> For the complete documentation index, see [llms.txt](https://support.parvenu.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://support.parvenu.io/features/flows/sources.md). # Sources Steps that bring new leads into the list. They run once per run, not per lead. Source steps go out and find leads, then write them into the Flow's list. Every Source has a **Save results into list** setting. | Step | What it does | | ------------------------------------ | ----------------------------------------------------------------------------------------------- | | Lead Search | Finds leads by job title, location, industry and company size. | | LinkedIn Search | Imports people from a normal LinkedIn search URL. | | Sales Navigator Search | Imports people from a Sales Navigator search URL. | | Company Search | Finds companies by industry, size, technology or location. | | Maps Search | Finds local storefront businesses and their address, phone, rating and reviews. | | Lookalike Audience | Finds companies similar to your seed companies and adds them to the list. | | Funding Rounds | Finds companies that just raised money and adds them to the list with their round. | | Job Search into List | Finds LinkedIn job posts matching your filters. | | Domain Enrichment | Turns a list of website domains into contacts. | | Enrich Companies From Domains | Fills in company details for every row in the list. | | Find More Contacts | Finds more people at the companies already in the list. | | Take the next leads from a list | Hands the flow the next batch of leads from a list you already have, never the same lead twice. | | Get Next TAM Slice | Takes the next unused slice of a saved TAM audience. | | Check the sample | Scores what the last search found against an ICP and decides whether to keep going. | | Keep going until the list has enough | Counts new leads and loops back for another TAM slice if there aren't enough. | {% hint style="info" %} The credit cost of each step is shown in the Flow builder, and an estimate for the whole run is available before you start it. {% endhint %} ## Lead Search The main way to source people. Filter by job title, location, industry, company keywords, company domains and company size. Turning **Find emails too** on finds a deliverable email for each lead as it goes. Turn it off if you only want the records. **Unique contacts** is on by default: people recently added anywhere in your account are skipped and not charged, and nobody is ever added to the same list twice. Turn it off to fetch everyone again. ### How many leads to pull * **A fixed number** (the default) - pulls the number you type. * **Match the campaign's sending capacity** - reads a Smartlead campaign and sizes the pull to what its inboxes can send. It adds up the daily sending limit of every working inbox (connected, no errors) and multiplies it by **Days of sending to pull**. Leave **Smartlead campaign** as `{{trigger.campaign_id}}` to size the pull for whichever campaign started the run. A campaign with no working inboxes pulls nothing and the run says why. This check is free. **Count follow-ups** (off unless you turn it on, campaign capacity only): each lead gets one email per step in the campaign's sequence, so first emails and follow-ups share the same sending capacity. With it on, Lead Search pulls: (daily sends x days) / emails per lead, minus the leads still waiting for their first email For example, 10 inboxes x 20 a day x 3 days with a 4-step sequence and 20 leads waiting pulls 130 leads, not 600. Emails per lead is the number of steps in the Smartlead sequence; A/B versions of a step do not count. If the campaign already has enough leads waiting, nothing is pulled and the run says so. If the sequence has no steps, or Smartlead cannot be read, the step stops before any search is bought. ## LinkedIn Search and Sales Navigator Search Paste a search URL and choose how many pages to import. A LinkedIn page holds up to 10 people, a Sales Navigator page up to 25. Both can find emails as they import. Re-running a search into the same list never adds anyone twice - people already on the list only have their empty fields filled in. The pages are still fetched and charged again, so only re-run a search when you want new results. ## Company Search Finds companies rather than people. Filter by AI keywords, industries, employee bands and country. Up to 100 per page. ## Maps Search For local businesses with a storefront. Search by what they do and where they are. * **Find contacts** adds people at each business, up to 10 per place and 500 in total. * **Contact job titles** is required when Find contacts is on, and only then. Enter at least one, such as owner, president or general manager. If it is empty, checking the Flow warns you, and a run stops at this step with "Add at least one job title to find contacts". Nothing is charged. A search for businesses only needs no titles. * **Most contacts to return** sets a lower ceiling on contacts for the whole run. The run estimate counts contacts as well as places, so set this to keep the estimate close to what you expect to spend. * You can require the business to have a website, and restrict contacts to the same city. ## Lookalike Audience Finds companies similar to companies you already have, such as your best customers, and adds them to the list as company rows. * **Seeds** - pick