DataToLeads / AvocaData
The multi-tenant Data-as-a-Service platform we built from scratch — turning millions of spreadsheet-bound records into a white-label business that runs on 427 million.
What this project was for
Millions of lead records stuck in Excel — two or three agencies tried and failed before us. We built a four-interface, multi-tenant DaaS platform with a master→child credit economy and sub-second search across 425M+ records.
The problem
The client was sitting on an asset he couldn't operate. Years of scraping and automation had left him with hundreds of millions of lead records — and a spreadsheet to run them from. He was operating a real data business by hand in Excel: pulling lists, tracking buyers, fulfilling orders, all manual, at a scale where a spreadsheet stops being a tool and becomes the bottleneck. An entire community of vendors and agencies was stuck the same way.
What we built
Everything from scratch — raw idea, feature definition, architecture, infrastructure, and the scaling that holds it together. It started as a single end-user app, then grew into a full multi-tenant platform across four integrated interfaces working off one credit economy:
Agency white-label control panel
any agency resells the platform under its own brand and domain, with an 8-step onboarding wizard, page builder, agency-defined plans, dual payment gateways, and a master→child credit economy across thousands of downstream customers.
End-user application
two complete dashboard modes over one dataset (power-user filter table + guided card mode), natural-language audience search, masked previews until purchase, saved/enriched lists.
One-click CRM sync (GoHighLevel)
each agency connects its own isolated GHL account; end users sync any purchased list in a single click, no exports.
A vendor-uploaded data marketplace
vendors upload raw CSVs; the platform infers schema, converts types, validates, and turns each into a fast queryable database behind an approval gate. Of 427M records, 316M were uploaded this way.
Public marketplace & storefront
100+ vendors list products with per-row pricing, accuracy scores, freshness metrics, and credit checkout.
Built to run in production
Every agency on its own domain, zero manual setup
per-agency white-labeling on Kubernetes Services + Ingress Classes, provisioned automatically.
Dedicated compute per agency
dedicated K8s nodes per agency via a Python package driving kubectl, pods allocated by plan so no noisy neighbors.
Sub-second search across 400M+ records
Quickwit inverted-index search, sharded across nodes, fed by a daily Celery ETL pipeline.
Any CSV, schema figures itself out
a custom ClickHouse ORM infers schema with intelligent type conversion and high-performance inserts.
Results
Records live today, searchable in under a second.
Uploaded by data vendors self-serve, through the marketplace.
White-label agencies and 8,000+ active users, each on their own domain.
100+ vendors monetizing their data — still built and run by BrainsLogic.
How it was built
The platform was built as a multi-tenant system from the first line of code, so a single deployment could serve independent agencies — each on its own domain and brand — without per-client forks. The tenancy model runs through the data layer, the master→child credit economy, and the infrastructure itself.
Search across hundreds of millions of records is the product, so it drove the architecture. A Quickwit inverted index, sharded across nodes and fed by a daily Celery ETL pipeline, keeps queries under a second at 400M+ records. A custom ClickHouse ORM ingests arbitrary vendor CSVs, infers the schema with intelligent type conversion, and runs high-performance inserts behind an approval gate.
Per-agency isolation is handled at the orchestration layer: Kubernetes Services and Ingress Classes provision each agency's domain automatically, and dedicated nodes per agency — driven from a Python package over kubectl and allocated by plan — keep tenants from competing for compute.
Automated per-tenant provisioning
Kubernetes Services + Ingress Classes give every agency its own domain and brand with zero manual setup.
Sub-second search at scale
A Quickwit inverted index sharded across nodes, fed by a daily Celery ETL pipeline over 400M+ records.
Schema-inferring ingestion
A custom ClickHouse ORM converts arbitrary vendor CSVs into fast, queryable tables with type conversion.
Master→child credit economy
One credit model spans agencies and thousands of downstream customers, with dual payment gateways.
What it runs on
Worse, we had cycled through standard development agencies that promised the world but simply didn't deliver. Then we partnered with BrainsLogic. Together over the years, we didn't just fix a few bugs; we completely re-engineered how big data is distributed and commercialized. DataToLeads is the work I am proudest of, and it wouldn't be the powerhouse it is today without the brilliant minds at BrainsLogic. If you want to build something that actually shifts an industry, this is the team you hire.
Need a production system like this?
Tell us what you're building, fixing, automating, or scaling. Book a technical call to review scope, architecture, and timeline — or get an estimate and we'll come back with honest feedback on feasibility and approach.