How to Start Selling AI-Generated Goods | SUZURI vs BASE Comparison and 5 Steps

Selling goods made with AI-generated images is an accessible side hustle that requires no inventory, but once you actually get moving, a common sticking point is deciding whether to start with SUZURI or BASE. This article covers how to turn images created with Midjourney or Stable Diffusion into products, where to sell them so you can keep things sustainable, and how to map the whole process from a beginner's perspective.

My own approach has been to test-list on SUZURI first, then take only the designs that get a real response and develop them on BASE with a cohesive brand identity. This part matters a lot: even small changes to pricing or product descriptions can shift sales significantly. Rather than swinging for a home run, validating what sells on a small scale before expanding makes this side hustle far less likely to fail.

What follows covers five steps you can complete today to get your first listing live, profit calculations factoring in SUZURI's tribun system and BASE's fees, plus the practical details that trip people up: copyright and commercial use rights, tax filing obligations, and workplace policy considerations.

What Is the AI Goods Side Hustle? A Zero-Inventory Business Model

How It Works

The workflow for selling AI-generated goods is surprisingly straightforward. Picture it as a straight line: generate an AI image, place it on a product, wait for an order to trigger printing and shipping, collect your profit. The engine behind this is POD (print-on-demand), where products are printed and shipped only after someone buys. You never need to stockpile T-shirts or mugs in your room.

In practice, you start by creating goods-ready images with Midjourney or Stable Diffusion. Next, upload those images to a POD-enabled service like SUZURI and place them on T-shirts, stickers, mugs, and other items. Once you publish the listing, the service handles production and shipping when a purchase comes in, and you receive whatever profit margin you set. Think of it as "turning design files into products where the factory only moves after an order." It feels closer to selling designs than running a physical goods business.

When I first tested a listing on SUZURI, the biggest win was how low the psychological barrier felt. No need to forecast inventory, no packing supplies to buy, no shipping tasks to stall your evenings. Even listing a single item on a weeknight is a perfectly viable way to start. The scariest part of any side hustle is often not failure but the friction of getting started. With a zero-inventory model, that friction drops dramatically.

According to SUZURI's official guidance, you can upload a single image and start selling from one unit, with SUZURI handling everything from production to delivery. There are no upfront costs or monthly fees, and the platform claims you can go from signup to live listing in as little as two minutes. The profit model is also easy to grasp: you set a tribun (the markup you add on top of the base cost), and that tribun becomes your earnings per sale. The maximum tribun is 5,000 yen (~$33 USD). SUZURI offers dozens of product types, covering staples like T-shirts, stickers, and mugs that are easy to expand into. (Check the official page for the latest item count.)

Market conditions are also favorable. While there are no dedicated public statistics for AI-generated goods specifically, IMARC's research projects the global digital printing market at $30.5 billion USD in 2024, growing to $49.3 billion USD by 2033 at a CAGR of 5.5%. This doesn't map perfectly to POD, but the broader trend toward efficient small-batch, high-variety production is worth noting as the foundation for zero-inventory goods selling.

The important thing here is not to mass-produce AI images blindly but to decide who you're making them for first. Even with cat illustrations, the marketplace, price point, and product type shift depending on whether you're going for relaxed line drawings on everyday stickers or cyberpunk aesthetics on T-shirts. AI image generation is just the entry point; whether something sells is largely determined by how you design the product around it.

How SUZURI and BASE Serve Different Roles

SUZURI and BASE are often mentioned together, but they fill very different roles. SUZURI is a POD-integrated service built for "make it and sell it immediately." BASE is an ASP-type platform (a hosted storefront builder) designed for "own your shop." The former's strength is low listing friction; the latter's is branding flexibility.

What sets SUZURI apart is how short the distance is from image to product. You place your image on T-shirts, stickers, mugs, and more, then publish the listing. Production and shipping are handled for you, so as a seller your focus stays on design and pricing. In my experience, the speed of getting that first product out into the world was incredibly valuable. Rather than waiting until something is 100% polished, publishing at 60% and reading the response is a much better fit for a test-and-learn approach.

BASE, on the other hand, lets you set up an independent online store for free on its standard plan. You can customize your shop's design, refine how products are presented, and build a consistent visual identity. It's better suited to "growing a brand" than one-off listings. BASE reports over 1.7 million shops opened, and the Pay ID app has reached 16 million cumulative registered users. There is some discovery potential through the platform, but the expectation is that you drive traffic yourself through social media and content.

The fee structures differ too. SUZURI uses a markup model where you add tribun to the base cost to create your profit, which is intuitive. BASE deducts fees from your selling price. On the Standard Plan, the main fees are a 3.6% payment processing fee plus 40 yen (~$0.27 USD) per transaction, and a 3% service fee. Put simply, SUZURI is a "stack profit on top" model, while BASE is a "subtract costs from the selling price" model.

If you're using both, think of SUZURI as your testing ground and BASE as your brand headquarters. I found that adopting this order removed a lot of decision fatigue. Testing on SUZURI first reveals which designs work better as T-shirts and which as stickers. Once you have that data, giving those winners a series name and concept on BASE transforms single-item purchases into "I want more from this line" buying behavior.

Who This Works For (and Who It Doesn't)

This side hustle fits well for people who want to test ideas without committing money upfront. With platforms like SUZURI charging zero initial costs and zero monthly fees, expenses don't pile up before you've made a sale, and it's easy to set a walk-away point. In a side hustle, this ability to start small and validate cheaply is a major advantage.

It also works for people with little or no design experience. AI image generation handles the visual heavy lifting, and SUZURI's templates handle the product formatting. You'll still need an eye for adjustments as you go, but the entry point is far lower than dealing with print shops and inventory management from day one. For office workers who can only carve out time on weeknights or weekends, not having to deal with shipping is a significant relief. At the supplemental-income stage, a setup where you only need to protect your creative time is much easier to sustain.

