The same piece of machinery can be built two different ways. Same code. Same infrastructure. Same AI engine. One produces noise. The other produces signal. The difference is not the machinery. It is what gets fed in.
This is the Noise Machine Test. It is a diagnostic for businesses running AI content operations that wonder why the output does not sound like the brand. Why consistency breaks down. Why scaling hurts more than helps. The answer is almost always the same: generic inputs in, generic outputs out.
The test has four parts. They correspond to the four quadrants of content strategy. Before the test runs, there is a meta-question. Everything else is secondary to it.
The Meta-Question: Does a System Seed Exist?
If nobody can point to a single document, or a structured data set, that contains the organisation's validated expertise, brand voice, verified facts, and customer understanding, then the entire operation is running on generic inputs. That is the first diagnostic.
A System Seed is not a brand bible. It is not a style guide. It is not a list of talking points. It is the actual source material. The things the organisation knows. The position it holds. The data it has verified. The way customers actually think and talk. The mistakes they make before they hire it. The beliefs they hold that stop them from solving their problem.
If that material does not exist in one place, it is scattered. If it is scattered, the AI engine cannot find it. So the AI defaults to the same generic training data it uses for everyone else.
This question gets answered before any of the tests below. If the Seed does not exist, the diagnosis is simple: fix the Seed first. Everything else is temporary.
Testing Q1: Is the Strategy Genuine?
Q1 is quadrant one. It is where strategy lives. It is the position the organisation takes about its industry, its market, its customers, and the problem it solves.
Can the core position be stated in one sentence? Not a tagline. Not marketing language. A real belief about the industry. An example: "Most people don't understand their finances because the industry made it incomprehensible." That is a position. It is a claim about the world. It is something the organisation believes is true. In contrast, "We help people with their finances" is not a position. It is a category description. It describes what the organisation does, not what it believes.
The question is whether a position exists at all. Not whether it can be articulated beautifully. Just whether it exists.
The second part of the test is harder. Would the competitors say the same thing? If yes, it is not a position. It is a category description. Every mortgage broker says they help people with mortgages. That is not a position. That is what the category is called. The System Seed requires something only one organisation believes, or at least something it believes more deeply than anyone else.
The third part is the honest part. When was the last time Q1 thinking made the leadership team uncomfortable? Real strategy requires taking a position. Positions are uncomfortable because they exclude people. They create friction. A position so comfortable that nobody could disagree is not a position at all. It is consensus, which is another word for generic.
Testing Q2: Seeded or Generic?
Q2 is quadrant two. It is where the AI runs. It is where strategy turns into content.
Take the last five pieces of AI-generated content. Remove the brand name. Could they belong to a competitor? If yes, Q2 is running on generic inputs. This is the easiest test to fail and the most useful diagnostic. It reveals immediately whether the machine is producing noise or signal.
The difference is not in the AI engine. The difference is in what gets fed in. There is a fundamental gap between "write a blog post about mortgage rates" and "write a blog post from the position that most people don't understand their finances because the industry deliberately made it incomprehensible, using our verified rate data and in our specific tone." Both go into an AI. One produces noise. The other produces signal.
The test is specificity. Look at the prompt. Look at the brief. Look at what the AI is actually being given to work with. Generic inputs produce generic outputs. Seeded inputs produce content that sounds like someone. Content that could only come from one organisation. Content that, if a customer read it, would be recognisable before they saw the brand name.
If a team cannot describe the difference between its last piece of content and a competitor's last piece of content, it has not seeded Q2. It has just used the AI as a faster way to produce what everyone else produces.
Testing Q3: Is Validation Real?
Q3 is quadrant three. It is where someone reads the output before it goes live. It is where the machine's output gets checked against intent.
When was the last time someone read a piece of AI-generated content before it was published? Not skimmed. Read. With the question "would we say this?" in mind. This is the validation step. It is where the System Seed meets the generated content and someone asks: does this represent us accurately?
Most organisations have a checkbox here instead of a process. A checkbox is "looks fine, publish." A process is: check the facts against the Seed. Check the voice against the Seed. Check the logic against first principles. Check whether a customer would recognise the thinking as coming from this organisation. These steps take time.
