How we work
Audiences first.
Then everything else.
PRISM is the tool I'm building to run The Digital Alchemists on one rule: work out who your buyers are before touching a page or an ad. It pulls together your search terms, reviews, website, analytics and, if you share it, sales data, and turns them into audiences with the evidence attached. Then every page, ad and line of copy is checked against them, partly by hand for now.
Written by Dan Picken, founder. Last checked 2 November 2026.Why audiences first
Most marketing starts
with the tool.
Most marketing starts with a channel. 'We should be on TikTok.' 'Let's redo the website.' 'Boost that post.' The audience comes up later, usually as a persona with a stock photo that nobody opens again.
That's backwards. The same page can work for one buyer and lose another. A first-home buyer and an investor can type almost the same search and need opposite answers. Until you know who's buying, and why, every tool is guessing.
Asking buyers isn't enough on its own, either. People are often poor judges of why they chose something: in Nisbett and Wilson's well-known 1977 experiments, people gave confident reasons for choices that the researchers had shown were driven by something else. So audiences here are built mostly from what people do, search and write, with what they say as one input among several.
So the order here never changes. Audiences first. Then the channels where they look. Then what each page and ad should say to them. Then measuring whether it worked, and feeding that back into the audiences.
How it works
Evidence in.
Audiences out.
What goes in
- Search termsFrom Google Ads and Search Console: the exact words people use, and the questions behind them.
- ReviewsYours and your competitors': what buyers worry about and praise, in their own words.
- Your websiteWhat you say now, page by page, so the gaps between what buyers need and what you tell them show up.
- AnalyticsWhich pages and paths lead to an enquiry or a sale, and where people drop out.
- Ad resultsWhich messages people respond to. Used to test the audience picture, not to invent it.
- Sales data, if you share itWho actually buys and what they are worth. Optional: no names, emails or phone numbers needed.
What it's being built to do
- Page scoreEvery page scored on the 8 signals, per audience. A page can work for one buyer and lose another.
- Ad and landing page checkDoes the ad say what this audience needs to hear, and does the page keep the promise? See message match.
- Copy for each audienceHeadlines, ads and page copy written for one audience at a time, using their words.
- AI search checkCan ChatGPT and Google's AI read your pages and answer this audience's questions from them?
- Compliance gapsFor regulated industries, wording checked against the regulator’s rules for health, finance and legal ads. See the advertising rules guide.
What comes out
Audiences you can
check the working on.
Who they are
Each audience described from the evidence: what they're trying to get done, what triggers the search, what they compare, what stops them.
What moves them
The buyer psychology that applies to each audience, like loss aversion or social proof, from the psychology library, with the reason it fits.
Where the evidence came from
Every claim about an audience shows its source and how confident it is. A guess is labelled as a guess.
Who you’re missing
Audiences that are searching, or buying from competitors, that your site and ads never speak to.
The psychology is research, not my invention. Each principle links to the studies it comes from in the psychology library, and only stays in if it holds up against real results. I'm a marketer, not a psychologist.
Your turn
How well do you
know your buyers?
Tick the evidence you already have. It shows how much of your picture of your buyers is evidence and how much is guesswork, and what each missing piece would add.
What do you have?
Your audience picture is
Guesswork
The biggest gaps
The weighting is my rule of thumb for how much each source tells you about who buys and why. Nothing you tick leaves your browser.
What I see most often
Where audience work
goes wrong.
- 1
Personas from a workshop
Written from what the team believes, not from what buyers search, say or buy. They feel right and change nothing.
- 2
Audiences by demographics only
'Women 35 to 54' says who might see an ad. It says nothing about why they'd act, or what would stop them.
- 3
One message for everyone
The home page tries to talk to every buyer at once, so it speaks clearly to none of them.
- 4
The search terms never read
Months of buyers typing exactly what they want, sitting in a report nobody opens.
The honest bit
Still being built.
On purpose.
PRISM isn't finished, and it isn't for sale. Parts of it work, parts are being tested, and anything that doesn't hold up against real results gets cut rather than kept for the sales pitch.
The approach doesn't wait for the software. The same steps, done partly by hand, go into every free audit, every piece of buyer research and every Market Blueprint. PRISM makes them faster and more thorough.
And it won't fix a weak offer. If buyers don't want what you sell at the price you sell it, knowing them better just tells you that sooner. Which is still worth knowing.
Questions
people ask.
What is PRISM?
PRISM is the audience-first tool I'm building for The Digital Alchemists. It works out who your buyers are from your own data and what's public, then checks your website, ads and copy against each audience. It's still being built, and it isn't sold separately.
Can I buy or try PRISM?
Not yet. It's built for my own client work first. The same thinking goes into every free audit and every Market Blueprint.
How is this different from buyer personas?
Most personas are written in a workshop and given a stock photo and a made-up name. Here, every line about an audience points to the evidence behind it, and how confident it is. If the evidence changes, the audience changes.
Do you need my customer data?
No. Search terms, reviews and your website go a long way. Sales data makes it much sharper, because it shows who actually buys rather than who visits. If you share it, names, emails and phone numbers aren't needed.
Is the AI making things up about my customers?
That's the risk with any AI-written persona, and the reason every claim carries its source. Where there's no evidence, it says so. A labelled guess is useful. An unlabelled one isn't.
Related
Keep
going.
Start with
your buyers.
Takes a minute. Audits come back within 1 to 2 business days. For anything else, I'll get back to you as soon as possible.
What happens next
- 1I'll ask for read-only access, or just your website for a search audit.
- 2I go through it and write up the 3 biggest fixes.
- 3A 20-minute call to walk you through it. No pitch unless you ask.