On the flip side, this is not a fit for people who want to buy in bulk and ship at scale. POD lets you sell from a single unit, but the per-unit economics don't compete with wholesale purchasing. If you're planning for event sales or wholesale distribution and need tight control over unit costs, a different commerce model makes more sense.

It's also a limited fit for people who want to sell high-priced one-of-a-kind pieces. AI-generated goods are inherently reproducible. The value proposition is different from original artwork or handmade items where uniqueness is the selling point. You can build a brand, but the business model is fundamentally distinct from "only one of these exists."

One more factor that's easy to overlook: people who can't dedicate time to marketing and iteration. SUZURI is easy to start with, but listings don't sell themselves automatically. With BASE, this is even more true. You need time to refine your shop, think about how you present things on social media, and adjust based on what gets traction. AI goods selling is a relatively low-effort side hustle, but it's not a set-and-forget one.

This side hustle sits at the intersection of creative work and commerce. You get the satisfaction of making things, without carrying inventory or handling shipments. In return, the design decisions that matter most are about who the audience is and where you present the work. People who can create AI images with a sales perspective in mind are the ones who get the most out of SUZURI and BASE.

SUZURI vs BASE Compared | Which Should a Beginner Start With?

Comparison Table

Both SUZURI and BASE are services for selling online, but they serve very different functions. SUZURI is where you upload images and turn them into zero-inventory goods. BASE is where you build your own shop and control the full presentation. It's completely natural for beginners to feel torn, and the better question isn't "which is superior?" but "which one has the features I need right now?"

The numbers make the personality difference even clearer. SUZURI's creator guide highlights zero upfront costs and a two-minute path from signup to live listing, with profits built through the tribun system. The tribun cap is 5,000 yen (~$33 USD). (Reference figures as of June 2024.) SUZURI carries dozens of product types, and expanding from staples like T-shirts, stickers, and mugs is straightforward.

BASE, meanwhile, charges a combined 3.6% + 40 yen (~$0.27 USD) payment fee and 3% service fee on the Standard Plan, as outlined in their fee guide. With over 1.7 million shops opened and 16 million cumulative Pay ID registrations, the platform isn't a complete visibility vacuum. That said, your real sales engine will be your own social media presence, content, and brand-building.

A less obvious but meaningful difference is the distance between you and the buyer. SUZURI keeps operations light: low listing friction, production and shipping handled for you. The trade-off is less room to craft a detailed brand experience compared to BASE. BASE lets you design page layouts, logos, color schemes, product descriptions, and featured collections, essentially building "the store itself." That power becomes especially valuable once you know which products are working.

The Beginner's Starting Pattern

If I had to boil the beginner recommendation down to one line: start with SUZURI to validate on a small scale, then move only the winners to BASE for brand-building. The reasoning is simple. In the early stage, knowing "which themes get a reaction" matters more than knowing "what will sell."

This is exactly how I operate. When I listed several "animal x minimal" themes on SUZURI, only two designs got any real traction. I moved just those two to BASE, aligned the logo, key colors, and description tone, and built them into a brand page. The shift was noticeable: instead of one-off purchases, buyers started wanting to collect the series. Honestly, I didn't change the products much. Unifying the brand identity was what lifted average order value and repeat purchases. That's where BASE shines.

With this staged approach, the validation phase carries zero inventory and zero shipping burden. SUZURI handles production through delivery, so you're not adding operational costs while you're still figuring out what works. Instead of forcing growth on underperforming designs, you focus on what gets saved, clicked, and actually purchased, then decide where to invest next. The more focused your selection is when you move to BASE, the less scattered your self-driven marketing efforts become.

As a starting pattern, this clean two-step structure keeps things simple:

1. List multiple themes on SUZURI with zero inventory and observe which designs and products get traction
2. Take only the proven performers to BASE, refine the presentation, and build them as a brand

This sequence works because SUZURI and BASE have complementary strengths. SUZURI excels at speed of validation; BASE excels at depth of shopping experience. Jumping straight to BASE alone makes it harder to tell whether weak sales are caused by the product, the traffic, or the presentation. Finding your winners on SUZURI first makes everything about running BASE much more manageable.

Fees and platform terms do get updated, so treat any specific numbers here as a snapshot. SUZURI's payout fees and BASE's plan conditions are especially prone to changes, meaning 2026 figures should be verified against official sources. This section is based on the conditions confirmed at the time of writing.

Side Note: Where BOOTH and pixivFACTORY Fit In

If SUZURI and BASE are the main pillars, BOOTH and pixivFACTORY make more sense as supporting options.

BOOTH is a sales channel with strong ties to the pixiv creator ecosystem. It's well-suited to illustration, doujin, and fan art culture, offering natural touchpoints with users who are actively browsing for creative work. It occupies a middle ground between an independent shop and a creator marketplace. If your style aligns with that audience, it's worth considering. The fee structure is straightforward: BOOTH's help documentation lists the base service fee at 5.6% of the product price plus 45 yen (~$0.30 USD).

pixivFACTORY is better understood as a production platform than a storefront. Account creation and product design are free; costs only arise when items are manufactured. It's suited for creators who want to evaluate product quality and specifications before committing. Think of it less as a place to sell and more as a manufacturing backbone you can pair with other channels.

A practical way to think about the division: SUZURI for the fastest path to zero-inventory sales, BASE for your brand headquarters, BOOTH for a creator-focused distribution channel, and pixivFACTORY for quality-first production. In AI goods selling, what matters more than where you list is which work you show in which context. You don't need to lock everything down with your first choice. Thinking in terms of swapping roles across platforms is far more practical in day-to-day operations.

