# Connection requests for leads imported from CSV
> How to turn a CSV of LinkedIn profile URLs into connection requests: cleaning the list, writing the note, pacing sends and handling the replies.
- HTML: https://getkinlead.com/connection-requests-for-leads-from-csv-imports
- Markdown: https://getkinlead.com/connection-requests-for-leads-from-csv-imports.md

**Short answer:** Put the full LinkedIn profile URL for each person in one column of a spreadsheet, save it as a CSV, and import it into whatever tool sends your invites. The URL is the only reliable identifier, because names repeat and people change them. Clean the list first, then send at a pace you can actually keep up with.

## Where the profile URLs come from

A CSV of profile URLs is only as good as the list behind it. Most people start from one of four places: a Sales Navigator search they exported, the commenters on a competitor's post, an attendee list from an event, or profiles they saved by hand over a few weeks. Each one needs different cleanup before it's worth importing.

Sales Navigator exports are the easiest starting point because the file already has a profile URL column next to name, company and title. Pulling links out of a post's comments is slower, since you're copying them one at a time, and it gets painful past a few dozen people.

Hand-collected lists are the messiest. Someone pastes "Sarah Chen" and assumes the tool will work out which Sarah Chen. It won't. Names repeat, people change them after a job move, and two people at the same company can have near-identical profiles.

Whatever the source, the file needs one column of full URLs and nothing else that the importer has to interpret. Everything else in the sheet is for you, not the tool.

```
profile_url,first_name,company,why_they_are_here
https://www.linkedin.com/in/sarah-chen-logistics,Sarah,Northwind,Commented on the routing post
https://www.linkedin.com/in/marcus-webb-acme,Marcus,Acme,Met at the warehouse event
https://www.linkedin.com/in/priya-nair-northwind,Priya,Northwind,Downloaded the freight report
```

## Clean the list before you import it

Strip everything after the username. LinkedIn URLs pick up tracking strings when you copy them from a feed, so you end up with something like linkedin.com/in/sarah-chen?trk=public_post_comment. Some importers handle that and some don't, and a row that fails quietly is a wasted invite.

Check the URL type. /in/ is a person. /company/, /school/, /posts/ and /pulse/ are not. A file that mixes them will either error on import or, worse, send a request to something that isn't a human.

Deduplicate before you upload. The same person often appears twice with different tracking strings, and sending two invites to one person is one of the fastest ways to look automated. A quick dedupe on the username column catches almost all of it.

If you're doing this often, the cleanup is the part worth automating. There's a walkthrough on [automating LinkedIn lead cleanup](/automating-linkedin-lead-cleanup) that covers the checks in order.

**Rows that break the import**

- Sarah Chen, Northwind
- linkedin.com/in/sarah-chen?trk=public_post_comment
- https://www.linkedin.com/company/northwind
- linkedin.com/in/sarah-chen (duplicate of an earlier row)

**Rows that import cleanly**

- https://www.linkedin.com/in/sarah-chen-logistics
- https://www.linkedin.com/in/marcus-webb-acme
- https://www.linkedin.com/in/priya-nair-northwind

> One column, one URL per row, no spaces. If a row has a name and no URL, it isn't a lead yet.

## Sending by hand versus sending with a tool

Manual sending is fine for a short list you built this morning. You open each profile, click connect, write the note, and move on. It's slow, and the slowness is the point: nothing about it looks like a script.

The trouble starts when the list is bigger than the time you have. That's when people either let the file rot in a folder or start clicking fast enough to trip something. Neither is a plan.

A tool doesn't make the outreach better on its own. It makes it repeatable. If your note is weak or your list is wrong, a tool just delivers the mistake at volume.

| | What changes | Sending by hand | Sending with a tool |
| --- | --- | --- | --- |
| Time for fifty invites | Most of an afternoon | Minutes of setup, then it runs |
| Consistency | Whatever your calendar allows | The same volume every day |
| Personalisation | Whatever you remember to write | Pulled from the campaign brief you wrote once |
| Tracking | A spreadsheet you forget to update | Logged per lead as it happens |
| Risk | Low, if you're honest about pace | Entirely down to the tool's pacing |

## Writing a note people accept

A blank request is a coin flip. A note that says "I'd love to connect and learn about your business" is worse than blank, because it reads like a template and the reader knows it.

The note that works names the reason this specific person is on your list. They commented on a post about warehouse routing, they spoke at an event, they run a team that just changed tools. One sentence of context, one sentence of who you are, stop.

Don't pitch in the note. The invite is a door, not a meeting. Save the offer for the message after they accept, when they've already said yes to hearing from you.

- **draft:** Hi Sarah, your comment on the last-mile routing post stuck with me, we ran into the same thing at Northwind. Would be glad to connect.
- **them:** Thanks for reaching out. Which part did you hit? We're still arguing about it internally.
- **you:** The bit about drivers ignoring the optimised sequence. We ended up changing how routes get assigned rather than the routes themselves. Happy to share what that looked like if it's useful.
- **them:** Yes please. Send it over.

