True story: an 11-agent inmobiliaria buried in 14 spreadsheets and 3 disconnected tools. We built them a single internal SaaS in 9 weeks. Here's exactly what we built, and the math on why it paid back in 4 months.
Tell us your idea and we'll send you a fixed quote in less than 24 hours. No commitment.
When Inmobiliaria Levante (real client, name changed) hired us last summer, their daily operations looked like a small ecosystem maintained by hand:
Every Monday morning the office manager spent 2.5 hours copying data from the Excels into a "weekly KPI report" PDF the boss read while he drank his coffee.
Eleven months later we had quietly replaced almost all of it with a single, custom-built internal app. 2,400 hours/year saved. Listings went up the same day they were captured instead of waiting an average of 4 days. The boss stopped asking for the Monday PDF — because the dashboard was live on the wall TV.
This is the story of that build, what we did right, what we got wrong, and why custom software still beats off-the-shelf for a specific kind of company.
The first thing we did wasn't to write code. It was to rule out existing solutions. We asked the team:
This is the honest test for custom software: if 3 mainstream SaaS already sort of fit, custom is the wrong answer. Buy SaaS, accept 80% fit, move on.
But for Inmobiliaria Levante, three things tipped it:
Custom won.
We worked in 9 sprints of 1 week each, every Friday demoing to the whole team in their office over coffee.
We started with the unsexy infrastructure. Auth with email + Google for agents. Role-based access (admin, agent, mortgage partner, reform partner, read-only owner). Audit log on every write. PostgreSQL with row-level security for partner separation.
This week looked like nothing visible. Boss was nervous. We held our ground.
The single most-used module. We built:
Each buyer has a search profile. The engine matches every new listing against all active profiles in milliseconds. Three minutes after a property is captured, an automatic WhatsApp message goes to the right buyers with a personal note from their assigned agent.
The WhatsApp Business API integration (via the official Cloud API) was the single biggest behavioral change in the company. Open rates: 94%. Reply rates: 38%. Compare to email blasts they used to do (12% / 1.5%).
Replaced Trello. Each property has a vertical timeline of offers, counter-offers, files, signatures, with Slack-style mentions to keep all internal comms in context. We pulled in Docusign via API for digital signatures.
The dreaded one. Holded's API is functional but not the smoothest in Spain. We built a bidirectional sync: deals closed in our app create draft invoices in Holded with the right VAT, the right commission split, and the right tag for the boss's quarterly tax filing. The accountant cried tears of joy.
This is where custom destroyed the off-the-shelf options. We modeled their exact rule: agent gets 40% of the commission normally, 50% if the buyer came from their personal network, capped at €X/month, with a quarterly bonus if a team-level revenue target is hit. The system calculates this automatically and produces a payout report on the 1st of each month, ready for SEPA.
Live KPIs streamed to a dashboard pinned to a TV in the office. Visits this week, listings captured, offers in progress, pipeline value, agent leaderboard. Updates in real time. Within two weeks the team's behavior changed without a single management memo — competition, gamified.
A scoped-down view of certain listings exposed to mortgage brokers and reform companies. They can leave quotes, request photos, mark interest. Replaces an awful WhatsApp group of 19 people.
Algolia-powered search, a one-click GDPR data-export endpoint, and a clean admin panel for the boss. We also did a small piece for compliance: automatic deletion of buyer data 24 months after their last activity, a checkbox he didn't know was missing until we mentioned it.
We won't pretend everything was smooth. Two real mistakes:
| Before | After | |
|---|---|---|
| Time from capture → publish | 4 days avg | same day |
| Agent admin hours / week | ~9 / agent | ~2 / agent |
| WhatsApp open rate | 12% (email) | 94% |
| Listings published / month | 38 | 71 |
| Offers managed in tool | 0 (Trello) | 100% |
| Monthly KPI report | 2.5h manual | live, real-time |
| Tooling cost / month | €647 (mixed) | €89 (hosting) |
| Hours saved / year (team) | — | ~2,400 |
| Average sale cycle | 71 days | 52 days |
At an average loaded cost of €18/hour for an agent's admin time, the saved hours alone are €43,200/year. Build cost was paid back in 4.1 months.
Build custom when:
Buy off-the-shelf when:
We will tell clients to buy SaaS about a third of the time. No, that's not a typo. We turn down those projects gladly.
We do a free 2-hour discovery workshop for SMBs who think they've outgrown spreadsheets. We'll map your real processes, run the buy-vs-build math with you, and tell you the honest answer — even if it's "stay on Notion for another year".
Book a discovery call or explore our custom software services.
Spreadsheets don't scale. Process does.