Our first customer. The next test begins.

September 24: our first paying customer.

The creative work became real campaigns and a live buying experience. On September 24, a customer purchased a 12 × 18 inch Ghost Protocol canvas for $49 plus $15 shipping. Payment was verified in Shopify and the matching Printful order is waiting for fulfillment. Shopify recorded an organic Google visit to the About page; Meta also reported a purchase. These are overlapping signals, not proof that one channel caused the sale.

One paid order proves someone bought the product. It does not prove profitable acquisition, a winning ad, or repeatable demand.

Follow this order through delivery, reconcile attribution, and work toward 10 distinct paying customers before scaling a demonstrated winner.

The brief was bigger than “make some ads.”

Once there was a brand and a body of artwork, the next question was how a stranger would discover it, understand it, and decide to buy. AI developed a marketing plan that connected the customer’s interest, the product presentation, the channel, the landing page, and the economics of an order.

Ethan supplied direction and reviewed the output. The work produced research, a structured strategy, and a working creative library. This page explains that thinking and the checks behind it. The actual ads, campaign copy, targeting instructions, and operating settings stay private.

Research started with observable evidence.

The research combined public advertising-library records, a sample of social posts, profile presentation, and selected motion studies. The retained inventory covers 123 advertising records across four canvas and wall-art brands, plus 24 sampled social posts. Selected videos received closer frame-level analysis; not every video in the inventory was watched frame by frame.

The useful questions were about presentation: what catches attention first, how clearly the artwork reads, when a room or person helps establish scale, and what information reduces uncertainty. We separated those observations from hypotheses about why a format might work.

Public visibility has limits. An active ad listing is not proof of a profitable campaign, and likes are not sales. We did not have the other businesses’ acquisition costs, spend, purchase conversion rates, or return on advertising. The plan therefore treats the research as a starting point for tests, not a shortcut to claiming a winning formula.

Start with a reason to want the artwork.

The customer work focused on situations and interests: someone finishing a workspace, looking for a personal gift, or choosing a focal point for a room. Those are hypotheses about purchase motivation, not established demographic facts about INKRALLY customers.

From there, the plan connected the emotional appeal of a piece with practical reassurance. A buyer may like an image but still need to understand its scale, material, edges, and how it will look in a real setting. The storefront and product photography have to answer those questions as clearly as the first message attracts attention.

One business, several ways to be discovered.

Each platform has a different job. The strategy is to match the reason someone is there with the way we show the canvas, then connect that promise to the next page. These are the approaches being tested, not proven recipes for sales.

The channel cards below separate the intended customer journey from the latest verified operating state. Status is a September 18 snapshot. A prepared campaign is not the same as a delivering campaign.

Meta: make the artwork stop the scroll.

The discovery hypothesis is straightforward: someone may not be searching for wall art, but a collection that reflects their interests can earn attention. We use artwork-led videos and a browsable carousel across Facebook and Instagram. The opening needs to communicate the artwork quickly, while the surrounding message makes clear that this is a physical, ready-to-hang canvas.

The videos point to the collection they show. Carousel cards point to the individual products. That keeps the promise of the ad intact after the click: a person interested in a specific piece should not have to find it again inside a general catalog.

This is a purchase-oriented test. The delivery goal is website purchases, not video engagement. When the initial spread of creative produced weak shopping signals, AI reviewed individual ads and placements, reduced the active test to a smaller group, and improved mobile buying visibility. Ethan’s challenge about traffic quality changed the next test.

The immediate diagnostic question is whether people arrive, look at more artwork, choose a size, add to cart, and reach checkout. A video that earns views without moving people along that path is not a sales winner. Meta has delivered traffic, but genuine purchase attribution and sustainable acquisition are not established.

Pinterest: help someone picture the room.

Pinterest starts from a different motivation: collecting ideas for a space. Here the canvas is shown within an office, bedroom, or living room so the viewer can judge mood, scale, and how the artwork relates to furniture and wall color. The concept is less about a fast edit and more about a room someone wants to create.

The prepared campaign uses room-photo Pins linked to the exact featured artwork. The first test is designed to learn whether those settings bring interested visitors to the product pages. It uses an outbound-visit objective; that should not be confused with the purchase optimization used for Meta.

AI selected the room assets, prepared platform crops, checked that the artwork remained readable, wrote product-relevant descriptions, and connected the destinations and measurement. Ethan handled account and billing steps that required the owner. Approval has been reported, while actual campaign delivery remains unconfirmed in this update.

Google: meet a shopper who is already looking.

Shopping ads are built around the product catalog rather than a beat-cut film. The product image, title, price, availability, and destination all help describe what is for sale. The intended match is a person looking for a type of canvas and a product that actually fits that interest.

AI prepared the catalog connection and a curated pilot, checked product pages, and worked on descriptive collection and guide pages for organic discovery. The goal is relevance: space art should lead to space artwork, and the image and price in a listing should agree with what the shopper sees on arrival.

This channel is blocked at the account level in the current snapshot. Google Merchant has flagged misrepresentation. We have not established the specific cause or a successful remedy, and we are not presenting earlier product approvals as proof that Shopping is delivering. Fixing eligibility is a different task from optimizing an ad.

OpenAI Ads: test discovery in a relevant conversation.

The hypothesis is that someone discussing a workspace, an interest, or a gift may be receptive to a carefully matched art selection. AI prepared themed landing pages that give those visitors a focused starting point rather than asking every person to browse the same full catalog.

This is an emerging-channel experiment, separate from Meta and Google. Campaign setup, account verification, billing, review, and delivery have been distinct steps. Support is investigating the delivery issues. Until actual delivery and shopping activity are confirmed, the strategy remains a hypothesis with prepared destinations, not a working source of customers.

