We have produced a reasonable amount of sponsored editorial and measured completion carefully. The spread between the best and worst performers is enormous, and the differences are consistent enough to describe.
The failure mode is always the same
Pieces that are abandoned early are abandoned for one reason: the reader worked out within a few sentences that the piece exists to sell something and contains nothing else.
The signals that trigger that are recognisable. The brand named in the first paragraph. A problem described in terms that obviously set up the product. Language that has been through an approvals process.
Readers detect this immediately and it is not close. Completion rates on pieces with those characteristics were a fraction of the others.
What the successful pieces had in common
Four things, consistently.
They were genuinely useful independent of the product. Somebody could read the whole thing, not buy anything, and be glad they read it.
The brand appeared late and lightly. In our best performer the sponsor was mentioned once, near the end, and the disclosure at the top did the identification work.
They said something specific rather than general. A real number, a real example, a real admission of a limitation.
And they were written by somebody who could write, in a recognisable voice, rather than assembled from approved messaging.
The disclosure paradox
The finding that surprised us and that I now think is the most important.
Clear prominent disclosure did not reduce completion rates meaningfully in our pieces.
What reduced completion was the piece being bad, and readers discovering it was promotional after investing attention, which produces a sense of having been misled.
Disclosure up front sets an expectation. A good piece then exceeds it and the reader continues. A poor piece confirms a suspicion they had already formed.
Which means the argument for ambiguous labelling is not even commercially sound, quite apart from the regulatory position. Ambiguity buys a click and costs the engagement.
What the brand actually gets
The question that determines whether the format is worth using.
Not direct response. Sponsored editorial converts poorly compared with any direct format, and briefing it against conversion targets produces the bad version.
What it delivers is association and credibility. A reader who finishes a genuinely useful piece attributes some competence to the sponsor.
That is a brand outcome and should be measured as one — recall, association, favourability — rather than by clicks to a product page.
The programmes that work are the ones where the brief was to be useful, and the commercial return was accepted as indirect.
The publisher studio question
Worth addressing since it affects quality substantially.
Pieces produced by a publisher's own content studio consistently outperformed pieces we supplied, in our experience and in what others report.
The reason is straightforward. They know their audience, they write in the voice that audience expects, and they have editorial judgement about what will be read.
The trade is less control over the message, which is exactly the thing that makes it work. Brands that insist on approval over every line end up with the version that does not get read.
The workable arrangement we found was agreeing the territory and the factual claims, and leaving the writing alone.
Length
A practical finding that ran against expectation.
Longer pieces had higher completion rates than shorter ones in our set, which seems backwards and probably reflects selection rather than causation — the longer pieces were the ones with something to say.
What did correlate with abandonment was thin content padded to length, which is detectable within a couple of paragraphs.
Which points back at the same conclusion. The variable is whether there is anything there, and every other characteristic is downstream of it.
The test I would apply
Before commissioning anything: would this piece be worth publishing if no brand were paying for it.
If the answer is no, the piece will not be read, and the money is better spent on a format that does not depend on being read.
If the answer is yes, the sponsorship is buying association with something good, which is what the format was always supposed to do.
Distribution is half the cost
A commercial point worth stating plainly.
Producing a good sponsored piece and relying on the publisher's organic placement to distribute it generally produces disappointing numbers.
Most packages include paid amplification, and the split between production and distribution budget is worth negotiating explicitly.
Our experience is that a good piece with proper distribution outperforms two mediocre pieces with minimal distribution by a wide margin, which argues for fewer, better and pushed harder.
What to measure
Since the wrong metrics produce the wrong content.
Completion or scroll depth, which tells you whether it was read.
Time on page against a benchmark for the publisher's editorial, which tells you whether it held up against what surrounds it.
And a brand study where the budget justifies it, since the outcome is a brand outcome.
Click-through to a product page is the metric most commonly used and the least informative, because optimising for it produces exactly the promotional tone that stops people reading.