Skip to main content

MaoMuse case study: 1,281 search clicks a month and almost no sales

Published · Updated

TL;DR: MaoMuse is an AI pet-portrait shop that RedClaw owns and runs in Taiwan. It went from signed-off plan to live site in 3 days (live 12 July 2026), with 62 articles and 12 free tools online the next day. Monthly Google clicks grew from 79 to 1,281 in three months, but confirmed paid orders stayed at 2, worth NT$663 (about USD 21). We spent USD 369.04 on Meta ads and stopped in week two. The traffic landed on pet-name generators and pet end-of-life guides, which are a long way from buying a portrait. Since 21 September 2026 the shop sells digital files only and earns through affiliate referrals.

Some Taiwan context first

  • ECPay (綠界) and NewebPay (藍新) are local payment gateways that small Taiwanese web shops apply to for card, bank-transfer and convenience-store payments. Both review your website before you can collect money. We applied to both; between them they sent back 8 items to fix.
  • LINE is Taiwan's default messaging app. Our order center, notifications and customer-service bot run inside a LINE Official Account.
  • Prices are in New Taiwan dollars. We convert at 31 TWD per USD, the fixed rate used in RedClaw's reporting.

Store facts

ItemItem
ProductCustom pet portraits: upload a photo, get an AI-generated portrait. Originally digital files plus printed canvases (white-label print-and-ship partner). Digital files only since 21 Sep 2026, plus affiliate referrals
Live since12 Jul 2026 (plan signed off 9 Jul)
StackNext.js 15, TypeScript, Tailwind, Vercel, Firestore, Cloud Storage. Images generated with OpenAI gpt-image-2, with three rate limits (150 generations a day site-wide, 6 per IP, 3 per email)
PaymentsDesigned for ECPay and NewebPay in parallel. ECPay sent back 5 items, NewebPay 3. NewebPay confirmed live on 9 Aug
FulfilmentPrinted canvases shipped by the print partner, dispatched automatically after payment. Digital delivery only since 21 Sep
Customer channelLINE Official Account with order center, notifications, menu and an AI support bot

What happened after a 3-day launch?

Date (2026)What happened
9 JulPlan signed off, 10 decisions made, brand named
10 JulSite skeleton, checkout and ECPay module built
12 Julmaomuse.com live; AI generation tested on production
13 Jul62 articles and 12 tools published (about 400,000 Chinese characters). LINE account, GTM, GA4, Meta Pixel and Conversions API connected
14 JulReal-card test failed: the ECPay merchant account hadn't been activated yet. The code was fine
24 JulECPay sent back 5 items and NewebPay 3; we started on the paperwork
27 JulMeta ads started. That evening, the first confirmed real order: a digital file, NT$349
29 JulSecond real order, digital file, NT$314
8 AugLast day of Meta spend
9 AugDecision: stop paid ads, keep SEO only. NewebPay confirmed for live payments
17–20 SepStarted a 32-article cat series; 9 breed pages live
20 SepFirst affiliate link live (a cat-litter group buy)
21 SepPrinted canvases discontinued; digital files only, revenue moves to affiliate referrals
27 SepStrengthened the pet end-of-life content cluster

Monthly clicks went from 79 to 1,281. What about orders?

Data pulled on 7 October 2026. Orders from our own order database, ad spend from RedClaw's daily Meta sync, search data from Google Search Console.

MetricValue
Confirmed real paid orders2 (NT$349 on 27 Jul, NT$314 on 29 Jul), NT$663 total, both from Meta
New paid orders after 29 Jul0
Meta ad spend (27 Jul – 8 Aug)USD 369.04
Meta clicks1,836
AI portraits generated111
Users linked via LINE19
Affiliate link clicks15
Google organic clicks (12–31 Jul / Aug / Sep)79 / 296 / 1,281
Google organic impressions (same periods)2,718 / 9,327 / 55,439
Average position (same periods)16.0 / 9.9 / 8.6
Top pages by clicks, last 4 weeksPet-name generator 329, euthanasia cost guide 124, pet anniversary guide 115
Clicks to the purchase page /create, same 4 weeks28
SEO audit score54 → 68 → about 82

Our estimates (31 TWD per USD):

EstimateFormulaResult
Meta spend in TWD369.04 × 31≈ NT$11,440
Ad cost per real order11,440 ÷ 2≈ NT$5,720 (≈ USD 185)
ROAS663 ÷ 11,440≈ 0.06

The launch plan said blended ROAS should reach 1.3 within 4 months on a total budget of NT$160,000 (plan figures). We stopped in week two instead.

