Ecom landing pages skill

Ecom Landing Pages

Turn a store brief into two deliverables: a ranked list of landing-page concepts, and a build blueprint for the ones worth shipping. The method was reverse-engineered from Mill.com, a DTC brand running ~40 production /lp/ pages off a single product. That spread is the whole idea in one sentence: one product becomes many landing pages by varying the angle, not the product. A person who clicks a "feed your chickens" ad and a person searching "nyc curbside compost" want the same device but need different first sentences.

This is the one platform-neutral skill in the collection: landing-page thinking is the same on any CMS, so the ecom- prefix, not shopify-. Execution notes that differ by platform are called out where they matter.

When this fires

• A store needs landing-page concepts for paid traffic, a campaign, a new segment, a season, or an event.
• A concept got chosen and needs building into a real page.
• An existing landing page needs an audit against best practice.

Not for full-site IA, PDPs, collection pages, or blog/SEO articles: those live inside the browse flow and compete for attention. A landing page has one goal and strips the chrome. Different rules.

The loop

1. Intake from the live storefront

Read the store's real site: homepage, key product pages, collections, the sitemap if it helps. Never invent products, audiences, or offers: an empty intake field is a signal, not a license to make something up. Pull:

• Product / offer: what sells, price, the single most important thing it does.
• Core value prop: the one-sentence promise.
• Distinct audiences: buyers segmented by identity, not demographics.
• Top pains: the problems that drive purchase, in the customer's own words.
• Use-cases: the different jobs the product does.
• Geos / markets: cities or regions with concentrated demand or local programs.
• Seasonal / event calendar: sale moments, in-person events, launches.
• Offers: trials, guarantees, discounts, bundles, financing.
• Gift / occasion fit: is it giftable? which occasions?
• B2B / segments: institutional buyers with a longer cycle and different criteria.
• Partner / affiliate channels: creators, trade partners, referrers.

2. Generate against the 8 archetypes

Every landing page is one of these angles. Walk each one against the intake and ask "is there a page here?" A page can stack archetypes; lead with the dominant one, because the headline only gets to make one promise.

#	Archetype	When to reach for it
1	Audience / persona	A distinct buyer identity with its own motivation; the headline names them
2	Use-case	The product does several jobs and a segment cares about exactly one
3	Pain / emotional reframe	Cold top-of-funnel traffic that doesn't know the category yet
4	Location / geo	Geo-targeted ads, local programs, regional logistics or PR
5	Event / seasonal / promo	A campaign window, a sale, or an in-person moment; urgency-driven
6	Offer / trial / guarantee	Conversion-stage traffic that needs a nudge or risk reversal
7	Occasion / gift	Holiday and gifting; reframes the reader as the giver
8	Segment / B2B + partner	Institutional buyer or referral partner; metrics and process, not lifestyle

Push every archetype at least once before ranking. The failure mode is generating the same angle ten times. The full taxonomy with Mill examples and the stacking rules is in references/build-blueprint.md.

3. Key each candidate to a traffic source

Traffic source is the primary angle selector. It dictates the angle and the headline's job. Drop any idea that can't name a source: a page with no campaign behind it is an orphan.

• Paid social, cold → pain/emotional or audience. The viewer wasn't looking for you; earn the click with a felt problem or an identity hook.
• Search, intent → use-case, geo, or offer. Match the query language exactly.
• Local / geo ads → location, with local proof and local logistics.
• Email / retargeting, hot → offer/guarantee or bundle. They know you; remove friction.
• Partner / affiliate → segment/partner, usually a different conversion goal (a sign-up, not a purchase).

4. Prioritize on 4 axes

Score each candidate 1–5 and rank. Intent match carries the most weight.

• Intent match: how precisely the page answers the traffic source's mindset.
• Demand / traffic availability: real search volume, an ad audience, or a live event to send traffic. No traffic is a no.
• Business value: margin, AOV, strategic segment, or lifetime value of who converts.
• Build cost: copy + assets + (for B2B) a lead-capture/CRM path. Cheaper wins ship first.

5. Output: lead with the ranked table

One row per concept. This table is the deliverable: put it first.

#	Concept	Archetype	Target	Core promise	Primary CTA	Traffic source
1	"Feed your flock"	Audience/use-case	Backyard chicken owners	Turn scraps into chicken snacks	Shop now	Paid social
2	"NYC: skip the trash chute"	Geo + pain	NYC apartment dwellers	No more takeout-trash trips	Shop now	Search + geo ads

Then, below the table:

• Deliberately deprioritized: the archetypes or ideas you cut and why (e.g. "geo is thin for a national DTC with no local programs"). Never a silent cap.
• Build first: the top 2–3 with one line each on why they lead.
• Offer to take one concept through the build blueprint. Stop at the ranked list unless asked to build.

Building a chosen concept

When a concept is picked, build it on the 10-section page skeleton in references/build-blueprint.md: hero → benefit pillars → how it works → use-cases → tiered social proof → risk reversal → comparison → FAQ → cross-sell → final CTA. That reference also carries the copy patterns, the archetype-specific structural deltas (B2B swaps lifestyle for metrics; promo moves price up top; partner leads with terms), and the pre-launch checklist. Worked teardowns of live Mill pages are in references/mill-teardown.md.

Red flags: stop and correct

• Inventing the store's products or audiences instead of reading the site → go read the live storefront and re-ground every concept.
• A flat list with no archetype spread, or concepts with no traffic source → you skipped steps 2–3. Force every archetype once; every row names a source.
• A table where everything wins → you're listing, not prioritizing. Rank honestly and state what you cut.
• Jumping straight to building the page → the deliverable is the ranked list first. Build is a separate step, on request.
• Same angle repeated → the whole point is angle variety off one product. If three rows are the same archetype with a new noun, collapse them.

References

• references/build-blueprint.md: full 8-archetype taxonomy, the 10-section page anatomy, copy patterns, archetype deltas, and the pre-launch checklist.
• references/mill-teardown.md: archetype→URL map and section-by-section teardowns of live Mill.com landing pages.

Provenance and maintenance

Last verified: 2026-07-05. The method is reverse-engineered from a public, observable system, so a stranger can re-verify it read-only: visit mill.com and browse its /lp/ landing pages, then map each live page to one of the 8 archetypes above. If the pages still cluster into audience, use-case, pain, geo, event, offer, gift, and B2B/partner angles off a single product, the taxonomy holds. Landing-page best practice drifts slowly; the archetypes and the section skeleton are durable, but re-confirm any Mill-specific claim (page counts, exact headlines) against the live site before quoting it, since a brand can restructure its funnel at any time.