> ## Documentation Index
> Fetch the complete documentation index at: https://docs.salesfinity.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Build a Look-Alike Account List

> Seed a search with your best customers, a pasted set of companies, or a single competitor, and get a list of companies that look like them.

**You will end up with:** a company table of look-alikes, appended to or separate from the
seed list, ready to score and to find contacts at.

**Before you start:** similarity search needs a look-alike data provider connected under
**Settings → Connections & API → Data Providers**, such as Ocean. Without one the companion
falls back to semantic discovery and says so. Seeding from closed-won needs a CRM connected.

## Three ways to seed

| Seed                                           | Prompt shape                                            |
| ---------------------------------------------- | ------------------------------------------------------- |
| Your closed-won accounts                       | "companies like our closed-won accounts from last year" |
| A list you paste or already have on the canvas | "more companies like the ones on this list"             |
| One named company                              | "competitors of Stripe", "alternatives to Gong"         |

Every one of those phrasings is the same ask, similarity, and routes to the same search.

## Steps

<Steps>
  <Step title="Load the seed">
    ```text theme={null}
    Load our closed-won accounts from the last 18 months.
    ```

    Or paste names:

    ```text theme={null}
    Put these on a list: Rippling, Deel, Gusto, Justworks.
    ```
  </Step>

  <Step title="Ask for look-alikes">
    ```text theme={null}
    Find 100 companies that look like these.
    ```

    The whole seed table's domains are used at once. Add constraints when you want them:

    ```text theme={null}
    Find 100 companies like these in the US, 51 to 500 employees, in SaaS.
    ```
  </Step>

  <Step title="Or seed with a single competitor">
    ```text theme={null}
    Who competes with Orum? Build a list.
    ```

    The companion resolves the company to its real website first, so a same-named business is
    not substituted.
  </Step>

  <Step title="Remove the ones you already know">
    ```text theme={null}
    Remove any that are already in HubSpot, and any that are on the seed list.
    ```

    Both checks are free. Overlap with another table and CRM matching use the columns and IDs
    already on the rows.
  </Step>

  <Step title="Score and fan out">
    ```text theme={null}
    Score these against our "Outbound dialer ICP" profile, then find the VP of Sales and
    Director of Sales Development at the top 30.
    ```
  </Step>
</Steps>

## Look-alike people

The same idea works for people. Seed with a few of your best contacts and ask for more like
them; the companion returns people with similar roles and backgrounds for you to review.

```text theme={null}
Find people like our top 10 champions on this list.
```

## What happens behind the scenes

Similarity search sends the seed domains to the provider, which returns companies close to
them in industry, size, and description. It is discovery only: it never returns phones or
emails, so it is always followed by a contact search and, if you want, enrichment. The
resulting rows carry firmographics so you can trim by size or geography before spending on
contacts.

## Prompts to try

```text theme={null}
Alternatives to Salesloft with under 200 employees, then score them.
```

```text theme={null}
Companies like the 20 accounts we booked meetings with last quarter.
```

```text theme={null}
Start a new list: look-alikes of our enterprise wins only, excluding anything already in Salesforce.
```

## Next

<CardGroup cols={2}>
  <Card title="Score your accounts" icon="ranking-star" href="/tutorials/score-accounts">
    Rank the look-alikes before you spend on contacts.
  </Card>

  <Card title="Find contacts at target accounts" icon="users" href="/tutorials/find-contacts">
    Multi-thread the ones that score well.
  </Card>
</CardGroup>