claude skills / E-commerce pack / pub-produit / en français

Product Ad.

Three new hooks every week, drawn from your customers' words, to replace ads that wear out.

See what it produces
  • Meta Ads
  • Instagram
  • TikTok
  • YouTube Shorts
  • Phone
3hooks per week, on three different angles
2 sto stop the scroll, 8 words at most, understood without sound
1single variable tested: the hook, the rest of the script is shared

orders of magnitude, depending on your volume

the problem

What gets in the way today.

  • 01The ad that worked last month does nothing now, and nobody has a fresh idea.
  • 02Everything changes from one video to the next, so there's no way to tell what worked.
  • 03The best lines are in customer reviews, and nobody reads them again.

input, output

You give this. You get this.

Customer reviews and last week's results

Atelier Lisière, Fig candle 180 g, product page attached. Test budget: €70 per version.
Reviews: "I've been using the pot as a vase for six months", "I thought it would be too strong, it's very gentle".
Frequent question: "Does it smoke?"
Last week: making-of 31% 3 s view rate, 1.1% click rate, €18 per purchase. Gift 22%, 0.8%, €29. Mood 19%, 0.6%, €34.

$ claude "Prepare this week's three hooks and the test plan."

Product ad: Fig candle, week of 23 Novemberexample

Last week's results

Making-of: 31% 3 s view rate, 1.1% click rate, €18 per purchase. Kept.

Gift: 22%, 0.8%, €29. Stopped.

Mood: 19%, 0.6%, €34. Stopped.

What we learned: hands-at-work footage stops the scroll better than mood footage.

The three hooks

A, moment of use: a hand slips three sprigs of eucalyptus into the empty pot. Text: "I've been using the pot as a vase for six months".

B, objection: close-up on a trimmed wick, a match comes closer. Text: Does it smoke?

C, surprising detail: someone leans toward the pot and smiles. Text: "Too strong? No, very gentle."

Shared script (22 s)

0 to 2 s: hook A, B or C

2 to 6 s: the lit candle, calm flame. Rapeseed wax, cotton wick.

6 to 11 s: hands pouring wax in the workshop. Hand-poured.

11 to 15 s: candle on a shelf. About 40 h burn time.

15 to 19 s: empty pot turned into a vase. Keep the pot.

19 to 22 s: product and name. Fig candle, on our website.

Test plan

Audience: the same as the "making-of" ad we kept | Budget: €70 per version | Duration: 7 days | Read-out: Monday 30 November

Naming: figuier_detail-vase_2026-11-23, figuier_objection-fumee_2026-11-23, figuier_objection-parfum_2026-11-23

Landing page: the Fig candle product page.

the method

What Claude does, in order.

  1. step 1Read the customer material

    Reviews and messages sorted: problem before purchase, moment of use, surprising detail, objection removed.

  2. step 2Choose three angles

    Three angles not tested recently, each with its source line, a real review or a product fact.

  3. step 3Write the hooks and the script

    First two seconds per angle, then a shared 15 to 25 second body, shot by shot.

  4. step 4Plan and read the test

    Same audience, same budget, same duration, clear naming and a read-out date. The winner stays.

demo in French / click, try it

Your turn.

An interactive demo of the skill. It is in French, the pack works in any language: ask Claude to answer in English.

Gros plan, mèche recoupée, une allumette approche

Est-ce que ça fume ?

0:00

the film / 44 seconds, with sound

The pack film.

the pack film

In 8 seconds.

the guardrails

What the skill never does.

  • Every hook is based on a real review or a product fact provided, never on an invented testimonial.
  • No claim without proof and no misleading before-and-after.
  • An "Ad" or "Paid partnership" label for any creator who is paid or received the product.
  • No invented performance thresholds: the three versions are compared with each other and with the best current ad.

Questions

Why change only the hook?

Because if everything changes, you can't tell what worked. The first two seconds change, and the body of the script stays the same for all three versions.

How much budget do I need?

You set it based on your volume. The skill splits the same budget across the three versions and doesn't name a winner before each one has received it.

What if I have very few purchases?

The skill then judges on the 3-second view rate and the click rate, and says so clearly in the results.

Can an actor play a customer?

An acted testimonial is never presented as a real customer. The skill starts from your customers' real words, with written consent from the people filmed.

the pack

In the pack: E-commerce pack.