The acronym was coined in May 2010 by someone else. The post everyone credits does not contain it, and the trademark declares a first use three years later.
The four-part media model is usually credited to a 2014 book. The acronym appears in an archived blog post from May 2010, written by somebody else, and the post everyone cites as the origin does not contain it.
This is not a pedantic point about credit. It matters because the model is one of the few genuinely useful planning frameworks in marketing, and a framework whose provenance is wrong tends to be taught with the reasoning stripped out.
The genealogy is short, fully documented in web archives and trademark filings, and more interesting than the version in circulation.
What the record actually shows
Four dates, each verifiable, in order.
March 2009: a three-part model, published by a manufacturer’s employee. A post titled around owned, bought and earned media, stating: “At Nokia we have been using the following simple model for our Digital Media Planning for about a year now.” The internal use claim cannot be verified; the publication date can.
December 2009: an analyst firm formalises the three-part version. And credits its predecessor directly: “Nokia was an early pioneer in this space”, adding that “‘Earned media’ is an old PR term.”
May 2010: the four-part acronym appears. In a post by a vice president of digital research at a public relations firm, captured by a web archive three days after publication: “We capture this new scope and integration in a model we refer to as PESO, Paid/Earned/Shared/Owned. Our PESO model predates the similar Forrester model (Paid/Earned/Owned) and is different in an important way.”
June 2013: the post usually credited as the origin. It sets out “four forms of media: Paid, earned, owned, and shared” and describes each. Searching the archived capture for the acronym returns nothing. The letters do not appear.
February 2014: the book. Published 25 February 2014, and the source of the attribution that circulates today.
And 2023: the trademark. A registration for the model name, filed March 2022, registered February 2023, declaring a first use in commerce of 1 April 2013.
Which is the detail that settles the sequence. The declared first use is almost three years after the acronym was published by someone else.
What the trademark actually covers, and what it cannot
The model name is registered, which is worth understanding precisely because it changes what you may say rather than what you may do.
The registration. Filed March 2022, published for opposition November 2022, registered February 2023, in the services class, with a disclaimer on the generic word in the name.
The declared first use in commerce is 1 April 2013. That date is a sworn statement to the trademark office by the applicant, and it is the applicant’s own account of when they began using the name commercially.
Which is the arithmetic that matters. The acronym was published in May 2010, in a post archived three days after publication. The declared first use is almost three years later.
What a registration in that class covers. The commercial use of a name for services, typically training, certification and consulting. It does not cover the underlying idea, which is not the kind of thing trademark law protects.
So you may describe your media planning using the four categories freely. Naming the categories is not using the mark, and a disclaimer on a generic component of a mark exists precisely because generic words cannot be monopolised.
What you should not do is present the framework as somebody’s invention when the record says otherwise. That is a sourcing question rather than a legal one, and it is the point of this article.
Getting the origin right restores a distinction that the popular version lost.
The 2010 author claims a specific difference, not just an extra letter. His post says the four-part model “is different in an important way” from the three-part one. The shared category is not an addition to a list. It is the recognition that some distribution happens through people who are neither paid nor covering you journalistically.
Which is why shared and earned are not the same thing. Earned media is coverage somebody chose to produce about you. Shared is distribution somebody chose to give something you produced. The mechanisms are different, the people are different, and conflating them produces a plan that measures neither.
The three-part precursor is honest about its own lineage too. It notes that “‘Earned media’ is an old PR term”, which is true and which the four-part version does not change.
And the manufacturer’s original framing was about planning, not measurement. “We have been using the following simple model for our Digital Media Planning.” It was a way to allocate a budget across channel types, which is still the best use of it.
One correction that follows. The model is not a content strategy. It is a media planning taxonomy, and using it to categorise pieces of content rather than routes to an audience is what makes it feel unhelpful.
Four categories, and the question each one actually answers.
Paid asks what it costs to reach people who do not know you. Its property is that it stops when the money stops, which makes it a tap rather than an asset.
Owned asks what you would still have if every platform closed. Your site, your list, your documents. It compounds slowly and it is the only category you control.
Earned asks who will say something about you without being asked. Slow, unschedulable, and the only category with third-party credibility attached.
And shared asks who will pass on what you produced. Distinct from earned because the artefact is yours and the distribution is theirs.
The planning value is in the ratios, not the categories. A programme that is entirely paid has no asset. One that is entirely owned has no reach. One that relies on earned has no schedule. Deciding those ratios belongs to nobody when each category has its own supplier, which is the argument for planning brand, site and paid against one budget rather than three.
Which produces one honest test for a B2B plan. If paid stopped on Monday, what would still deliver on Friday, and what would be gone by lunchtime.
Use the model to allocate budget across routes to an audience, not to sort your content into four bins. That is what its original author was doing with it, and it is the use that still works.
Keep shared and earned separate in your reporting. One is somebody choosing to distribute your artefact, the other is somebody choosing to produce their own. Conflating them hides which of the two you are actually getting.
Run the Monday test once a quarter. Write down what would still be delivering on Friday if paid stopped, and how much of your pipeline that represents. The answer is the honest measure of how much asset you have built.
And when you cite the model, cite it correctly. The acronym is from May 2010, published by a research executive at a public relations firm, and the post that introduced it claimed a substantive difference from the three-part version rather than just an extra category.
The earliest verified use is a blog post dated 12 May 2010 by a research executive at a public relations firm, archived three days later. It states the model 'predates the similar Forrester model'.
Did the widely credited 2013 post introduce it?
No. That post sets out four forms of media by name but the string PESO does not appear in it. The archived capture confirms this.
Is the model trademarked?
The model name was registered in February 2023, filed in March 2022, with a declared first use in commerce of 1 April 2013. That is almost three years after the earliest verified use of the acronym.
Where did the three-part version come from?
An analyst firm published it in December 2009, and that post itself credits a mobile manufacturer as an early pioneer, whose own version was published in March 2009.