Where did the time and money go?

1. The traffic grew in the wrong place

  • What happened: September brought 1,281 organic clicks, 4.3 times August. Most went to a pet-name generator and to guides on euthanasia costs and pet anniversaries. The page where you actually order a portrait got 28 clicks in four weeks.
  • Cost: around 400,000 characters of content built a nice traffic curve, and no new paid orders after 29 July.
  • Fix: from 21 September we sell digital files only and put affiliate offers on the informational pages that do get traffic, for example a cat-litter group buy on the cat-cost page. Partner slots on the end-of-life pages are built; we're still looking for the right partner.

2. Payment reviews check that the site matches the merchant records

  • What happened: ECPay returned 5 items: product category didn't match the registration, no pricing information on the site, contact email didn't match, and they wanted photos of the business premises and of stock. NewebPay asked for 3: a published refund policy, matching customer-service details, and contract and supporting documents.
  • Cost: on 14 July ECPay still wasn't active, and NewebPay only confirmed live payments on 9 August.
  • Fix: worked through each item. A custom product has no stock to photograph, so we sent finished-product photos plus the print partner's shipping records instead.

3. We set prices before getting the supplier's quote

  • What happened: at the original prices, the 40×60 cm canvas and the framed version would have lost money.
  • Cost: a site-wide price change once the print partner's quote came back (14 July).
  • Fix: repriced from the real quote. Now the order is cost first, then price.

4. The site promised a generation time we couldn't hit

  • What happened: the site said "ready in 30 seconds". Real runs took 150–265 seconds.
  • Cost: customers waited more than five times longer than the page said.
  • Fix: changed every mention to the measured time.

5. Bugs that only show up in production

  • What happened: the serverless runtime had no Chinese-capable fonts, so the watermark text on previews rendered as empty boxes online. Locally it looked fine.
  • Cost: until the fix, every real customer saw a broken watermark on their preview.
  • Fix: the watermark is now pre-rendered as an image, so production doesn't need fonts at all. On a related note, new pages weren't picked up by Google on their own. The 9 pages we submitted by hand were all recrawled within 13 minutes, and one moved from position 94–98 to 30.6. The control group we didn't submit didn't move.

How is the store doing now?

  • Paid ads stopped on 8 August and haven't restarted. USD 369 bought 2 digital-file orders, an estimated ROAS of about 0.06.
  • Organic search is still growing. In late September: 36–58 clicks and 1,300–4,000 impressions a day, average position 7–12.
  • The business model has changed. Printed canvases are gone; the site sells digital files, and the main revenue bet is affiliate referrals. 15 affiliate clicks so far, no conversion or commission figures yet.
  • One known bug: the order page for paid digital orders errors out because the high-resolution file is no longer in storage.

What we're working on next (not finished; numbers will be updated):

  1. Put affiliate offers that fit the topic on the high-traffic name-generator and end-of-life pages, so informational visitors have somewhere useful to go.
  2. Publish the rest of the cat series on schedule and submit each page to Google by hand.
  3. Fix the order-page bug.

What this means for your store

  1. Before writing anything, ask whether someone searching that phrase would buy from you next. We wrote 62 articles and 12 tools and passed 1,000 clicks a month, but very few of those visitors wanted a portrait. When we plan content for a client, we first map each keyword's distance to the purchase page. See our e-commerce coaching service.
  2. Before a payment review, check your pricing, refund policy and contact email against your merchant records. Our two gateways sent back 8 items between them, mostly about what the site disclosed. ECPay publishes its review criteria, the documents you need and its fees (Chinese only). Our other store, Taolilife, ran into the same kind of problem.
  3. Agree on a stop-loss for ads before you start. We stopped in week two because spend and orders were too far apart; there was no need to wait four months to know. All three stores are covered in our e-commerce hub.

FAQ

With 1,281 clicks a month, why only 2 orders?

The clicks went to a pet-name generator and guides on euthanasia costs and anniversaries. The order page got 28 clicks in the same four weeks. People searching those topics mostly weren't shopping for a portrait.

Why stop ads in week two?

USD 369.04 brought in 2 digital-file orders worth NT$663, an estimated ROAS of about 0.06. The gap was too wide to justify waiting.

How does the site make money now?

Digital files, plus affiliate offers on the informational pages that get traffic. As of 7 October 2026 that's 15 affiliate clicks, with no conversion or commission figures yet.

Is your traffic growing while orders aren't? Send us your URL on Telegram: https://t.me/eric16888999