Gruhome: what two years of owning everything teaches you.
A premium home-furnishings retailer in Delhi NCR, thousands of fabric SKUs, a non-technical sales team, and one person responsible for everything digital. From Jul 2022 to May 2024, that person was me.
The situation
Gruhome sells fabrics, curtains, and soft furnishings: a business where a single product can have dozens of material and colour variants, and where the sales conversation happens on the shop floor, on a phone, with a customer waiting. There was no digital infrastructure and no budget for an agency.
Being the sole owner meant every decision came with its full cost attached: if I chose a custom build, I maintained it; if I chose a platform, I lived inside its constraints. That accountability is the most PM-shaping thing that has happened to me.
The decision log
-
Shopify over a custom storefront
The call: build on Shopify's basic plan and write every Liquid section from scratch, benchmarking against Sixpenny, instead of a custom stack I'd be maintaining alone forever.
Why: a custom build optimises for my résumé; a platform optimises for the business surviving my eventual departure. The basic-plan constraint was real (no metafield UI, variant caps), so the work became designing around it.
Custom tabbed product pages and a metafield-driven fabric selector that non-technical staff edit themselves.
-
A zero-cost PWA when every tool hit its limit
The call: after Excel, SharePoint, and third-party tools all failed the shop floor, build a PWA catalog with Google Sheets as the live database, GitHub for versioning, and free static hosting.
Why: the sales team needed instant SKU lookup on phones with zero training and zero recurring cost. Every off-the-shelf option failed at least one of those three.
2,000+ products searchable on the floor, no licence fees, in daily use since launch. It eliminated inventory-search interruptions entirely.
-
Warn, don't block, on duplicate dispatches
The call: when the dispatch tracker detects a fabric book already out with another client, it warns the salesperson instead of blocking the dispatch.
Why: the floor sometimes genuinely needs to send a duplicate (two clients, one popular book). A hard block would push staff back to WhatsApp and untracked dispatches: the exact failure that cost the business money across ₹13L+ in catalogue assets.
The system stays the source of truth because it never gets in the way of a sale.
-
Kill the admin feature that blurred roles
The call: after one walkthrough with the approvers, I cut an admin-side feature I'd already built; it let admins edit dispatches in a way that made approvals ambiguous.
Why: the tracker's value is a clean answer to "who has the book and who signed it out." A feature that blurs the approval boundary erodes the one guarantee the system makes.
Sunk cost eaten, role boundaries intact. Shipping less kept the product trustworthy.
The demand side
Tools were half the job; the other half was demand. Across 2022–2024 I ran all performance marketing through Meta Ads Manager: audience targeting, creative iteration, and budget calls on a spend the owner could feel personally.
Small budgets are a forcing function: at this scale you can't buy your way out of a weak creative, so every campaign became an experiment with a hypothesis and a kill threshold.
- Meta campaigns run 12
- Impressions 1.2M+
- Link clicks 49,500+
- Average CPC on ₹35K+ spend ₹0.94
What it taught me
Everything I now do deliberately as a PM, I first did at Gruhome out of necessity: scoping against real constraints, choosing boring technology on purpose, watching a non-technical user struggle with something I thought was obvious, and killing work I was proud of because the walkthrough said so.
The systems are still in production, which is the only metric of the four I'm actually sentimental about.