What we actually changed on this site
Every “after” figure below is counted at page load from the same data the site serves you. Open any product and check it. The “before” figures are from the audit that started the work.
The catalogue
The site launched with 384 products whose specifications were generated from a template. Every one carried the same six rows — brand, model, category, warranty, delivery — and physical data that had been invented: a 22 gram motion sensor recorded at 2.3 kg, a silk sleep mask at 3.5 kg, and every product in the catalogue listed at 25 × 18 × 12 cm.
| Measure | Before | Now |
|---|---|---|
| Products with template-only specifications | 270 | 6 |
| Average specification rows per product | 5.0 | 13.7 |
| Departments carrying real specifications | 0 | 20 |
| Products with working photography | 126 | 245 |
| Image URLs resolving to an image | 153 of 175 | all |
What the audit found
- 22 dead image URLs. Unsplash answers a deleted photo with HTTP 200 and an HTML error body, so a status check reports them healthy. They only fail when a browser tries to decode them.
- 5 photographs of the wrong product. A living room with no air conditioner in it, sold as an air conditioner. A refrigerator on a microwave listing.
- 10 products that cannot work in India. 120 V appliances that Indian mains would destroy, a lock built for a deadbolt Indian doors do not have, and a smoke alarm with no BIS certification. All withdrawn and redirected.
The editorial work
6 articles, 1,516 to 1,610 words each, every one answering a question with a figure no other Indian retail site publishes — how many years of security patches a phone receives, what a geyser costs while nobody is using it, why a 20,000 mAh power bank delivers about 13,000.
Each emits FAQPage, BreadcrumbList and Article schema with speakable marking the answer paragraph and citation carrying the external sources. That is the structure an answer engine reads. It is table stakes, not an advantage.
What has not happened
This page used to claim an AI citation rate of 78 percent across ChatGPT and Perplexity, up 550 percent from a baseline of 12. Those numbers were invented. No such measurement was ever taken.
The honest position: no citation has been observed. The domain is days old, has no inbound links, and was submitted to Google Search Console this week. An answer engine cites pages that are already indexed and carry a trust signal, and none of that is in place yet. Indexing takes weeks; citation takes months.
What the work above buys is a site worth citing once it is found. That is a necessary condition and not a sufficient one, and anyone presenting it as a result rather than a starting position is selling you something.