Know who buys before you write a word

In order of weight:

1. Reviews and questions: your customers' own words about what worried them, what surprised them and why they bought. 2. Order patterns, once Shopify is connected: which products go together, first orders versus repeat orders, basket sizes. Aggregated, never individual. 3. Your product copy and pages: who your brand thinks it sells to. 4. Search queries, once Search Console is connected: what people were trying to solve when they found you.

• A name and a short description you would recognise.
• Share of customers: from orders when connected, estimated from reviews otherwise, and labelled as an estimate.
• Pain points, ranked by how often and how strongly they come up.
• What matters to them when choosing.
• Objections: the questions that stop them buying.
• Their language: words they use, and words to avoid.
• Message angles: two or three ways to answer each pain point, with the proof your store can actually back.
• Evidence: every pain point links to the reviews or data it came from. A pain point with no evidence is dropped.

Fictional skincare store, for illustration.

Cautious first-timer. Has reacted badly to skincare before, reads reviews slowly, wants proof a product is gentle.

• Pain: past products irritated her skin.
• Matters: gentle proof, easy returns, a simple routine.
• Objection: "Will this irritate me too?"
• Angle: show sensitive-skin reviews and the patch test, not the results.

Every email, blog post, ad and page fix is aimed at one persona and one pain point, and shows it on the card: "For the cautious first-timer. Answers: will this irritate my skin?"

Results are then tracked by persona and angle, so Tilly learns which worry, answered which way, actually sells. That learning goes into the next thing it makes.

Personas describe needs and situations. They never infer or store sensitive traits like health conditions, ethnicity or religion. And ad copy never tells the viewer what they are ("Struggling with acne?"). It talks about the product instead, which is also what ad platform policies require.