5 Steps to Get Started | From AI Image Creation to Your First Listing

The estimated time for your first listing breaks down roughly as follows: 30 minutes for theme selection, 60 minutes for image generation, 30 minutes for rights verification, 45 minutes for product setup and pricing, and 45 minutes for writing descriptions through to publishing and promoting. Total: around 3 to 4 hours. In reality, the more clearly you've nailed down "who is this for and what am I making" before generating images, the fewer do-overs you'll face. I usually build this flow with the assumption of testing on SUZURI first. It's the easiest place to check whether something gets saved, clicked, or viewed all the way to the product page.

Step 1: Pick Your Theme

The first decision isn't about the artwork itself; it's about who will use this, and in what situation. If you skip this and jump straight to "making cool images," you'll end up with designs that look good but don't land as products. "For people who like cats" is too vague. "A mug with understated cat motifs that brightens up a remote worker's desk without being too cute" gives you a clear direction on style, product type, and tone.

In practice, jotting down three things about your target customer is enough: what they like, what vibe they want to avoid, and where they'll use the product. For example: likes Nordic color palettes, dislikes overtly childish character art, uses it on their office desk. With those three points, your image generation prompts and product descriptions will stay consistent.

The most common stumble here is going too broad. "Animals," "flowers," or "cute" alone leave too many directions open, making it hard to even compare your generated results. When that happens, pick one theme and branch from there. Locking in "retro cafe x java sparrow" and varying the presentation for mugs, stickers, and T-shirts makes it much easier to find what resonates.

Your next action is to distill your theme into a single sentence. If you can articulate "who uses this, in what setting, for what purpose" in one line, you'll have a much easier time in the image generation phase.

Step 2: Generate Images

With your theme set, resist the urge to obsess over one perfect image. Batch out 10 variations to start. Producing a range of directionally different options makes it much easier to spot which ones have commercial potential. Whether you're using Midjourney or a Stable Diffusion variant, treat this stage as generating comparison material, not finished products.

It helps to think about prompts in three buckets: photorealistic, illustration, and pattern. For photorealistic, something like "a calm Nordic-style desk with a white mug featuring a minimal blue bird motif, natural light, clean photographic feel." For illustration: "soft lines, flat colors, muted blue and beige, a java sparrow illustration with generous whitespace, simple and adult-oriented." For patterns: "small motifs arranged in a regular repeat, white background, a continuous design suitable for stickers and fabric accessories."

This is critical: if you're designing for print, resolution and margins need to be on your mind from the start. For product-ready images, aim for 300 dpi as a baseline, and avoid pushing fine details or text all the way to the edges. On T-shirts and stickers, lines that look crisp on screen can appear weak at actual size. SUZURI's standard T-shirt uses 100% cotton at 5.6 oz, a solid everyday weight, so centered motifs with strong visual presence tend to pair well with it.

Print-related roughness is a common trip-up. Images that look fine on screen can show noticeable degradation at product scale. When that happens, re-generating at a higher resolution or running an upscaler to refine the details is faster than trying to fix it downstream. My own experience is that images with post-upscale margin and contour adjustments consistently perform better on product pages than raw outputs.

Your next action: from the 10 options, select the 3 you'd most want to turn into products, and resize them for the item types you're targeting.

Step 3: Commercial Use and Rights Check

Before taking your images to market, verify the commercial use terms of the AI tool you used. Midjourney permits commercial use on paid plans. For Stable Diffusion variants, the licensing depends on the specific model and distributor. Product descriptions generated with ChatGPT are also usable commercially within OpenAI's terms of service, though responsibility for the output rests with you as the user.

The biggest thing to avoid on the rights front is output that strongly resembles existing characters, celebrities, or recognizable works. Even without using a name directly, an image that clearly evokes a specific franchise is a product you should pull before listing. As Japan's Agency for Cultural Affairs outlines in its "AI and Copyright" guidance, AI-generated outputs are not automatically free from copyright concerns. In practice, if an image makes you think "this might look like something," regenerating a new version is faster and safer.

I take this step seriously. The reasoning is simple: fixing something after it's listed is far more costly than reworking it before publication. Designs with high viral potential on social media are exactly the ones where recognizable motifs create the most risk. Even if a design seems likely to perform well, anything that evokes a known source is hard to build into a long-term product.

Your next action: from the 3 candidates, narrow down to 1 or 2 images that have no ambiguity on the rights front.

Step 4: Create Products and Set Prices

Once your images are finalized, it's time to build the products on your chosen platform. On SUZURI, the flow follows the official "How item sales work" guide: upload your image, select items, set your tribun, and publish. At the "just get it out there" stage, this simplicity is a huge advantage. Even just placing a design on staples like T-shirts, mugs, and stickers lets you compare how the same image performs across different product types. SUZURI's mugs hold roughly 310 to 320 ml (~10.5 to 10.8 oz), which is a natural fit for a full cup of morning coffee and pairs well with desk-themed designs. Stickers work best with graphics that read clearly even at a small size, while T-shirts favor compositions where the main motif is easily identifiable from a distance.

For BASE, the process involves registering products, configuring payment and shipping settings, and then publishing. There are more setup steps than SUZURI, but in exchange you get much more control over the overall shop presentation. As covered earlier, BASE pairs best with the approach of moving proven sellers in from SUZURI.

On pricing, avoid the temptation to maximize profit from day one. SUZURI's tribun model means your markup directly raises the selling price, so during the validation phase, keeping it modest makes it easier to separate "this design doesn't move because of the price" from "this design just isn't connecting." When prices are too high, you'll see saves and views but not purchases. My early approach was to set tribun low, then adjust upward only for designs that already showed traction.

Your next action: for each design, pick the two best-fitting product types and prepare them for publishing.

