Monday morning, you open the same tab again. Same query typed by hand, same records copied one by one into the same spreadsheet, same fifteen minutes spent hunting for a contact address on a site that hides it. By Friday the file is usable. The following week you start over on the next town. This is not a motivation problem: it is a task that has no business occupying a human being.
An automated prospecting tool exists for exactly that: to replace a sequence of gestures with a sequence of blocks that hand the file to one another. So the useful question is not whether such a thing exists, but which ones, exactly, and how many. Most pages answering this query sell a vague capability. This one does the opposite: it publishes the inventory, by name, with the date it was read.
Read on 15 September 2026The 20 blocks of an automated prospecting tool, named one by one
On 15 September 2026, the outsend library holds 20 block types. That is not a marketing number: it is the number of entries in the registry the product itself serves for public reading, and you can recount it by opening the public outsend block documentation. Each block shows its current status inside the tool.
| Block | Family | What it adds to the row |
|---|---|---|
| Google Maps scrape | Start | Name, address, phone, website, rating and review count, category, listing link. |
| Companies by activity | Start | Establishments drawn from the French Sirene register held by INSEE, by activity and area. |
| Import list | Start | Your own pasted file, a public spreadsheet, or the output of a recent extraction. |
| Email discovery | Enrich | Contact addresses found from the establishment's website. |
| Social profiles | Enrich | Public social accounts linked from the site. |
| Extra phone numbers | Enrich | Numbers present on the site but missing from the listing. |
| Number details | Enrich | Line type and carrier of the number already on the row. |
| SIRET / SIREN | Enrich | French legal identifiers spotted on the company's website. |
| Company data | Enrich | Legal form, dates, declared headcount, from public open data. |
| Legal notice | Enrich | Contents of the site's legal notice page, when one exists. |
| Tech stack | Enrich | Technologies detected on the public website. |
| Ad footprint | Enrich | Traces of running advertising campaigns. |
| Logo & brand assets | Enrich | Logo and brand elements pulled from the site. |
| Pricing | Enrich | The rate card published on the site, when it is public. |
| Google reviews | Enrich | Reviews attached to the listing and whether the owner replied. |
| Bounce check | Verify | Status of an address already sitting in the email column. |
| Dead sites | Verify | Sites that no longer answer, error out, or have expired. |
| PageSpeed | Verify | Site performance score measured by Google's public tool. |
| Filter | Transform | Keeps only the rows matching a rule you write. |
| Sort & cap | Transform | Reorders the file on one column and cuts it to the size you want. |
Three of these blocks are starting points: they create the list. The other seventeen work on a list that already exists. That distinction is what matters when you build your first sequence, and it is also why you cannot begin with any block you like.
The assembly rules a block count never tells you
Short answer: one mandatory start out of three, one predecessor per block, as many successors as you want, twenty blocks maximum in the same run, and no column lost along the way. These constraints are published in the product documentation, not inferred from use.
- Everything begins with one of the three starting blocks. A Google Maps extraction, a selection of companies by activity from the French Sirene register, or an import of your own file. An enrichment block cannot go first: it would have nothing to read.
- A block receives from a single block upstream. Two branches never merge. One block can, however, feed several different continuations, which lets you run two distinct treatments from a single extraction instead of launching it twice.
- Twenty blocks maximum per run. Beyond that, the sequence is rejected. The ceiling is deliberately low: a twenty step sequence is already hard to read back, and two short sequences are easier to fix than one long one.
- No block ever drops a column. Every step outputs everything it received, in the original order, and appends its own at the end. Custom columns from an imported file travel through untouched. Only the operations you explicitly ask for, a filter or a cap, remove rows, and never columns.
That last rule is what separates a tool from a series of exports reloaded by hand. When you assemble an extractor, an address finder and a verifier yourself, every handover between them is a place where a column vanishes, where the order shifts, where you go back to the spreadsheet. That is the work the outsend all in one prospecting platform removes, far more than the extraction time itself.
What the law asks of an automated prospecting tool
Automating how a file is built does not change the legal status of that file. Two texts set the rules, and both can be read in full.