a Parvenu list whose company websites are the seeds, or type a comma-separated list of domains. Use one or the other. * **Max results** - the most similar companies to add in one run. * The seed companies are never added. * It adds companies only. To get people at them, put a **Find More Contacts** step after it. More on how it works: [Lookalike Audience](/features/sources/lookalike-audience.md). ## Funding Rounds Finds companies that recently announced a funding round and adds them to the list as company rows, one row per company. * **Round types** - such as Seed, Series A or Growth. Leave it empty for all of them. * **Include other round types** - on by default. Also brings debt, grants, rounds with no type given, rounds not yet confirmed, and raises still in progress. * **Sectors** - such as AI or fintech. Leave it empty for all of them. * **Country** and **US state** - where the company is based. A US state also means the United States. With either one set, companies whose location can't be read are left out. * **Smallest round** and **Largest round** - in US dollars. With a largest round set, rounds with no amount given are left out. * **Investor** - only rounds with this investor, leading or joining. * **Keyword** - words to find in the company name or description. * **How far back the first run looks** - today, the last 7 days or the last 30 days. Later runs into the same list bring only rounds the list has not had yet. * **Most rounds per run** - the most rounds to add in one run. * **Save results into list** - the list the companies go into. How it fits into a Flow: * It adds companies only. To get people at them, put a **Find Contacts** step after it. * The step has no repeat setting of its own. To run it every day or week, use the Flow's **Schedule** trigger. * A company that raises again keeps its one row, gets the new round, and runs through the rest of the Flow again. * Later steps can use `latest_funding_stage`, `latest_funding_amount_usd`, `latest_funding_date` and `latest_funding_investors`, for example in a condition or an AI prompt. More on how it works: [Funding Rounds](/features/sources/funding-rounds.md). ## Job Search into List Finds LinkedIn job posts - by keywords or a LinkedIn jobs URL - filtered by when they were posted, job type, experience level and minimum salary. There is an option to exclude staffing agencies. Minimum 10 results, 50 by default. **Unique jobs only** is on by default, so job posts already saved on the list are skipped and not charged. ### Job titles (searched together) Type several job titles, separated by commas or one per line, for example `marketing coordinator, social media manager`. They are searched together in one LinkedIn search ("A" OR "B" ...), so one step does the work of one job search step per title. * Up to 15 titles go into one search. More than 15 are split into searches of 15 that run one after the other into the same list and share the step's **Max jobs**. * If **Keywords** is also set, a job must match one of the titles AND the keywords. A step with only Keywords works the same as before. * One lead per hiring company is still the default. * The run's "jobs found" total counts every job on the list, including jobs saved by earlier runs, not only what this search returned. ## Take the next leads from a list Use this when the leads already sit in a Parvenu list and a campaign should get the next batch each time it runs. It pairs well with the **Smartlead campaign running low** trigger. * **List** - the Parvenu list to take leads from. * **How many leads per run** - 100 by default, from 1 to 2000. With the Smartlead campaign running low trigger you can enter `{{trigger.need}}` to take exactly what the campaign is short. * **Order** - **By name A to Z**, or **Shuffled** (a mixed order that stays the same every run). * **Saved view** (optional) - only take leads that match a saved view of the list. The flow never takes the same lead twice: it remembers which leads it has taken from that list. Two flows on the same list each keep their own place. When the list is used up, the run finishes and says so. On a manual run, the run's lead limit also caps the batch, so no lead is marked as taken without being worked. ## Domain Enrichment and company sources * **Domain Enrichment** - paste a comma-separated list of domains and get contacts back. * **Enrich Companies From Domains** - fills in company details for rows you already have. * **Find More Contacts** - finds additional people at companies already in your list. ## TAM steps These three work together to work through a large saved audience a slice at a time. {% hint style="info" %} **Get Next TAM Slice** never hands out the same slice twice. It publishes the slice as `{{tam.*}}` for the next step to use. {% endhint %} * **Get Next TAM Slice** - takes the next unused slice of a saved TAM audience. You can tell it to stop when the audience is used up. * **Check the sample** - scores the leads the last search just added against an ICP. If enough of them fit, the Flow carries on; if not, it goes back for the next slice. You set the minimum score, how many of the sample must hit it, and how many times to retry. * **Keep going until the list has enough** - counts the new leads this run has produced and loops back for another slice until it has enough, or until it runs out of tries.