Gadget EC owned media｜1.5 million monthly PV
Overview
| Item | Detail |
|---|---|
| Industry | Gadget EC |
| Price | About ¥10,000–¥50,000 |
| Result | ① Monthly PV from 700,000 to 1.5 million ② Sales about 2x |
| Sales | Owned EC only |
| Scope | From owned-media strategy through execution |
We ran a WordPress owned media after designing who we speak to, and what we deliver, first. Monthly PV from 700,000 to 1.5 million. Sales about 2x as well.
Marketing strategy
| Strategy | Detail |
|---|---|
| Who / What | Design who we speak to, and what we deliver, first |
| How | Run owned media on WordPress |
| Focus | Because the category is niche, how the team is built |
Strategy and execution, both. Who and What first, then How, then WordPress owned media.
Volume needs a team. People who already know a niche gadget are rare. People who can also write and direct are rarer.
After trying other setups, we landed on this: even without much writing or directing experience, put people who love that gadget on the team. Gather people who know it. That is what stabilized the team most.
Marketing tactics
| Tactic | What we did |
|---|---|
| Content SEO | Understand the gadgets we sell, then run to a manual |
| Technical SEO | A specialist consultant in, plus engineers, for structured data and other basics |
| Building and running AI agents | Contractors to one-third or less. Delivery that took a month per piece down to about 1.5 hours |
Content SEO
Train the team on the gadgets we sell. Run the operation to a manual. The base is people who can write, and who like those gadgets.
Technical SEO
A specialist consultant in, working with engineers, for structured data and other basics.
Building and running AI agents
Before AI, contractors including writers and proofreaders sat at 30–40 people. After AI agents, the operation ran with 6–7. Contractors to one-third or less. Quality held. Delivery that took about a month (3–4 weeks) per piece moved to about 1.5 hours.
This was early in the AI wave. We started with prompt engineering for simple speed. Then context engineering with AI agents. Finally harness engineering: we built the agents and ran them.