Batang
From material-choice uncertainty to a working service.
Choosing materials for my home was less about finding attractive products than deciding what suited my space and budget. I defined the audience, planned comparison flows, used AI to implement the service, created introductory content and designed behavioral measurement. Batang is a personal project testing whether content-driven visitors can move from browsing to comparing and selecting materials.
- Period
- Sep 2026 — Present
- Role
- Personal project · Product planning · UX/UI · Content · Measurement design
01 · BACKGROUND
A clearer starting point before speaking to an installer
While preparing my home, I wanted to keep the existing flooring and find plaster-textured wallpaper within a practical budget. Similar-looking materials used different names, specifications and pricing units across brands. It was difficult to judge both the overall look and total cost. I started with a question: could consumers make better decisions if they could shortlist materials and establish a budget before a consultation? This began with my own experience, and I am testing the need through the service rather than treating it as proven market demand.
02 · PRODUCT STRATEGY
Start with a shortlist of wallpaper options
The initial audience is newlyweds and people preparing to move who have a desired look and budget but limited knowledge of materials. I started with wallpaper, where browsing and comparison criteria are relatively clear. The flow is: enter home details → explore materials → compare up to three options → select a preferred material → review estimated costs → share the selection. The questionnaire is optional. Products without confirmed prices are marked as requiring an inquiry, and material prices are separated from estimated costs including installation.
03 · MY CONTRIBUTION
From problem definition to a working service
I defined the problem, feature priorities, information architecture, filters, comparison experience, service copy and content direction. I used AI tools to implement, test and revise the service, translating requirements into clear behavior and exception-handling rules. I expanded browsing from wallpaper to flooring, films and tiles, separated comparison from the basket, and added material search and surface filters to the 3D preview. My focus is on how people interpret information and the conditions that produce misleading results.
04 · ACQUISITION
Begin with the customer’s problem
Using my own home photography and renovation experience, I created a Reel about the difficulty of choosing materials and why I built Batang. I published the introduction on October 10, 2026. Future content will test two different concerns: price differences between similar wallpapers and finding the right plaster texture. I distinguish organic content from paid advertising. The measurement plan uses content-specific UTMs to connect acquisition with comparison and selection behavior. Paid-media efficiency results will be added only after the underlying campaign data is verified.
05 · MEASUREMENT
Decide what to measure beyond visits
I use GA4 for acquisition, sessions and page views, and Amplitude for material exploration, comparison and selection. The event plan separates category browsing, filters, product details, comparison candidates, the comparison screen, preferred-material selection and share-link creation. Search Console and Naver Search Advisor help check search acquisition and indexing. The current conversion is material selection, not a purchase. For paid campaigns, I plan to review comparison and selection behavior and cost per selecting user alongside CTR and CPC. Creating a share URL is measured separately from someone opening it. My website growth report brings together the portfolio, photography site and Batang, recording each site’s objective, source, period and filters. I do not sum daily unique users or combine GA4 and Amplitude users. Missing data is not treated as zero, and recent figures are rechecked for reporting delays.
06 · EARLY OBSERVATION
Use early numbers to identify the next questions
In the October 13, 2026 report, Batang recorded 11 GA4 users, 33 sessions and 337 page views for October 6–12. Only known qa/test acquisition was excluded; these figures are not a verified count of real customers with all internal activity removed. For October 4–12, Amplitude showed an ordered funnel of category exploration 12 → comparison screen 4 → material selection 2, using the same identifier within 24 hours. This is an early sample that includes testing. Its period and identity rules differ from GA4, so I did not combine the figures to calculate a conversion rate. The observations informed the next questions: can people find comparison, understand preferred-material selection versus the basket, and access sharing at the right moment? These numbers establish a measurement baseline, not evidence of growth.
07 · ITERATION
Turn implementation problems into clearer rules
The 3D preview originally required leaving the screen to change flooring. I brought search and surface filters into the same view so materials can be changed and compared in context. Repeating product thumbnails as textures could turn furniture in staged photographs into floor patterns. I introduced a rule to use reviewed surface images as textures and label other previews as approximate colors or estimated textures. A product-family name containing “green” was incorrectly interpreted as a color. I changed the priority to image-derived color → explicit color information → color words excluding product-family names → manual review, and aligned search filters and the 3D preview. These are confirmed usability and representation fixes, not verified conversion improvements.
08 · NEXT EXPERIMENT
Help people move from a click to a selection
The first planned test addresses the step from adding comparison candidates to opening the comparison screen. The hypothesis is that a label such as “Compare 3 materials,” combining the count and action, will make the next step clearer. With limited traffic, I will prioritize before-and-after observation and task-based usability checks over a concurrent A/B test. I plan to track the share of unique users who open comparison within 24 hours of adding a candidate, together with candidate removals and selection cancellations, reporting both numerator and denominator under comparable acquisition conditions and time windows. Internal and test activity must first be separated and event duplication and gaps checked. The proposed wording and experiment are not yet implemented or validated. Content-message tests will follow this measurement work.
09 · ROADMAP
Build a more trustworthy decision experience
Near-term work focuses on reviewing surface images, clarifying pricing units and estimate assumptions, and improving mobile browsing and 3D loading. Next, I will examine how content-driven visits lead to comparison, selection and sharing, and prioritize confirmed problems. Longer term, I would like to extend the service to furniture, appliances and lighting. The current decision-support hypothesis and data quality come first, alongside clear standards for image rights and product information. Batang connects customer concerns with content, product flows and measurement. Service: https://penggwon.com/batang/