since it establishes that most people reading any given published piece are not going to purchase during the current quarter, which means the piece does not function as a sales asset and cannot reasonably be evaluated as one. This changes the editorial test that should be applied, replacing the question of whether a piece generates a lead with the question of whether the piece will still be true and useful to a person who encounters it eighteen months later at the point when that person is in market. It also changes the selection of subjects, favouring topics with a long shelf life, positions the organisation can defend over time, and explanations of things that do not change on a quarterly basis, which is the opposite of writing to a news cycle. One consequence follows for measurement, namely that a piece whose function is to be remembered eighteen months after publication cannot be evaluated within the month in which it is published, and this mismatch between the horizon of the work and the horizon of its assessment is why editorial programmes are commonly cancelled around the fourth month.Most readers are not buyingThe derivation, which is arithmeticChange supplier every five years, so a fifth are in market in a year, so a twentieth in a quarter.A deduction from an assumed cycle. Not a count of anybody.Legitimate useReframing the audience, and thereforethe test you apply to a piece.Illegitimate useAs a measurement, or as the basis ofany forecast.The editorial test it producesNot “does this generate a lead” but “will this still be useful to someone in eighteen months.”Which is why editorial programmes get cancelled at month four. The horizons do not match.
An arithmetic deduction, not a count. Useful for setting the editorial test, useless as a forecast. Source : Ehrenberg-Bass Institute, on the derivation of the in-market proportion (2026)

Writing for people who are not buying

One framing is worth adopting explicitly, provided it is cited for what it is.

The observation. If companies change supplier roughly every five years, then about a fifth are in market in a given year, and about a twentieth in a given quarter.

What that is. An arithmetic deduction from an assumed purchase cycle. It is not a count of buyers, and it should not be presented as a measurement.

What it is useful for. It reframes the audience. Most people reading anything you publish are not going to buy this quarter, which means the piece is not a sales asset and cannot be judged as one.

Which changes the editorial test. Not “does this generate a lead” but “will this still be true and useful to someone who reads it eighteen months from now, when they are in market.”

And it changes what to write about. Subjects with a long shelf life, positions you can defend, and explanations of things that do not change quarterly. The opposite of a news cycle.

One consequence for measurement. A piece whose job is to be remembered eighteen months later cannot be measured in the month it is published, which is why editorial programmes get cancelled at month four.

Permissible claims arising from each category of content marketing evidenceThe claims that may legitimately be drawn from each category of evidence commonly used in content marketing decisions, distinguished by what each source actually measures. A survey of marketers reporting on their own activity permits a claim about what the profession is currently doing, and about the prevalence of particular practices among practitioners, but permits no claim whatsoever about audience preference, because the audience was not surveyed. A survey of buyers reporting their own content consumption permits a claim about what buyers report doing, subject to the reliability of retrospective self report, but where the published response bands overlap as they do in the principal such survey, no reproducible aggregate can be constructed from it, and where the most recent edition publishes no methodology at all no claim of any kind can be supported. Platform documentation describing ranking and demotion permits a claim about what a distribution system is designed to promote or suppress, which is directly actionable because it describes the mechanism rather than an outcome, and one platform names its demotion categories explicitly including clickbait links and engagement bait. Regulatory filings permit claims about audited user counts under stated definitions. Probability samples with published margins of error permit claims about audience composition, such as the proportion of a national adult population using a platform and its distribution by education level. What no available source permits is a claim about which content format a specific business audience prefers, since no published study measures that, and no published study measures what proportion of a downloaded document is actually read. The practical consequence is that format decisions must be made on reasoning and on the organisation’s own observation rather than on published evidence, and that this is a constraint rather than a failure, because it applies equally to every competitor consulting the same absent literature.What each source lets you saySourceWhat you may claim from itSurvey of marketersWhat the profession is doing. Nothing about audiences.Survey of buyers, overlapping bandsNothing reproducible. The aggregate cannot be rebuilt.Platform ranking documentationWhat the system promotes and suppresses. Actionable.Probability sample with a published marginWho is present on a platform. The best available input.Which format your audience prefersNo source. Nobody has published this.Which is a constraint rather than a failure, because it applies to every competitor reading the same absent literature.
Three sources, three different permissible claims. None of them tells you which format your audience prefers. Source : Method, over the sources examined in this article (2026)

What to do with this

Write the exclusions first. Three subjects you will not cover, three claims you will not make, one format you will not produce. That document is shorter than a content strategy and more useful than one.

Then set your standard of proof and publish it. In a field where the most repeated statistics do not survive being traced, a stated rule about sourcing is a genuine differentiator and costs nothing.

Stop optimising for reactions. One platform names engagement bait as a demotion category, and the other publishes nothing you could optimise against with confidence.

And judge pieces against the eighteen-month test rather than the month they appear in. If most of your readers are not buying now, the only sensible question is whether they will remember this when they are. An editorial line inherits its subjects and its register from the brand above it, so settling the positioning and the messaging first saves writing the line twice.

The related pieces are does a white paper still work and organic social benchmarks do not exist.