Step 5: Write Descriptions, Publish, and Promote

Using ChatGPT to draft product descriptions speeds things up considerably. But rather than posting them as-is, adding your own specifics about intended use, target audience, sizing, and care notes makes the listing much stronger. For a mug, the description changes depending on whether you're positioning it for an office desk or for quiet mornings at home. Mentioning that while it's microwave-safe, prolonged high heat may affect the printed surface shows thoughtfulness about the actual user experience.

My own observation: descriptions that lead with a usage scenario in the first paragraph tend to get better save rates and click-through. When someone can immediately see who the product is for and where it fits into their life, it becomes personal. "A mug designed for adults who love java sparrows, meant to bring a little calm to your work desk" sets a very different tone than a generic design description. Follow that with design details, sizing, and care notes.

Post-publication visibility differs by platform. On SUZURI, how you tag products affects discoverability, so include not just the motif but also the use case and aesthetic style. On BASE, the overall shop impression matters more than individual listings, so even just aligning the color temperature of product photos creates a stronger brand feel.

For promotion, splitting your content into three posts works better than a single "now available" announcement. Post one showing the rough or in-progress generation, a second with the finished image, and a third with the live product page. This sequence gives people context for the work, which tends to generate more genuine interest than showing only the end result. Showing how something was made converts curiosity into product page visits.

Your next action: publish the product pages, then prepare three separate social media posts covering the creation process, the finished image, and the live listing.

How to Turn a Profit | Pricing Strategy for SUZURI's Tribun and BASE's Fees

SUZURI Profit Calculation and Simulation

SUZURI's profit math is about as simple as it gets. The core formula: selling price = base cost + tribun; your profit = tribun. You decide your earnings per sale upfront and add that amount to the price. One of the first stumbling blocks in any side hustle is not understanding how much you actually keep. SUZURI removes that confusion, which makes it a great place to start. The official tribun cap is 5,000 yen (~$33 USD).

For a concrete example, take a T-shirt with a base cost of 2,000 yen (~$13 USD). With a tribun of 300 yen (~$2 USD), the selling price is 2,300 yen (~$15 USD), and your profit per sale is 300 yen. Sell 100 in a month: 300 yen x 100 = 30,000 yen (~$200 USD)/month. Set the tribun at 800 yen (~$5 USD) and the price becomes 2,800 yen (~$19 USD) with 800 yen profit per unit. Sell 60 at that price: 800 yen x 60 = 48,000 yen (~$320 USD)/month. These are hypothetical numbers, but the takeaway isn't "higher margin means more money." It's that monthly revenue is determined by the balance between volume and margin.

I've actually run something close to an A/B test, listing the same designs at 300 yen and 800 yen tribun without changing anything else about the presentation. At the lower price point, the conversion rate relative to page views moved noticeably. Sales context and design strength obviously matter too, so these numbers should be taken as illustrative. But at minimum, there are products where "just a bit more expensive" is enough to slow purchases. T-shirts, as a category where people readily compare options, are especially sensitive.

Given this dynamic, starting with a modest margin is a highly practical strategy on SUZURI. Begin with lower tribun to reduce price resistance while you learn "does anyone actually want this design?" Then widen your margin only on the designs that prove themselves. For a side hustle income target of 10,000 to 50,000 yen (~$65 to $330 USD) per month: at 500 yen margin, you need 100 sales for 50,000 yen; at 1,000 yen margin, 50 sales gets you there. The latter looks easier on paper, but higher prices typically mean fewer units sold. Prioritizing validation over margin optimization tends to be more sustainable.

Note that SUZURI applies fees at the payout stage as well. As of 2026, specific payout fees and conditions may have been updated, so avoid treating any single figure as permanent. At low sales volumes, payout costs have a proportionally larger effect on actual earnings.

BASE Profit Calculation and Simulation

BASE works differently from SUZURI: you calculate profit backward from the selling price. On the Standard Plan, the working formula is: profit = selling price - cost of goods - payment fee (3.6% + 40 yen / ~$0.27 USD) - service fee (3%) - shipping costs.

For a concrete example, say you're selling a mug with a cost of 1,800 yen (~$12 USD) at a price of 2,800 yen (~$19 USD) on the Standard Plan. Combined fees come to roughly 2,800 yen x 6.6% + 40 yen = approximately 225 yen (~$1.50 USD). That leaves a gross margin (before shipping) of 2,800 - 1,800 - 225 = approximately 775 yen (~$5 USD). At a selling price of 2,200 yen (~$15 USD), fees are roughly 2,200 x 6.6% + 40 = approximately 185 yen (~$1.25 USD), giving a margin of 2,200 - 1,800 - 185 = approximately 215 yen (~$1.40 USD). Both are hypothetical, but the pattern is clear: at lower price points, the fixed 40 yen component hits harder than you'd expect.

A 600 yen price difference produces a margin gap wider than 600 yen once fees are factored in. And on BASE, the entire shop experience, including photos, descriptions, and brand consistency, needs to convince people that your price is worth it. Unlike SUZURI where you ride the platform's pricing context, you're building that justification yourself. Even small pricing details like choosing between 2,180 yen, 2,280 yen, and 2,480 yen can affect perception, but pricing too low means the fixed fee component eats into your margins disproportionately.

To target 10,000 to 50,000 yen (~$65 to $330 USD) per month on BASE, per-item margin is the critical variable. At 500 yen gross profit, 100 units gets you to 50,000 yen. At 1,000 yen, you need 50. The math mirrors SUZURI, but on BASE you need to account for cost of goods, fees, and shipping design in your pricing. That's why expanding product range before you have a clear winner is riskier here. Focus on proving a few items first, then confirming they're profitable after all deductions.

BASE's fee structure and payout terms are also subject to updates, so as of 2026, verify current conditions on the official help pages. Payout-related conditions especially tend to create a noticeable gap in take-home pay when monthly revenue is still small.

