Native advertising exists because content that matches its surroundings gets engaged with more than content that announces itself as advertising.
That is the entire commercial proposition, and it is also the reason regulators pay attention to it. The tension is inherent rather than an implementation problem.
What the rules generally require
Regulatory frameworks differ by jurisdiction and converge on the same principle: a reasonable person should be able to tell that they are looking at advertising, before they engage with it.
In practice that has produced guidance about several things.
The disclosure must be clear and prominent, not in small light text at the bottom.
It must appear before or at the point of engagement, not after somebody has clicked through and read half of it.
The language must be understood. Terms like sponsored and advertisement test well. Terms like presented by, partner content, promoted and brought to you by test considerably worse in consumer research, and several regulators have said so explicitly.
And it must be visually distinguishable, which usually means the unit cannot be identical in every respect to surrounding editorial.
Where the industry actually sits
Being honest about it, compliance is uneven.
The large publishers with legal departments generally do this properly, with clear labelling and visual differentiation.
The problems concentrate in the content recommendation units at the bottom of articles, where labelling is frequently minimal, where the units mimic editorial headlines closely, and where the destination is often a considerably lower-quality experience than the site the user was on.
Those units are also where the most aggressive creative sits, because performance is measured on click rate and the formats that get clicked are the ones that most resemble editorial.
Regulators in several jurisdictions have taken enforcement action in this area, and the direction of travel is toward stricter requirements rather than looser.
The publisher's trade
The commercial calculation that drives the whole thing.
These units pay, and for many publishers they are a meaningful share of revenue at a time when other revenue has declined.
The cost is that they degrade the user experience and, more seriously, they associate the publisher's brand with whatever the destination turns out to be.
Several publishers have removed them entirely and reported that the revenue loss was smaller than expected because other inventory performed better without the clutter, and because the audience metrics improved.
Others have kept them and applied editorial standards to what can be promoted, which is a middle position that requires ongoing work.
What good native looks like
It exists and it is worth describing, because the format is not inherently a problem.
Clearly labelled, in language a reader understands, before engagement.
Genuinely useful content that the reader is glad to have read, which is the same standard as anything else.
Produced to the publisher's editorial standards, frequently by the publisher's own studio, which is why those units generally perform better and generate fewer complaints.
And relevant to the audience rather than merely targeted at them, which is a distinction that shows up clearly in engagement metrics.
The measurement question
A practical point that is worth raising because it affects incentives.
Native is usually measured on click rate, which rewards the executions that most resemble editorial, because those get clicked by people who thought they were clicking something else.
Those clicks are worth very little. The bounce rate on misled traffic is extremely high and the brand effect is plausibly negative.
Measuring time on the destination, scroll depth and any downstream action instead changes which creative wins, and it changes it toward the honest executions.
We made that change and our click rates fell while our cost per meaningful engagement improved, which was an awkward conversation and the right outcome.
The position I have come to
Disclosure that is genuinely clear costs some performance, and the amount it costs is a reasonable measure of how much of your performance depended on people not knowing what they were looking at.
If a unit performs well when clearly labelled, it is working because the content is good.
If it only works when the labelling is ambiguous, the format is not the problem and neither is the regulator.
The brand safety side
A related risk that sits with the advertiser rather than the publisher.
Content recommendation networks place units across large numbers of sites, and the advertiser frequently has limited visibility into where.
Which means a brand can find itself adjacent to content it would never have chosen, on domains it has never heard of.
Requesting a placement report, and applying an exclusion list to it, is basic diligence that a surprising number of advertisers do not do.
The lists are usually long and the top fifty domains typically account for most of the delivery, so reviewing them is an hour rather than a project.