How Rodz Produces Its Own Intent Signals

The Collection Infrastructure

Scrapers running continuously

An intent signal doesn’t appear out of thin air. Behind every detected event sits a large collection infrastructure. Rodz runs its own scrapers, which continuously query public data sources at scale. Each scraper targets a specific source type, whether that’s a business registry, a job board, a professional social network, or a sector-specific database, and it’s tuned to extract relevant information without hammering the source server.

These scrapers run around the clock. Query frequency varies by source: some registries update daily, others in real time. The point is to catch every event within hours of publication. Rodz rebuilds its scrapers regularly, because sources change their structure constantly and a scraper that isn’t maintained drifts back to producing noise.

Cross-referenced data sources

The sources queried cover the full digital footprint of a business.

Legal and administrative sources

SIRENE: registrations, address changes, statutory modifications
Commercial registries: financial filings, capital increases
BODACC: legal notices, transfers, liquidations
Legal announcement journals: official publications

Employment and HR sources

General job boards (Indeed, HelloWork) and specialist ones (Welcome to the Jungle, Talent.io)
LinkedIn: job postings, executive role changes
Company career pages

Financial sources

Press releases: funding rounds, partnerships, financial results
Investor databases: funding rounds, valuations
Industry publications

Media and social sources

LinkedIn: posts, comments, engagement signals
Industry press: articles mentioning target companies
Google Maps: new locations, address changes

Cross-referencing sources

The value Rodz produces doesn’t come from any single source in isolation. Anyone can check SIRENE or Indeed, or spin up a scraper on Apify. What changes when sources are cross-referenced: a hiring surge on Indeed, matched against a funding round picked up in a press release, matched again against a leadership change on LinkedIn, forms a composite signal that’s considerably stronger than any one of those events read alone. That stacking is where the operational value actually comes from.

The Producer-vs-Reseller Dynamic in Practice

It’s worth being concrete about how this works in the market, because it has direct implications for anyone buying signal data.

Several B2B platforms resell Rodz signals under their own brand, surfacing them inside their own interface as if the detection were native to their product. Users of those platforms are, in effect, consuming Rodz-produced data, typically at a significant markup. The signals appear with the reseller’s logo, but the production infrastructure behind them is Rodz’s: the scrapers, the deduplication, the enrichment cascade, the 48-hour delivery window. None of that is rebuilt by the reseller. They buy access to the output and sell it on.

This is the practical difference between a producer and a reseller. The reseller adds distribution and a user interface. Rodz adds the data itself. When something goes wrong upstream, whether a source changes structure, a scraper breaks, or a signal type requires recalibration, the reseller has no lever to pull. Rodz rebuilds the scraper.

The job-offer signal makes this visible in an unexpected way. Rodz’s aggregation across 6,500+ job boards and career pages has made it, by volume of job listings processed, one of the largest job posters in France. That’s not a positioning claim; it’s an artifact of production scale. No reseller achieves that position by redistributing a feed someone else built.

For buyers, the implication is simple: if you’re already using a platform that surfaces intent signals, there’s a reasonable chance you’re already using Rodz data, just at the price the reseller chose to charge for it. Coming to the source means access to the full signal catalog, at production cost, with configuration control the reseller layer typically doesn’t expose.

The 46 Signal Types

A complete taxonomy of commercial events

Rodz categorizes events into 46 signal types organized into major families:

Growth signals: hiring, capital increases, office openings, headcount growth
Transformation signals: executive appointments, mergers/acquisitions, reorganizations, strategy changes
Intent signals: solution research, event attendance, tech stack changes
Regulatory signals: compliance initiatives, certifications, patent filings
Financial signals: funding rounds, published results, investments

Each signal type has its own characteristics: how often it’s detected, how strong a need it reveals, how long it stays relevant, and which sectors it applies to.

