Agency IP: how to build a proprietary framework clients believe

Andy Last co-founded Salt Communications in 2000 and sold it to IPG seventeen years later. Asked what he would do differently if he were building a PR agency today, his first answer was intellectual property.

"What lasts is proprietary thinking - frameworks, models, methodologies that clients can't replicate internally and that make your agency feel irreplaceable," he wrote in PRmoment. The models Salt developed, and the books and platforms they fuelled, became its best new-business tools. His regret is the timing: he wishes the agency had invested in that intellectual capital from year two rather than year six.

The rest of his list - pricing on value, picking a niche earlier, treating culture as a commercial asset, knowing your acquirers before you need them - is worth reading. IP comes first for a reason. It makes most of the others easier.

Most agency founders reading that will assume they already have it. They have a methodology with a name, a diagram in the credentials deck, perhaps a registered mark. Very few have what Last is describing.

Most agency methodologies are renamed workflows

The typical agency framework is a description of how the agency already works, given a proprietary-sounding name and drawn as a circle, a funnel or a set of interlocking shapes. Underneath the branding it is usually some version of discover, define, make and measure.

There is nothing wrong with having a process. The problem is presenting a process as thinking. A client comparing five agencies sees five diagrams describing broadly the same sequence, and correctly concludes that none of them is proprietary. The name was supposed to create distinctiveness. It ends up advertising the lack of it.

A workflow tells a client what order you will do things in, and owned thinking tells them something about their market they did not know.

That is the whole difference, and it is why so much agency IP fails to register with the people it was designed for.

What genuine agency IP is

Genuine IP is a claim about how the client's world works, which the agency can support with evidence and which leads to different decisions from the ones the client would otherwise make.

Each part of that definition does work. A claim about the world, rather than about the agency, is what makes it interesting to someone who has not hired you yet. Evidence is what separates it from an opinion anyone could hold. And a change in decisions is the test of whether it matters: if the client would do exactly the same things with or without it, the framework is decoration.

Last's own phrase is the sharpest filter available. Thinking clients cannot replicate internally. A client can replicate any workflow by hiring the right people and reading the right books. What they cannot replicate is the pattern the agency has seen across dozens of clients in a category, or the data it gathered that nobody else thought to gather.

A useful recent illustration of the shape, whatever you make of the substance: when M+C Saatchi Talk launched a strategic framework called The Conversation Converter in September 2026, Creative Boom's write-up described it through its premise - that coverage, reviews, creator content and wider brand conversations increasingly influence how brands are "understood, surfaced and recommended by search engines and AI platforms". It also named the two people who developed it, head of intelligence Marine Escande and analyst Chloé Walker. A premise about the market and named owners are two of the levers below. Whether a framework earns belief depends on the rest.

The tell-tales of fake IP

These patterns are easier to spot in your own work than to design out in advance.

The name arrived before the idea. Someone in a strategy session wanted a branded method for the credentials deck, and the thinking was reverse-engineered to fit the acronym. Readers can usually tell, because the stages are the letters of a word.

It describes the agency, not the market. Every element is about what the agency does - its stages, its teams, its deliverables. Nothing in it would interest a client who never intends to hire you.

It cannot be wrong. A framework that no result could contradict is a taxonomy. If every campaign outcome can be explained after the fact by the same diagram, the diagram is not predicting anything.

It changes shape between pitches. The model has four stages in one deck and six in another, and the names drift. That tells a buyer it is a presentation device rather than something the agency actually runs on.

Nobody outside the agency has ever used the name. No editor has quoted it, no client refers to it internally, no engine associates it with you. It exists only in documents the agency wrote about itself.

The founder cannot explain it without the slide. If the idea does not survive being said out loud in two sentences over coffee, it will not survive a trade editor or a procurement panel either.

Two or three of these together and the framework is doing mild reputational damage, because the buyers you most want are the ones sophisticated enough to recognise a renamed process.

The levers that make a framework believable

Credibility comes from a set of conditions that either hold or do not, and most agencies are strong on one and absent on the rest.