The Risk of Overpricing and Finding the Right Range

The stronger the urge to earn more, the more tempting it is to push prices up. This is important: raising prices isn't inherently wrong, but overpricing can reduce purchase rates enough to lower your total revenue. On SUZURI, increasing tribun directly inflates the selling price. On BASE, pushing for thicker margins widens the gap between your listing and comparable products elsewhere.

Especially in the early stages of AI goods selling, purchases are more likely driven by "this design is cool" or "this would be useful" than by brand loyalty. If your pricing gets ahead of your brand recognition, you'll see views and saves but not checkouts. In the lower price range, even a few hundred yen can shift behavior, and when traffic is healthy but sales aren't happening, the issue might be pricing rather than design quality.

When thinking about the right price range, it's less about being slightly cheaper than competitors and more about whether someone can look at the listing and immediately feel the price makes sense. For a T-shirt: "would I grab this for casual wear at this price?" For a mug: "is this reasonable for a small daily upgrade at my desk?" Products can sell at higher prices, but that usually requires a complete package: design quality, series cohesion, shop-wide consistency, and compelling descriptions all working together. For an unknown seller starting out, keeping prices accessible while you build traction produces more reliable validation data.

The practical approach: start with modest margins, read the response, then expand margin only on proven performers. This makes it easy to see whether your side hustle is actually viable, and it avoids the failure mode of leaving high prices on products nobody's buying. Profit isn't something you maximize in a single move. It's something you grow once you've found the price range where people actually buy.

Designs That Sell and Promotion Tactics

How to Approach Design

The most common failure in goods selling isn't bad design. It's launching products without a clear picture of who will use them and when. Before anything else, defining scenarios like "a mug a late-20s office worker uses at their desk" or "a sticker someone wants to slap on their laptop after a concert" or "a T-shirt for weekend errands" automatically narrows down color, motif, text density, and sizing. Without that anchor, you end up with a T-shirt that's minimalist, a mug that's loud, and a sticker in yet another style, creating a disjointed shop that makes it hard for any product to build momentum.

When working with AI-generated images, the first thing to nail down isn't prompt engineering technique but the visual identity you're building. Whether it's "muted-tone animal illustrations," "neon-lit futuristic cityscapes," or "Nordic-inspired botanical motifs," committing to a direction makes product expansion much smoother. At this stage, I typically limit myself to around three colors, a single motif family, and one font if text is involved. That constraint is what creates the sense of "this shop has a look" on both SUZURI and BASE.

Keeping your initial product count small is more practical too. Rather than listing dozens of SKUs right away, one design across three products is a much more testable starting point. A T-shirt, a mug, and a sticker is a strong combination: the T-shirt carries the design as wearable art, the mug works as a daily-use item, and the sticker serves as a low-cost entry point. Same design, different purchase motivations. SUZURI's standard T-shirt is 100% cotton at 5.6 oz, which avoids feeling flimsy and pairs well with a centered hero motif that reads clearly in everyday wear.

For mugs and stickers, whether a buyer can picture using the product is what determines the response. SUZURI's mugs hold around 310 to 320 ml (~10.5 to 10.8 oz), fitting a solid cup of morning coffee and looking right on an office desk. Stickers with waterproof labeling invite people to imagine them on tumblers, laptops, and storage cases. Designs that make the use case immediately obvious are the ones that convert.

Effective goods design doesn't need to be artistically complex. For merchandise, what matters is clarity at a distance. If the motif is identifiable in a small thumbnail, the text isn't overwhelming, and the color palette is cohesive, the listing already stands out in a browse view. AI images tend to be information-dense, so rather than using them raw, trimming the background, isolating the hero element, or adjusting for product-appropriate composition is where the real product-readiness work happens.

Getting More Visibility on SUZURI

SUZURI doesn't leave listings completely invisible, which is a real advantage, but visibility still needs to be engineered. The first lever is specificity in tags and descriptions. Using tags that connect to seasonal themes or current trends helps your listings surface for people who are actively searching. In summer, T-shirts with cool-toned colorways; around year-end, mugs positioned for gift-giving. Aligning with seasonal context keeps your products from blending into the background.

Product images also perform better when you use more than one. Showing different angles or context shots, like how the design sits on the product, reduces bounce after someone clicks through. For T-shirts, include an image that conveys the overall look plus one focused on print placement. For mugs, show the front view and something that suggests what it feels like in hand. For stickers, show a placement context. Since SUZURI carries a wide range of items, mechanically duplicating the same image across all of them creates less impact than making small adjustments per product type.

In descriptions, vague words like "cute" or "stylish" are far less effective than specifying where and by whom the product will be used. For a mug: "perfect for your desk," "pairs with your morning coffee routine," "a thoughtful small gift." For a T-shirt: "easy to pair with simple bottoms," "great for weekend wear." Weaving in material facts helps too. SUZURI's standard tee is 100% cotton at 5.6 oz, giving you a concrete detail about wearability. Mentioning a mug's capacity grounds the product as a real object rather than just a picture.

Campaign tie-ins are another underutilized tool on SUZURI. When the platform runs promotions or features, re-sharing the relevant items on social media with that context can boost movement. On platforms with built-in discovery, riding the official wave tends to be more efficient than going it alone. I've had cases where re-promoting an existing product during a campaign outperformed the initial launch post.

How you present work on social media matters just as much. In my experience, combining short behind-the-scenes video of the creation process with the product URL outperforms static image posts for click-through. A finished product image alone tends to end at "that's nice." But when someone can see the progression from rough generation to product-ready design, curiosity translates into clicking through to the actual listing. SUZURI's short purchase path means the "see it, get curious, visit the page" sequence is especially powerful.