A Q3 that takes less than five minutes per piece is not validation. It is rubber-stamping. Rubber-stamping at scale produces noise at volume. The moment when the AI misinterpreted the position does not get caught. The moment when the voice drifted does not get caught. The moment when a fact got slightly wrong and now the content contradicts earlier thinking does not get caught.
The harder test: is there an actual validation process, or just a person who looks at things and says yes? There is a difference. A process is repeatable. A process has steps. A process produces consistent output because the same thing happens every time. "The founder reads it" or "whoever is available checks it" is not a process. It is a person. When that person is busy, or not thinking clearly, or has read the same thing seventeen times that day and their attention drifts, the validation fails.
Testing Q4: Is Deployment Coherent?
Q4 is quadrant four. It is where the content actually lives. Website. Email. Social. Wherever the audience encounters the brand.
If pricing changed tomorrow, how long would it take to update every asset? If the answer is "weeks," there is no deployment system. There is Reactive Churn. The website gets updated. The email sequences get forgotten. Someone else updates the sales page. Nobody tells the social team. Six weeks later, ads are still running with the old pricing.
The test is coherence. Do all the channels say the same thing right now? Check the website against the email sequences against the social profiles. If they disagree, Q4 is fragmented. Fragmented deployment means the audience gets mixed signals. They see one thing on the website and another in the email. They conclude the organisation is disorganised or untrustworthy because the messages do not align.
Coherent deployment does not mean everything is identical. It means everything says the same thing. Website and email and social should all be expressing the same core position, the same offer, the same thinking. The form changes. The medium changes. The message stays consistent.
The Patterns: What Usually Goes Wrong
Most businesses running AI content operations fall into recognisable patterns. Knowing which pattern applies tells the team exactly what to fix first.
Pattern One is no System Seed at all. Everything is generic. The AI gets fed brief instructions. The output is usually fine but indistinguishable from everyone else's output. The organisation has decided to run on volume instead of signal. Fix Q1 first. Nothing else matters until the Seed exists. There is nothing to validate against. There is no standard for consistency. Build the Seed. Then everything else becomes possible.
Pattern Two is System Seed exists but Q3 is skipped. The organisation has figured out its position. It has built its Seed. But nobody is actually reading the output before it goes live. The errors are scaling. The content might be good but nobody knows because nobody is checking whether the AI interpreted the Seed correctly or whether the output drifted off-brand. Fix Q3 before scaling Q2 further. One person reading each piece. One person checking facts and voice and logic. It is the cheapest insurance policy available.
Pattern Three is Q1 and Q3 are solid but Q4 is fragmented. There is a real position. There is good content. But the channels do not talk to each other. The website says one thing. The email says another. The social is inconsistent. The content is good but the deployment is scattered. The audience gets confused because the same product appears with different messages. Fix the deployment coordination. Appoint someone who knows all the channels and makes sure they are saying the same thing.
Pattern Four is everything runs but the Seed is stale. Q1 has not been updated in months. The operation is producing content from outdated thinking. New customer insight surfaced last quarter. The position has shifted. New data is available. The System Seed is still pointing to the old truth. The content machine is running on information that is no longer accurate. Refresh Q1 regularly. Quarterly at minimum. The Seed is not a document written once. It is a document maintained.
The Honest Assessment
Most businesses running AI content are somewhere between Pattern One and Pattern Two. No Seed, or Seed exists but nobody actually checks the output before it goes live. That is the reality.
Fixing it is not complicated. It is disciplined. No new tools. No bigger team. A process.
Start with the System Seed. Take what is actually known. Position. Data. Customer understanding. Get it into one document. Get it structured. Then run one cycle through the framework. Feed the Seed to the AI. Let the AI generate something. Have someone read it. Ask: does this sound like us? Does this represent what we believe? Fix it if it does not. Then publish it. Then repeat.
The difference between the first piece generated from a Seed and the hundredth is the difference between noise and signal. The machine has not changed. The fuel has.
This is the framework, lifted clean from the businesses where it was built. Marketing Curious: Working the Noise traces the origin: the diagnostic patterns identified across dozens of operations that thought they were running AI content well and discovered which quadrant was actually broken. This page is the tool. The book is the receipt.
Part of the Marketing Universe. Explore Traffic Plus Offer : The Trust Algorithm : Opportunity and Authority.