Alexa — everything we need to ask (2026-08-03)

Consolidates the open questions scattered across DATA-REQUEST.md, METRICS.md, COMMANDS.md ▶ (3) and the Rosebank/online finding from 2026-08-03. Daniel sends these. Agents draft only ([[agents-never-send-to-clients]]).

Why it is staged, not one message

Five asks in one message reads like a form. They are also different kinds of ask — one is about her own site, one is a favour for Daniel's marketing, one is a favour with her professional network — and they get very different answers depending on order.

The sequencing that matters: ask about HPCSA advertising rules before asking for case-study numbers. If her rules restrict practitioner testimonials or quoted case studies, the numbers can't be published anyway and the ask was spent for nothing. Rules first, then data.

Two hard rules, both already recorded: colleagues only — never ask a psychologist about patients, or for an introduction to anyone she may have a clinical relationship with. And POPIA: aggregate counts only, never anything client-identifying.

▶ Message A — send now, alongside the R200 invoice

Practical, all about her own site. Nothing here is a favour.

Hi Alexa — two quick things and one small invoice.

First, I want to check something on the site. It currently says you practise in Cape Town and online, and in one place that your practice runs mainly online. But the page title still leads with "Therapy in Rosebank", which is what Google shows as the headline and what people see in their browser tab.

So that I aim it at the right people: are you seeing clients in the Rosebank rooms as well as online, or is it online only now? And if you are still in the rooms, how would you like that described — the suburb, or just Cape Town?

It matters more than it sounds. If most of your work is online you can be found by people anywhere in South Africa, not only people searching in your suburb, and I would set the site up quite differently for that.

Second, I have set the site up so you can edit the text on it yourself — click a paragraph, type, done, no training needed. If you would like that switched on, send me the email address you want to sign in with and I will set it up.

And the invoice is attached. It is the R200 we agreed, and the monthly care plan starts from go-live.

Also settled by her answer: whether a practice address belongs anywhere. INV-2026-002 currently carries no address, which is correct for an online practice, so nothing needs adding unless she says otherwise.

▶ Message B — a week or two later, once A is answered

The HPCSA question first, on its own, because the answer decides whether the rest of the case study is even possible. Genuinely useful to her too — she gets to set the boundary rather than be asked to approve something already written.

Hi Alexa — a question I should have asked earlier, and you are the right person to ask because you live under these rules.

I would like to write your site up as a short case study on my own site, as an example of what I build for private practices. Before I write a word: what do the HPCSA advertising rules actually allow a psychologist to have published about their practice? I am specifically unsure about client testimonials and about being quoted by name myself.

I would rather build within the rules than ask you to approve something that puts you in an awkward position. Whatever the limits are, I will work to them, and I will show you the page before it goes anywhere public. Happy to leave it out entirely if it is more trouble than it is worth.

If she says testimonials/quotes are restricted: the case study becomes after-only and qualitative. METRICS.md already concludes most before→after rows have no "before" anyway, so that is not much of a loss. Do not fill it with estimates.

If she says quoting is fine, then the numbers ask from DATA-REQUEST.md follows in the same thread. That draft is already written and still good — aggregate counts only, [X] for anything she cannot give, nothing estimated.

▶ Message C — separate, and only when the relationship is warm

The referral-names ask. This is the largest favour of the three and it is about her professional network, so it should not ride along with an invoice or a marketing request.

Hi Alexa — a favour, and a no is completely fine.

I am starting to reach out to a few other practices about the same kind of website work. Some of the names I have come across are people you might know professionally: Rianie De Bruin, Liizl Miller and Dr Annalize Green in Pretoria, Natalie Sadie in Stellenbosch, and Judy Grant in Claremont.

Do you recognise any of them as colleagues? And if you do, and only if you are comfortable with it, may I mention that you suggested I get in touch?

⚠️ Colleagues only. If any of these names turn out to be someone she has a clinical relationship with rather than a professional one, drop that name and do not ask again.

Not for Alexa — Daniel's own calls

• The share card image. og:image and twitter:image pointed at lovable.dev's promo graphic and twitter:site credited @lovable_dev — removed 2026-08-03 (commit 078233e), so shares currently render a text-only card. Candidates already in public/: hero-starlings-poster.jpg, murmuration-poster.jpg. Daniel picks.
• The <title>. Still leads with the suburb. Gated on Message A's answer, then it is an SEO decision, deliberately parked.
• Editing go-live needs the Supabase account (dashboard, never hand-inserted SQL) plus three Vercel env vars — COMMANDS.md ▶ "Berlein editable content".