Driving Traffic and Building Repeat Buyers on BASE

BASE doesn't generate sales passively, so building visibility into your presentation is essential. The most impactful starting points are product names and category structure. If these are vague, you're leaving search traffic on the table. "Blue flower mug" is far weaker than "Nordic Floral Illustration Mug" as a product name because the latter includes style and use context. For categories, grouping by "desk accessories," "apparel," and "gift ideas" rather than just "T-shirts" and "mugs" creates browsing paths that encourage exploration.

Your storefront's visual coherence directly affects revenue. Because BASE functions as an independent shop, the overall consistency of the store carries more weight than any single product listing. When thumbnail backgrounds, description tone, and banner aesthetics all align, people feel more confident about the price. When cute illustrations sit next to sterile product photos, even good work looks less like a brand. Moving proven designs from SUZURI to BASE requires shifting your mindset from "this item sells" to "this shop earns trust."

Pay ID can provide some supplementary discovery, but your primary traffic engine is social media. On X and Instagram, posts that combine creation process + product URL + usage suggestion create stronger purchase intent than product showcases alone. A "usage suggestion" might be something as simple as "great for a home office desk," "easy gift pick," or "looks sharp on a laptop." Putting the product in a life context moves people from admiration to imagining ownership.

Reviews and user-generated content carry significant weight on BASE. Setting up a post-purchase feedback prompt or providing a dedicated hashtag for customers creates secondary exposure. New shops with small followings often lack social proof, so even a handful of real-use photos or brief testimonials can shift perception. In the early stage, asking friends or existing followers to try a product and share their experience is a legitimate way to build that initial credibility, as long as you stay within platform guidelines.

For repeat purchases, adding new products matters less than creating a lineup people can buy into. If a floral design is performing, expand it from mugs into stickers and T-shirts. Or offer color variations within the same series. This creates a "I want the other items in this line" dynamic instead of one-and-done transactions. BASE's shop-level structure is ideal for building a following around series, so designing for collectibility beats chasing individual hits.

My approach is to treat SUZURI as the testing lab and BASE as the brand-building stage. Following this sequence means SUZURI handles exposure and response data while BASE handles deep presentation of the winning products. When a listing isn't selling, separating the problem into design quality, visibility, and traffic source makes it much easier to diagnose. Products that sell aren't just well-designed; they're designed to be found by the right person, in the right place, at the right moment.

Copyright, Commercial Use, and Side Hustle Considerations for Employees

AI and Copyright Fundamentals

The first thing to understand when turning AI images into products is that "made with AI" doesn't automatically mean "legally safe." As Japan's Agency for Cultural Affairs outlines in its guidance, copyright assessment ultimately comes down to the resulting expression and how it's used, not the tool that created it. An AI-generated work that is based on and closely resembles an existing copyrighted work can raise the same issues as any other potential infringement.

The most important thing to avoid is designing products that closely resemble existing characters or celebrities. An output with a hairstyle, color scheme, and outfit that's unmistakably reminiscent of a specific anime protagonist is obviously risky. But even unnamed images that strongly resemble a famous actor or idol aren't safe. Copyright, personality rights, publicity rights, and unfair competition law can all come into play simultaneously. I take a strict stance on this. The more commercially promising a design looks, the harder I check whether its appeal is riding on someone else's recognition.

Trademarks are another easy oversight. You don't need to put a logo directly in the image. Including another company's brand name in a product title or description can create confusion. Phrases like "Disney-style," "Ghibli-esque," or "looks like official [brand] merch" are all flags in product copy, even if the design itself is original. If the text evokes a known IP, the risk jumps sharply.

I never publish a generated image without verification. My process: run an automated similarity check first, then manually search using reverse image lookup and related keywords. After that, review the terms of service for the AI tool used, the license for any models or assets, and the listing platform's rules. Finally, scan the product description and tags one more time for any language that could evoke copyrighted or trademarked material. It's tedious, but this layered verification is the single most effective step. Between the creative work and the compliance work, the compliance side matters more for a sustainable side hustle.

Commercial Use: A Practical Checklist

In AI goods selling, "the AI tool allows commercial use" is not where the analysis ends. In practice, four separate layers need to clear: the tool, the assets, the fonts, and the sales platform. OpenAI's terms of service permit use of outputs within their policies. Midjourney centers commercial use permission around paid plans. Stable Diffusion variants require checking licenses at the individual base model and LoRA level. This is critical: two images "made with Stable Diffusion" can have completely different licensing depending on which derivative model was used.

Here's a checklist that covers the common gaps:

• AI tool terms of service (check the source directly)

Review the official usage terms of whatever AI service you used. Commercial use permissions, output ownership, and prohibited uses vary by provider. Always verify before proceeding to product creation.

• Additional model and LoRA licenses

These may carry separate terms from the base model. Check the distributor's documentation.

• Asset licenses

If you incorporated textures, stock photos, or brush packs, verify whether redistribution and product use are permitted.

• Font commercial use terms

Check permissions specifically for logo use, merchandise, embedding, and rasterization.

• Product descriptions and tags

Scan for any language that could evoke existing works, brand names, or trademarks.

• Platform rules

Review SUZURI's and BASE's prohibited items policies, rights infringement procedures, and content removal criteria.

Sales platform terms deserve their own review right before listing. If there are rights issues with your generated content, consequences can go beyond product removal to account restrictions. Platforms aren't making special accommodations for AI; they're prohibiting listings that create rights infringement or consumer confusion. The practical standard is: if you can't confidently explain that a product is your own original work, it's going to be hard to keep listed.

In my experience, the riskiest listings break down not because of the images themselves but because of the surrounding copy. A visually abstract design that's perfectly fine on its own becomes problematic the moment you add "inspired by [game title]," "[brand]-aesthetic," or "perfect for fans of [franchise]" to the description. Treat the product page as part of your rights compliance, not just your marketing.

