How does LinkedIn auto connect work?
Auto connect is software that sends LinkedIn connection requests from your own account on a schedule. You give it a list or a saved search and set a daily number. It then presses Connect for you, with or without a short note. Most tools can also send follow-up messages once someone accepts. Auto invite and auto add are the same thing under other names.
Which brand you pick matters less than three other choices:
- Who holds your login and session while the tool runs.
- What data the tool pulls, and who answers for it.
- Whether your campaign has earned more volume yet.
A tool can get your account restricted, and even a vendor says so. Linked Helper's own pricing page warns that, whatever the tool's settings, LinkedIn can still apply limits or restrictions to individual accounts (Linked Helper pricing).
I rank no tool here and quote no ban rate. A ban rate needs a method and a sample, and I will not borrow one from someone selling the tool.
Your buyer never sees the tool. They see a request from a person, then a profile, then a message. My line is simple: the clicking and scheduling can run on software, but the words the buyer reads and every reply get a human.
I treat the connection note as the first message, so I never let it pitch.
If you add a note, give one specific reason to connect in a single line. The offer waits for the first message after acceptance, and even there you sell a short call and keep the service for later.
Extension, cloud, desktop or GitHub bot: who holds your login?
An auto connect tool can run in four ways. Each one puts your account in different hands.
| Tool type | Where it runs | Who holds your login or session | Must your computer stay on? | Where collected data sits | Main risk |
|---|---|---|---|---|---|
| Chrome extension | Inside your browser | You, but the extension can read the LinkedIn pages you open, depending on its permissions | Yes, with the browser open | With the vendor, if the extension enriches or exports | Data collection you never planned, as the KASPR case below shows |
| Cloud tool | The vendor's servers | The vendor runs a live session for your account | No | The vendor's servers | A third party holds standing access to your account |
| Desktop app | Your computer or a server you rent | You | Yes, the machine and the app | Your machine, unless you pick a cloud storage option | Your own device and server security |
| GitHub script | Your computer | Whatever the script's settings file asks for, sometimes your password | Yes | Wherever the script writes it | Credentials in a plain file, no support, breaks when the page changes |
Linked Helper is a clear desktop example. Its pricing page describes desktop software that you run on your own machine or on a rented server (a VPS) under your control, and the machine must stay on with the app open. The same page offers local storage tied to one device, or cloud storage hosted by Linked Helper on US-based servers, so know which one your workspace uses.
Expandi is a cloud example. It describes itself as cloud-based software with a dedicated country-based IP address (Expandi pricing). That is convenient. It also means the vendor runs the session for you.
Open-source scripts need their own check. Before you run one, read its setup notes. One popular auto connector bot on GitHub has you type your username and password straight into its Python file. If it asks you to type your LinkedIn username and password into a settings file, anyone who opens that file, a backup of it or a shared folder holding it has your account. Keep such a file off shared drives and public repositories, and change the password when you stop.
Free connection boosters follow the same logic. A tool that costs nothing may be paid for in the data it collects or the access you grant it. Before you install one, ask who holds the session and where the profiles it reads end up. If the vendor's page cannot answer both, leave it uninstalled.
If your account gets flagged anyway, know what to do when LinkedIn restricts your account before it happens.
Why do vendors publish such different daily caps?
A vendor cap is a product setting, not a platform rule. Each number is the vendor's own choice, so compare how each product enforces its limit before you compare the numbers.
Linked Helper shows this clearly. At the time of writing, its $15 Standard plan caps a set of advanced actions at around 20 a day, and the Pro tier lifts those caps. Its limits count on a rolling 24-hour window instead of resetting at midnight. The page does not say why 20 was chosen. PhantomBuster's Auto Connect page recommends about 20 requests per working day and puts LinkedIn's weekly limit at about 100, depending on account activity. One open-source auto connector bot on GitHub tells users not to exceed 80 a week.
Expandi takes the opposite route. You set your own daily limits for connections, messages and InMails, and the tool will not go over them. There, the cap is your own risk appetite, enforced by software.
None of these is an official LinkedIn number.
Scale volume last, not first. On my own campaigns I touch the daily setting last, and only after acceptance holds at a low volume.
As a reference point, we ran LinkedIn outbound for a US business consulting and coaching firm, after optimizing its founder's profile. 38 of 98 connection requests were accepted, a 39% connection rate, and the campaign booked 204 sales calls over 4 months.