1. A premise someone could disagree with. The framework rests on a claim about the market that a reasonable competitor might reject. Consensus positions make poor IP because they are available to everyone. The discomfort of taking a side is the price of owning something.
2. Evidence the agency holds and others do not. Proprietary research, a pattern across a body of client work, a dataset built over years. The evidence is what lets the claim survive a sceptical question, and it is also the hardest part to fake, which is exactly why buyers look for it.
3. A named owner who argues for it in public. A framework attributed to "the agency" is attributed to nobody. A named strategist or founder who defends it in bylines, on stages and in comment is what gives it accountability, and accountability is what makes it quotable.
4. Use by people who are not you. A framework becomes believable at the point a client, an editor or another practitioner uses its name without being prompted. Until then it is a claim. Afterwards it is a reference.
5. Consistency across every surface, for years. The same name, the same premise and the same definition on the website, in the credentials deck, in bylines and in the founder's LinkedIn. Editors and AI engines both learn associations through repetition, and they learn nothing from a framework that is described differently each time it appears.
6. Narrow scope. Credible IP explains one thing well. A framework that claims to account for the entire marketing process is, by construction, a description of the entire marketing process, which brings you back to the renamed workflow.

Each of those is a lever, not a step. Pulling any one of them properly takes months, and the sequencing between them - which evidence to build first, whose name goes on it, when it is ready for a trade editor - is where most of the difficulty sits.

Where owned thinking earns its keep

In the pitch. A framework with a real premise gives a buyer something to take back to colleagues who were not in the room. It changes the comparison from which agency seemed most capable to which agency understood the problem, and that is a comparison an independent can win against a larger network.

In trade press. Editors run arguments, not service descriptions. A framework with a defensible premise gives a byline its spine and gives a journalist a reason to call for comment when the category moves. It is the difference between pitching a topic and pitching a position, which is the core of how a trade-press pitch actually lands.

In AI answers. Engines cite named concepts that are consistently attached to a named entity across multiple independent sources. A framework that appears only on your own website is unlikely to be surfaced. One that has been quoted by trade titles, referenced by clients and described the same way everywhere becomes part of how the engine understands what your agency is - which is what an AI entity audit is designed to test.

The same asset serves all three. That is why Last describes Salt's models as its best new-business tools, and why it compounds in a way that individual campaigns do not.

How IP relates to positioning

Positioning says who the agency is for and what it will be chosen for. IP is the evidence that the agency understands that world better than anyone else.

The two have to point in the same direction. A framework about retail media from an agency positioned on B2B technology is a distraction, and it confuses the buyers and engines trying to work out what you do. The best frameworks read as the natural expression of a positioning that was already sharp, which is why the positioning statement usually has to be settled first.

Owning thinking is different from launching a capability

These get confused because both involve an announcement and a name.

A capability launch asks the market to believe the agency can now do something it could not do before, and the burden of proof is delivery - a named expert, a first client, visible work. That problem is covered in launching an agency capability that gets believed.

Proprietary thinking asks the market to believe the agency knows something. The proof is different: evidence, a premise that holds up under pressure, and other people using the idea. An agency can launch a capability without owning any thinking, and plenty do. The reverse is rarer and more valuable, because a genuinely owned idea tends to create demand for capabilities the agency has not yet had to launch.

Why year two beats year six

Last's timing regret is the most useful part of his argument for anyone running a young agency.

The evidence underneath good IP accumulates with client work, and the association between a framework and an agency accumulates with time in public. Both compound. An agency that starts in year two has four extra years of pattern, citation and repetition by the time it reaches year six, and that lead is very difficult for a later entrant to close.

The mistake is waiting until the agency feels large enough to deserve a point of view. The agencies that end up owning an idea usually committed to it before they were sure, then spent years making it true in public.

Building that is part writing, part research and part discipline about what the agency will and will not say, and our method treats it as one programme rather than a branding exercise.

If you suspect your agency's framework is a workflow with a name, that is a useful conversation to have.