Tax Filing and Workplace Policies for Employees

If you're working a full-time job and selling AI-generated goods on the side, tax obligations and employer rules need to be on your radar from the start. In Japan, the general rule is that salaried employees must file a tax return if their total non-employment income exceeds 200,000 yen (~$1,325 USD) per year. Revenue from AI goods sales would typically be categorized as miscellaneous income or business income depending on continuity and scale. Since expense treatment varies by individual circumstances, specific tax decisions should be worked through with a tax professional or your local tax office.

One detail that trips people up: the threshold is based on "income" (revenue minus expenses), not gross sales. Subscription fees for AI tools, design software, and reference materials may qualify as deductible expenses depending on the context, but mixed personal and business use requires careful documentation. Even a casual start to a side hustle can become hard to account for retroactively if you haven't been keeping records.

Note: The tax rules described here are based on Japan's tax system. If you're in a different country, please check your local tax regulations for applicable thresholds and filing requirements.

Workplace policies carry equal weight. Some employers permit side work but require disclosure. Others have non-compete or information security provisions that add conditions. Using SUZURI or BASE under a pseudonym doesn't exempt you from your employer's rules. I want to be direct here: approaching this as a "side hustle they won't find out about" is not a strategy I'd recommend. A policy violation creates problems that have nothing to do with how much revenue you're generating. The three questions to sort out: Is side work permitted? Is disclosure required? Does the work conflict with your employer's business?

AI goods selling is easy to start because there's no inventory, which makes it tempting to assume "it's small enough not to matter." But copyright compliance, platform terms, tax obligations, and workplace rules are all areas where people who establish the foundation early tend to sustain the business longer. Building that operational base deserves as much attention as expanding your product line.

Common Mistakes and How to Avoid Them

Pricing Mistakes

The single most frequent error is setting tribun too aggressively from day one, resulting in product pages that get views but no purchases. The better a design looks, the easier it is to think "this should command a premium." But in AI goods selling, a product's appeal isn't just about visual quality; it's the combination of clear use-case communication and price credibility. On SUZURI, where pricing is transparent, even a small markup increase can widen the gap relative to comparable listings.

The fix isn't to chase maximum profit immediately but to start with modest tribun and observe. When views are healthy but purchases aren't following, the question to ask is whether the issue is margin per unit or margin multiplied by volume. A high-priced product that barely moves generates less revenue than a moderately priced one that sells consistently.

That said, dropping price alone doesn't solve most problems. What often preserves margin better is making the product description more specific about the use case. "A stylish mug" converts worse than "a mug sized for your desk coffee, holding about 310 to 320 ml (~10.5 to 10.8 oz), perfect for your morning routine at the office." SUZURI's mug really does fit that scenario, and framing it for gift use, workplace use, or home desk use lets you improve conversion without touching the price.

Quality oversights can also masquerade as pricing problems. If an image looks grainy, fine lines appear weak, or a motif gets awkwardly cut by seams or folds, buyers hesitate before the price even registers. Building images at 300 dpi, checking placement in the product preview, and being willing to regenerate and re-upload are small investments that prevent losses no pricing adjustment can recover.

Rights and Policy Violations

The more aggressively you pursue trending aesthetics, the higher the risk of inadvertently resembling existing characters, evoking trademarks, or incorporating assets that don't permit commercial use. This applies regardless of whether you're using AI. The most dangerous pattern isn't a single close resemblance but the accumulation of "inspired by" elements: hairstyle, color palette, silhouette, outfit details, and descriptive tags all pulling toward a recognizable source, even when no single element is a copy.

When I generate an image that feels familiar, I change direction early. This isn't a defensive reflex; it's genuinely the more productive path. Leaning on resemblance to something well-known constrains everything downstream: your product names, tags, and descriptions all have to tiptoe. Building original visual language gives you full creative freedom from product design through to marketing copy.

The practical framework is to treat Japan's Agency for Cultural Affairs copyright guidance and each tool's terms of service as separate checkpoints. Midjourney, Stable Diffusion variants, and OpenAI tools each handle output rights differently, and Stable Diffusion derivatives require checking all the way down to individual model and LoRA licenses. Fonts and texture assets with commercial restrictions can block a product even when the AI tool itself is cleared.

If a design feels close to an existing character, adjusting just the face isn't enough. You need to shift the color scheme, outfit, pose, and conceptual framing, essentially changing your starting point. Even without naming a franchise, a design that makes viewers think of a specific work is one to avoid. The products that last aren't the ones that flirt with the line; they're the ones that are easy to explain as original.

Traffic and Operational Stumbles

When listings aren't selling, the assumption is often "my designs aren't good enough," but in many cases the real issue is insufficient operational effort after publishing. SUZURI makes it easy to list and BASE gives you freedom, but both can create a false sense of completion at the moment of publishing. Without ongoing adjustment, you can't tell whether the problem is the design, the price, the description, or the discovery path.

A common early limitation is having only one product listed. A single design expanded across T-shirts, mugs, and stickers creates three different entry points for the same visual. T-shirts attract buyers for wearable reasons, mugs for daily utility, and stickers as low-commitment trial purchases. I've consistently found that spreading a design across product types gives much clearer signal on what's working than a single listing does.

Running social media without connecting it to the product page is another stall pattern. BASE especially relies on self-driven traffic, so product pages alone don't convey enough. Posting roughly three times a week with varying content, like a wear photo, a desk setup shot, or a color variant comparison, fills in the gaps that a static listing can't. The key isn't post frequency but consistency between what your social posts say and what the product page says. "Abstract art T-shirt" in a caption pointing to a page that says "minimal design for office-friendly everyday wear" creates a disconnect that costs clicks.