Those calls were meetings booked through LinkedIn. A campaign whose acceptance is far lower at low volume usually has a profile, list or note problem. Test it small first. A weak result at 100 requests will not improve at 1,000. A tool only copies the miss at scale.
You do not need thousands of requests to know where you stand. That 39% came from fewer than 100 requests. A batch of about 100 is enough for a first read. Acceptance also swings by audience. On one tracked campaign for a software development firm in Germany, 52 of 84 requests were accepted, 61.9%. That client booked 488 sales calls in 6 months.
That gives a simple warm-up rule:
- Send your first 100 or so requests at the lowest setting your tool allows, or by hand.
- Count acceptances from your target titles only.
- If acceptance sits well below the 39% to 62% those two campaigns reached, change the profile, the list or the note, and run another batch. Treat both as reference points, not a benchmark. A very senior or niche audience can accept less, so also compare against your own earlier batches.
- Raise the daily setting in small steps, and only while acceptance holds.
What do auto connect tools cost per month?
The table shows each vendor's published entry prices, taken from its own pricing page at the time of writing (PhantomBuster from its own pricing guide, dated July 2026). Currencies and billing bases are the vendor's own, and most plans price one sender.
| Tool | Published price at time of writing | What one price covers | Free trial |
|---|---|---|---|
| Linked Helper | Standard $15 a month; the page lists no separate monthly Pro price | One license = one LinkedIn account | 14 days of full Pro access, no card |
| Expandi | Business $99 a month, or $79 a month billed annually | Not stated per seat; agency pricing is custom | 7 days |
| SalesRobot | Basic $59, Advanced $79, Professional $99 a month; $39, $59 and $79 billed annually | Per LinkedIn account | 14 days, no card |
| Dripify | Pro $79 and Advanced $99 a month; $59 and $79 billed annually | Per user | 7 days, no card |
| HeyReach | Growth $79 a month, or $63 a month billed yearly | Per sender | Offered; length not given on the plan card |
| Meet Alfred | Individual $29, Teams $39, Business $49 a month, billed annually | Per user | Offered; terms not given |
| PhantomBuster | Start $69 a month, or $56 a month billed annually | Per workspace, shared by everyone in it | 14 days, no card |
| Waalaxy | Pro €19 (300 invitations a month), Advanced €49 (800), Business €69 | Per user, per month | 14 days; no free plan listed |
| Octopus CRM | Two price sets on the page, 9.99 to 39.99 and 6.99 to 24.99 a month, on Monthly and Annually (-35%) tabs; no currency symbol shown | Per month, per plan | 7 days, no card |
Watch two details. At the published 45% for a 12-month term, Linked Helper's $15 drops to $8.25 a month per license. And because each license covers one LinkedIn account, three people sending from three accounts need three licenses: $45 a month before any discount.
Find the billing basis before you compare headline numbers. Per user, per sender, per account and per workspace sound alike and price very differently once a team grows. A per-workspace plan can look expensive for one sender and turn cheap at five.
Can a LinkedIn tool get your company in trouble with regulators?
Yes, through the data it collects. Regulators have acted on enrichment features that pull contact details.
On 5 December 2024 the French data protection authority, the CNIL, fined KASPR 240,000 euros (CNIL). KASPR sells a paid Chrome extension that pulls contact details of LinkedIn users whose profiles its customers visit, into a database of about 160 million contacts. Three findings matter to anyone running a LinkedIn tool:
- It also collected details of people who had limited their visibility to 1st and 2nd-degree connections, which the CNIL found unlawful.
- It kept data for 5 years from each update, such as a job change, which the CNIL judged disproportionate.
- It told the people concerned only in 2022, four years after the extension launched, by an email in English.
KASPR was ordered to stop collecting that limited-visibility data and delete it, with a deadline of 18 June 2025. The decision covers data collection by an extension. It says nothing about connection requests themselves.
The GDPR articles behind a case like this apply to your own campaign data too (GDPR on EUR-Lex):
- Article 5(1)(c), data minimization: keep only data that is adequate, relevant and limited to what you need.
- Article 6(1)(f), legitimate interests: a lawful basis only where the person's interests and rights do not override yours.
- Article 14(3)(b): when you use data you did not get from the person to contact them, you must give them the required information at the latest at your first communication. Article 14(2)(f) adds that you tell them where the data came from.
The UK adds its own layer. Guidance from the UK Information Commissioner's Office (ICO) on B2B marketing states that direct messages on social media count as electronic mail (ICO).