Organic content builds familiarity alongside the paid tests.

The organic plan draws from the same approved artwork library but serves a different rhythm: collection films for discovery, swipeable selections for browsing, and room scenes that explain how a canvas can live in a space. Platform-ready files and the review library make it possible for Ethan to approve a piece before it is posted.

Preparing a library does not mean every post has been published. Instagram and Pinterest are the focus of the organic work; TikTok setup did not become a working publishing channel. Email is a follow-up opportunity for people who explicitly opt in, not a license to message every visitor.

The experiment’s audience is also separate from the art brand’s customer audience. Someone watching Ethan build the business may enjoy the process without wanting a canvas. We should learn from those visits without counting followers or lab watchers as evidence of product demand.

How AI turned approved artwork into the ad creative.

The starting material was the reviewed artwork, not a freshly generated approximation of it. Ethan’s feedback established which pieces to use, how fast the edits should feel, and which early attempts did not meet the standard. GPT-6 handled research, asset preparation, composition, and editing through image, browser, and video tools.

For motion, AI studied selected references frame by frame. Widely spaced screenshots had made the original interpretation too slow. Closer inspection revealed a stable product position, quick artwork changes, and background contrast that adapted to the image. Those observations became an original editing approach using INKRALLY’s own art.

The finished artwork was placed on a consistent digital stage. Soft, art-derived background colors separated it from the surroundings while keeping the whole composition visible. Cuts were aligned to musical beats, and alternate aspect ratios were deliberately recomposed rather than simply cropping off the artwork. These are edited films assembled from still art, not generated footage of a canvas moving in a room.

The room-photo direction used generated environments and product mockups to help explain scale and atmosphere. The selected artwork had to remain faithful to the original inside the scene. Unrealistic people, awkward handling, mismatched art, and unconvincing packaging were reasons to revise, not details to excuse. Those scenes are illustrative product imagery, not customer photographs or proof that a physical photoshoot occurred.

Carousels made the same library browsable one piece at a time. Each card needed a legible canvas and a destination that matched it. Stills, carousels, and films were organized for review so Ethan could approve the work and give feedback before use.

The final checks covered the exported file, not just the editing timeline: does the full artwork fit, do cuts still land with the sound, is text inside the usable area, and does the destination match? Ethan selected and approved the creative direction; tools rendered the assets and exports. Music suitability and permission remain separate from whether an edit sounds good.

1. 01

   Research

   Study public formats and selected edits. Treat apparent success as a hypothesis.

2. 02

   Choose

   Use approved INKRALLY art. Ethan directs the selection and reviews the work.

3. 03

   Compose

   Build an artwork stage, a product carousel, or an illustrative room scene.

4. 04

   Edit

   Align motion cuts with sound; adapt each format without losing the art.

5. 05

   Verify

   Inspect exports, readable artwork, crop, rhythm, and matching destinations.

6. 06

   Learn

   Compare actual shopping behavior, then revise the next test.

A sale has to leave something after the costs.

AI built an economics model around the cost of fulfilling an order, shipping, payment processing, and an allowance for problems such as refunds or reprints. That helped separate revenue from the amount left to support customer acquisition and the business itself.

The model used recorded provider quotes and clearly labeled assumptions. It was a planning tool, not observed profit. The broader lesson was to evaluate an offer and acquisition cost together: an appealing discount or a high revenue number can still leave too little contribution. We are keeping the actual thresholds, budget allocations, and campaign decision rules private.

Make the result explainable.

The creative work was organized around distinct hypotheses about attention and product understanding. The purpose of an experiment is to learn something specific, with the destination and measurement aligned to that question. A large folder of assets is not, by itself, a test plan.

A reusable production structure let AI change the artwork selection, opening, or format while preserving the approved source images. Ethan reviewed the results in a creative library. The overview below explains the production stages; it does not publish our exact prompts, audio files, timing maps, or manufacturing settings.

The strategy now is deliberately restrained: keep a coherent set of creative ideas, send visitors somewhere that matches the art they saw, and observe what they do next. A room image tests context; an artwork-led video tests the appeal of the collection. Neither earns more investment just by attracting inexpensive clicks.

Agree on what counts before looking at a dashboard.

The measurement plan follows the path from product interest to a completed order. It accounts for the fact that the custom storefront and hosted checkout are different surfaces, and that retries or repeated events must not become duplicate purchases.

It also separates traffic from Ethan’s audience from paid acquisition. A YouTube viewer arriving to follow the experiment has a different starting point from a first-time shopper seeing an ad. Lab watchers, storefront interest, paid customer orders, and repeat customers should not be collapsed into one success number.

The latest review found measurable browsing from Meta traffic, with Facebook visitors showing deeper recorded engagement than Instagram visitors. Those small samples are directional evidence, not a channel winner. Paid visits had not produced recorded carts or checkouts in the reviewed window, and completed-purchase attribution remains unproven.

GA4 sessions and Meta landing-page views still do not match one for one. They have different definitions, and previews, returning visits, and incomplete attribution can complicate comparisons. We keep that uncertainty visible instead of adding the numbers together or calling every event a customer.

The product is real. Finding customers is the current test.

The physical samples arrived and Ethan approved their quality. Now our first customer order is paid and waiting for fulfillment. Delivery and repeatable demand are the next checks. Earlier platform snapshots below describe the conditions at the time; they are not live account statuses.

Scaling is a later phase. A campaign needs verified purchases and economics that still work after production, shipping, payment fees, and customer support. A promising click-through rate is useful for diagnosis but cannot establish a profitable business.

The lab will distinguish what was prepared, what was approved, what actually delivered, and what customers bought. Channel interruptions and unsuccessful tests belong in the story alongside the progress. Exact spend limits, targeting, creative recipes, and operating decisions stay private.