> Keep the note under 200 characters. It gets cut off on mobile, and a truncated sentence reads like a mail merge.

## Pacing requests so the account stays healthy

LinkedIn doesn't publish a number for invites per day or per week. Anyone quoting one is guessing or repeating someone else's guess. What the platform does watch is behaviour: bursts of activity, acceptance rates that fall off a cliff, and a pending pile nobody ever answers.

The practical approach is to pick a daily number you can sustain for a month, spread it across your working hours, and stop when acceptance drops. A steady twenty that get accepted beats a hundred that get ignored.

Withdraw invites that have sat unanswered for a few weeks. A pending pool full of dead requests tells LinkedIn your outreach isn't landing, and it also blocks you from inviting those people again later.

More on the signals and how to read them is in [LinkedIn account safety](/features/linkedin-account-safety).

| | Signal | What it usually means | What to do |
| --- | --- | --- | --- |
| Acceptance rate falling | The note or the targeting is off | Rewrite the note, narrow the list |
| A large pending pool | People are ignoring rather than declining | Withdraw the oldest invites |
| A warning from LinkedIn | The account has been flagged | Stop sending on that account for a while |
| Replies but no meetings | The follow-up is the problem, not the invite | Change what you send after they accept |

> If LinkedIn pushes back, stop. Sending through a warning is how a temporary limit turns into a restriction.

## What happens after they accept

The invite was never the goal. The first message after acceptance is where most lists die, because people send a pitch that has nothing to do with the note they wrote two days earlier.

Reference the note. If you mentioned their comment about routing, open with the routing question. Then ask something small and specific: is it still a problem, who owns it now, have they changed tools since.

Keep the first message short enough to answer from a phone. Long messages get archived and forgotten, and you only get one first impression per person.

1. **Message within a day.** They just accepted, so your name is still on screen. A message a week later reads like a broadcast.
2. **Ask one question, not three.** A message with four questions gets answered with none. Pick the one you actually care about.
3. **Write down what they said.** Even a no is useful. The next campaign should either skip them or open with what you learned.
4. **Stop after two follow-ups.** If two follow-ups get nothing, the answer is no. Leave the thread alone and move on.

## Doing this with Kinlead

Kinlead imports a CSV of LinkedIn profile URLs into a [lead group](/features/lead-groups-and-scoring), and any campaign can point at that group. Leads are scored from 0 to 100 against the customer profile you describe, and anything below the fit line isn't kept, so a rough list gets filtered before you send anything.

Outreach runs from your own LinkedIn account through Unipile, and Kinlead never stores your password. Each account sends at most one action every 5 minutes, inside daily invite and message limits you set and inside a send window in the lead's time zone. Connection notes are AI-written, under 200 characters, and open with why that person became a lead.

1. **Import the file.** Upload your profile URLs into a lead group. Leads can also come from a search or from the Steal Customers agent, which reads comments on competitor posts.
2. **Set the campaign.** Choose the goal, tone and an optional sales playbook, then pick how replies are handled: hand off at the first reply, AI replies until there's real interest, or AI replies until a meeting is booked.
3. **Set your limits.** Pick daily invite and message limits and a send window. Stale pending invitations are withdrawn automatically to keep the pool healthy.
4. **Read before it sends.** Queued messages can be edited on the Actions page, and any [campaign](/features/campaigns-and-send-windows) can be paused.

> Check the first twenty queued notes before the campaign runs. If the AI misread your offer, it will misread it twenty times.

## Related

- [Lead groups and scoring](https://getkinlead.com/features/lead-groups-and-scoring.md)
- [LinkedIn account safety](https://getkinlead.com/features/linkedin-account-safety.md)
- [AI LinkedIn outreach](https://getkinlead.com/features/ai-linkedin-outreach.md)
- [Automating LinkedIn lead cleanup](https://getkinlead.com/automating-linkedin-lead-cleanup.md)

## Common questions

**Can I import names instead of profile URLs?**

No. Names aren't unique enough, and two people at the same company can have near-identical profiles. You need the specific URL to be sure the request reaches the right person.

**What happens if a URL in my CSV is broken?**

Most tools skip that row and carry on with the rest. Check the file for typos and stray tracking strings before you upload, because a skipped row is an invite you never sent.

**Do I need Sales Navigator to send requests from a CSV?**

No. A standard LinkedIn account can send connection requests to imported profile URLs. Sales Navigator is useful for building the list, not for sending from it.

**How many connection requests should I send a day?**

LinkedIn doesn't publish a number, so anyone giving you one is guessing. Pick a daily figure you can keep up for a month, spread it across working hours, and cut back if acceptance starts falling.

[Create a Kinlead account](https://getkinlead.com/signup) or [book a call](https://getkinlead.com/demo) to talk through pricing and setup.