A configuration for every signal

The same signal type doesn’t read the same way depending on context. Hiring three people doesn’t mean the same thing as hiring thirty. That’s why each signal is configured with its own filters:

Trigger thresholds: minimum number of hires, minimum funding amount
Sector filters: industry codes, keywords in business description
Geographic filters: departments, regions, countries
Size filters: minimum/maximum headcount, revenue
Role criteria: type of position being hired, seniority level of the new executive
Combinations: pairing multiple signals to produce composite signals

This granularity lets each company configure detection that actually fits their offering and market, rather than receiving a generic feed they have to filter manually afterward.

The Verification Process

From raw signal to actionable signal

A raw event isn’t an actionable signal. Between detection and delivery to the sales rep, several processing steps run in sequence.

Deduplication: the same event often surfaces across multiple sources. Rodz eliminates duplicates so only one signal per event gets delivered.
Contextualization: the signal is enriched with company information (size, sector, history) so the sales rep understands the context at a glance, without extra research.
Balance Scoring: a proprietary model assigns a score combining the nature of the signal (strength of revealed intent) and its recency. That recency coefficient decays after 48 hours.
Deep Search Enrichment: the right contact’s details are identified via a SIRENE, Google Maps, and LinkedIn cascade.
Delivery: the enriched, scored, and contextualized signal goes to the client’s CRM or appears in the Rodz interface.

The 48-hour rule

The full process, detection through delivery, must complete within 48 hours of the event. That deadline isn’t arbitrary. Past that threshold, a signal loses most of its value. The prospect has already been contacted by someone else, has started their own research, or has been pulled into other priorities. An intent signal is only valuable inside that window. Within it, the sooner the outreach goes out, the more likely a reply. Past it, the message lands like any other cold email.

Not All Signals Cost the Same to Produce

This matters for anyone comparing signal prices across suppliers. Different signal types have wildly different production costs, and that gap explains most of the pricing variance you’ll see in the market.

Consider a social engagement signal: tracking when someone in your target list reacts to a relevant LinkedIn post. To detect one new connection or engagement event on a profile, the scrapers have to read the profile’s full current state and compare it against yesterday’s snapshot. Take a contact with 25,000 connections. Day one: 25,000 rows processed. Day two: 25,001. That’s 50,000 rows of data produced and compared to surface a single qualified lead. The per-unit cost on that signal is real.

Compare that to a company registration signal. A new entry in SIRENE is a clean creation event: one row, one signal, no delta comparison needed. The production cost is a fraction of the social case.

This is why social signals carry a higher per-unit price than registry or job-offer signals, and why Rodz qualifies them hard before delivery. Rodz data puts the qualification rate on raw social reactions at roughly 0.8%. Selling unfiltered social data would mean billing clients for 99 rows of noise per qualified lead. That’s the difference between a producer that invests in disqualification and a reseller that pushes raw volume downstream.

The practical implication: when you see a “cheap” social intent signal from a static-database vendor, the question worth asking is whether the disqualification step happened at all, or whether that cost was simply passed to you in disguise, as manual filtering time.

The Numbers That Matter

What the Rodz infrastructure produces:

4 actionable signals per company per year on average
Less time spent per sales rep working out who to contact

And one KPI measures effectiveness: the positive response rate. Not open rates or click rates, which tracking pixels and security filters skew beyond usefulness.

Want to use this infrastructure yourself? The getting started guide with the Rodz API covers authentication and your first request, and the complete API reference lists all available endpoints.

Frequently Asked Questions

How long does it take to configure the signals?

The creation phase takes one to two weeks. It includes an audit of your offering, selection of the relevant signals from the 46 available, and setup of each signal’s filters. The first signals are delivered as soon as that phase is done.

Yes. Rodz only collects data from public sources: registries, press releases, job postings, professional profiles. Each of those publication events is, by construction, the legitimate interest that justifies the outreach. Deep Search enrichment uses professional contact details only. The right to object is offered in every outreach. For more on the regulatory framework, the CNIL guidance on legitimate interest provides useful context on how this basis applies in practice.

What happens if a signal is false or outdated?

Multi-source cross-referencing reduces false positives considerably. A signal detected on a single source gets verified against others before delivery. The Balance model’s recency coefficient automatically downgrades signals that age past the 48-hour threshold, so stale data doesn’t reach the sales rep as if it were fresh.