Sole traders and some partnerships are treated as individuals. Sending them electronic marketing requires consent. The soft opt-in is the other route. It covers details collected during a sale or sale talks, with a clear chance to opt out. Public data, social media included, is still personal data. Anyone can object to direct marketing, and that right is absolute. The ICO also notes that someone using a professional network in a personal capacity is not receiving B2B marketing.
What that means for your setup:
- Switch off enrichment add-ons, the features that look up emails and phone numbers, unless you need them, and export only the fields you will use.
- Plan how your first message, or a notice it links to, tells people where their details came from.
- Keep one exclusion list across every tool and every sender, and honor an objection the same day.
- In the UK, treat sole traders as individuals.
This is not legal advice.
When should you not automate connection requests at all?
| Situation | What happens if you automate | Verdict |
|---|---|---|
| Your profile reads like a CV | People open it, see a job history and leave | Fix first |
| Acceptance is well under your reference rate at low volume | The same miss, at ten times the volume | Fix first |
| The note or first message pitches your service | You get ignored faster and at scale | Fix first |
| Nobody reads and answers replies the same day | Accepted connections go cold in the inbox | Do not automate |
| You expect a brand-new campaign to bring clients by itself | Weeks of sending, then a hunt for a better tool | Do not automate |
| Acceptance holds, replies get answered, the message sells a call | More of what already works | Automate |
The profile row comes first because that is where every request lands. Open the sender's profile before the message, and judge it as a page built for clients. A job history does not do that job. A strong profile with an average message beats a perfect message sent from a profile that looks abandoned.
The reply row is the one founders skip. Software can send, but a two-word answer like "maybe later" still needs a person to decide whether it means yes, maybe or go away. One automation on its own moves nothing; the whole chain from list to reply has to work.
The fifth row is the classic failure: a campaign switched on, nothing back, and then a search for a different tool. The tool was never the problem. The list or the offer usually is. Push 5,000 "personalized" messages a week on broken foundations, and they still get ignored.
What a first message that sells the meeting looks like
A made-up example, from a software firm to a head of IT at a UK logistics company.
Connection note: "Hi Maria, liked your post on warehouse systems. Glad to connect."
First message after she accepts:
Thanks for connecting, Maria. Is warehouse or routing work planned for next year? If so, a 15-minute call next week could cover it.
One reason to talk and one small ask. The service list waits for the call.
How do you scale once the manual version works?
Scaling means adding senders to what already works: more LinkedIn accounts, each with its own license or seat, running the message that passed the acceptance test. If you would rather a senior team ran the sending and the replies, that is what our LinkedIn outreach service does.
Split the work before you add anyone. I divide territory per account and per rep, which keeps each lead with a single sender. Split by country, by vertical or by title band, whichever matches how your buyers cluster. Load the shared exclusion list into every account on day one.
Then order the channels. The LinkedIn touch comes first. The warm email, sent to someone who has already seen your name there, follows it. Never the other way round.
Fix the funnel first. Then scale. If your test batch passed, the price table shows what each extra sender will cost.
Thinking about automating LinkedIn requests and unsure your profile is ready for it? Book a call and send me the profile. I will tell you what I would fix before you switch anything on.
Questions founders ask before switching it on
Do team discounts change which tool is cheapest?
They can. Linked Helper gives 10% off at 10 to 19 licenses, rising to 50% at 1,000+, and multiplies that with its term discount. At the time of writing, its page prices 20 Standard licenses for 12 months at $1,584 instead of $3,600. HeyReach's Agency plan is $999 a month for 25 senders, under $40 a sender.
What happens when a free trial ends?
It depends on the vendor, so read the trial terms before you load a list. Linked Helper pauses your campaigns and charges nothing. Dripify charges nothing and the account goes inactive until you pick a plan. On PhantomBuster, moving to a paid plan during the trial ends the trial and starts billing at once.
How long can I hold profile data before telling people?
Under GDPR Article 14(3)(a), you must give people the required information within a reasonable period after you obtain their data, and never later than one month afterwards. If you contact them sooner, the deadline moves up to that first message. Data sitting unused in a tool still starts that one-month clock.
What should happen to collected data when a campaign ends?
Delete what you no longer need, because data minimization applies after the campaign as much as during it, and regulators look at retention periods. Keep the exclusion list, though. People who objected must stay excluded the next time anyone on your team runs a campaign, whichever tool they use.