For ongoing improvement, revising product names and descriptions on a short cycle is highly effective. Twice a week, I'll swap out just the title, the opening line, or the usage phrasing and watch for response shifts. That kind of micro-adjustment preserves profit margins better than repeatedly tweaking prices, and it often reveals which framing resonates most.

On the quality side, resolution issues and print-readiness gaps sometimes hide behind what looks like a traffic problem. A thumbnail might look sharp, but if a potential buyer zooms in and the detail feels soft, they won't complete the purchase. Building at 300 dpi, checking seam and fold lines in the preview, and regenerating when necessary are maintenance tasks that reduce a category of friction that no marketing fix can address.

Your First Week Action Plan

Day 1

Day one is about locking in a single theme before you create anything. Casting a wide net across "cute animals," "abstract art," "retro typography," and "botanical motifs" leads to scattered image generation and inconsistent descriptions. Go beyond "who am I selling to" and articulate "who uses this, in what setting." Something like "a mug that makes a remote worker's desk feel a little brighter" or "a T-shirt someone with a monochrome wardrobe would reach for" gives your downstream decisions a clear anchor.

At the same time, study three reference accounts and competitors. Three is enough to avoid analysis paralysis. Focus not on follower counts but on first-image presentation, description phrasing, and which product types they've expanded into. I jot quick notes at this stage about what caught my attention and what felt off. Skipping this leaves you without criteria when you're evaluating your own image outputs on Day 2.

Day 2

Day two is for generating 10 image variations. The goal isn't just producing 10 different pictures. It's creating intentional variation across color and layout. For the same motif, try: with background and without, centered and off-center, warm palette and cool palette, with text and without. These deliberate contrasts make it much easier to spot which direction has commercial potential.

Narrow down to three favorites by end of day. Selection criteria should go beyond personal preference: does this work as a centered T-shirt graphic? Is it too busy for a mug wrap? Will the sticker outline hold up at small scale? What looks best on screen and what translates best to a physical product are slightly different things. Images with fine-grained detail often need stronger linework and more breathing room to succeed as merchandise.

Day 3

Day three: publish one design across three products on SUZURI. The recommended combo is a T-shirt, a mug, and a sticker. Each serves a different purchase motivation, making it easy to compare how the same design performs across contexts. T-shirts are about visual impact, mugs about daily utility, and stickers about low-barrier trial purchases.

Draft descriptions with ChatGPT, then personalize with specifics about usage, audience, and sizing. For the mug, help buyers picture it on their desk with their morning coffee. SUZURI's mugs hold about 310 to 320 ml (~10.5 to 10.8 oz), just right for a generous cup and easy to frame as a daily-use item. For the T-shirt, the 100% cotton 5.6 oz fabric is a genuine selling point for everyday wearability. Work these details into the description for credibility.

Day 4

Day four: post three times on social media, combining creation process content with product URLs. Vary the timing to see which audience segments engage. Posts that show the generation process, rejected alternatives, or color experimentation tend to hold attention better than a standalone product shot.

Keep your social copy aligned with the product page. If the listing says "desk-friendly," your post shouldn't pitch it as "bold statement art." I've found that keeping the social framing close to the product page's opening line consistently reduces post-click bounce. Overpromising on social and underdelivering on the product page is a conversion killer.

Day 5

Day five: review your views, likes, and click data, then adjust the product name and primary image. This order matters. Jumping to price changes first obscures what's really happening. The first diagnostic: are people seeing the listing but not clicking, or clicking but not buying?

Make small, targeted changes. Swap an abstract product name for one that includes the use case. Replace a full-product hero shot with a tighter crop that shows the motif clearly. I often spend just five minutes on a weekday refining a single description element. These incremental improvements feel minor individually, but they compound into meaningful conversion gains. Adjusting title, opening line, and lead image one at a time also makes it possible to identify exactly which change drove the improvement.

Day 6

Day six: launch a second design across three products using the same structure. The important thing is to not copy Day 3 exactly. Change your hashtags or try a different positioning angle to create a second test condition. This way, you're not just evaluating design quality but also testing which messaging approach lands.

For example, if Design 1 was framed around "cute" and "relaxing," try framing Design 2 around "desk upgrade" or "gift pick." Stickers in particular tend to move when you show placement context, like a laptop or tumbler with the sticker applied. Waterproof stickers invite daily-use imagination in a way mugs and T-shirts don't. If response patterns differ between the two designs, the difference might be less about the artwork and more about how you've positioned it.

Day 7

Day seven: line up the full week's data and concentrate your energy on the one or two products that showed the strongest response. Trying to push everything equally scatters your effort. Sort products by which ones attracted views, which ones got clicks, and which ones generated engagement like saves or conversations, even if purchases haven't materialized yet.

From here, the natural next phase is to observe for another three weeks or so, then consider moving your winners to BASE for brand development. Setting up a logo, key color, and a polished top page transforms the experience from "individual products" to "a shop worth returning to." BASE has over 1.7 million shops and 16 million cumulative Pay ID registrations, but the real results still come from the traffic funnels you build yourself. My own rhythm is adding one new design on weekends and spending weekdays refining descriptions, steadily finding what resonates before committing to brand-building.

What You Can Do for Free

The free starting scope is well-defined: list on SUZURI and run social media promotion. Those two activities alone let you test design appeal, product-type fit, and content-to-commerce flow. SUZURI's integrated creation-to-sales pipeline and hands-off production and shipping make it ideal for initial validation.

Base your decision to invest money on KPIs, not gut feeling. Spending on BASE design upgrades or premium features makes sense once you have at least one sale per week for two consecutive weeks, which signals sustained demand. Before reaching that threshold, your time is better spent on product count, description quality, lead image selection, and social-to-listing consistency. The principle isn't "build it before they come" but "only invest in building around the things people are already responding to." That mindset fits this side hustle perfectly.