For business to business outreach, the French data protection authority publishes its position on commercial prospecting by electronic means, updated on 10 June 2026. It states that prospecting aimed at professionals may rest on the legitimate interest of the organisation when the subject of the approach relates to the profession of the person contacted, and it sets two rules: the person must be informed, for example when their address or number is collected, and objecting must be simple and free of charge. The same page adds two points most articles on the subject leave out. First, where the data is already held by the company or acquired from third parties, you must make sure the person was informed and is able to object. Second, generic addresses such as contact@ or info@, which belong to legal entities rather than to individuals, fall outside those principles.
The second text sets a deadline, and it applies well beyond France. When personal data has not been obtained from the person themselves, which is the case for any list built from public sources, Article 14 of the GDPR requires informing them within a reasonable period after obtaining the data, and at the latest within one month, or, if the data is used to communicate with that person, at the latest at the time of the first communication. The full text is readable in Chapter III of the Regulation, Article 14(3). In practice this means the first message carries the information itself: where the data came from, why you are writing, and how to stop receiving messages. Our definition of cold email and its GDPR framing walks through the email case in detail.
Two sequences a solo operator can rebuild
Find the local businesses whose website needs redoing
Google Maps scrape then Dead sites then PageSpeed then Filter then Email discovery. You start from a trade and a town, drop the addresses that no longer answer, keep the slowest sites, and only look for a contact address on what is left. Five blocks, one start, and the Google Maps extraction block ran exactly once.
Revive an eighteen month old file
Import list then Dead sites then SIRET / SIREN then Company data then Sort & cap. The file you built by hand comes back with its original columns intact, plus the real state of every website and the current legal identity of every company. It is the same work described on the SIRET and SIREN enrichment page, applied to the whole list instead of one row.
In both cases the output is a file you export as CSV, XLSX or JSON, carrying the starting columns and everything each block appended. Nothing is sent to anyone: the tool builds the list, sending stays with you.
How to recount this inventory yourself
This count of 20 is dated because it moves. One block added next month takes it to 21, and this page becomes wrong that same day if nobody rereads it. Which is precisely what happened here: the previous version of this page announced 8 blocks, a figure that was correct the day it was written and has been wrong for a long time.
So you do not have to take our word for it. The public block documentation lists every type with what it accepts as input and what it produces as output: count the rows of the "Node types" table. The registry the product serves for reading is where that table comes from, and it answers without an account.
A method note while we are here: the same demand applies to the tools you compare us against. A connector count printed on a sales page with no date is not verifiable. Our comparison of prospecting tools applies that rule to fourteen products, figure by figure and source by source. And if you would rather judge the output than the promise, the shortest route is to request free alpha access and run a sequence on your own area.
Frequently asked questions
Do you need to write code to assemble these blocks?
No. Blocks are dropped and connected in a visual editor, and each one opens a settings panel with named fields: a trade and an area for the Google Maps extraction, a rule for the filter, a column and a number for the sort. The constraints described above are checked the moment you save, with a message saying what is wrong.
Can you start from a file you already own?
Yes, through the import block, one of the three possible starts: a pasted CSV, a public spreadsheet or the output of a recent extraction. Your custom columns travel through the whole sequence untouched.
Does an automated prospecting tool send the messages?
The outsend blocks stop at the file. They build, enrich and check the list, then export it. Sending happens afterwards in the messaging tool of your choice, which keeps at your end the record of what went out and the handling of opt out requests, two things regulators expect from the sender rather than from the extraction tool.
What happens when a block finds nothing on a row?
The row continues with an empty column. No step deletes a row that failed to answer it: the filter, and only the filter, removes rows, on the rule you wrote. That is what lets you measure the real coverage rate of an enrichment instead of guessing it.
Are all 20 blocks useful to everyone?
No, and that is normal. A property agent will never touch the technology detection block, while a web agency will build half its lists around it, as described on the website technology detection page. The full inventory mostly helps you spot the two or three blocks that match your trade.
Will this count of 20 still hold in six months?
It holds for 15 September 2026, the date printed at the top of this page. The library keeps growing, so the number will rise. The only honest answer to this question is the one in the previous section: the current list is readable in the public registry the product itself serves, of which the documentation is only a summary.