Jorge LewisYou can get LinkedIn outbound results in your terminal by asking a coding agent that has the GTM Bud MCP server to call get_campaign_results. The tool returns the results of one campaign, and each rate comes with its numerator, its denominator, its definition and an as-of time, so the number explains itself without a dashboard.
This guide is for a GTM engineer who works in Claude Code or Cursor and wants outbound results as data. It covers what the tool returns, how to ask for a weekly review across campaigns, how to read the numbers, and the limits. For everything else the agent can do, see AI LinkedIn outreach for GTM engineers.
What does get_campaign_results return?
get_campaign_results is a read tool that returns the results of one campaign: the counts, the rates that are calculated from them, and the time of the calculation. It changes nothing, so the agent can call it as often as you ask.
The counts cover the steps of a LinkedIn campaign: leads in outreach, leads reached and conversations by channel, connection requests sent and accepted, and replies. Replies are labeled Interested, Warm or Not interested, and the results also show a count of replies that are not yet labeled. Each rate comes as four parts.
| Part | What it is | Why you want it |
|---|---|---|
| Rate | The percentage | The headline number |
| Numerator | The count on top | Shows how many events are behind the rate |
| Denominator | The count below | Shows if the rate rests on 12 leads or on 1,200 |
| Definition | One sentence on what is counted | Stops the argument about what “reply rate” means |
The fifth part belongs to the whole result: the as-of time, which is the time the numbers were calculated.
Here is a sketch of one answer as the agent prints it. The layout and the rate names are illustrative, and the counts are left out because they would be your numbers.
campaign UK and Ireland legal leaders (active)
as of 2026-10-05 09:12 UTC
acceptance rate
numerator connection requests accepted
denominator connection requests sent
definition stated by the tool in one sentence
interested reply rate
numerator interested replies
denominator conversations
definition stated by the tool in one sentenceWhy the denominator matters most. A rate from a dashboard screenshot hides its base. Two people can both say “reply rate” and divide by different things: all leads, or only the leads who accepted. When the definition travels with the number, that mistake cannot survive a copy and paste into a document.
How do you ask an agent for a weekly review across campaigns?
Ask the agent to list the campaigns, call the results tool once for each one, and write one table with the as-of time in each row. Results are per campaign, so the loop is the agent’s job.
A session looks like this. The tool names are the real ones.
you Weekly review. List my campaigns, get the results for each
active one, and write a table to outbound/weekly/2026-10-05.md.
One row per campaign. Include numerator and denominator for
each rate, and the as-of time.
agent list_campaigns
4 campaigns: 3 active, 1 ended.
agent get_campaign_results x3
Results returned for 3 campaigns.
agent Wrote outbound/weekly/2026-10-05.md
you Compare with last week's file. Which counts moved the most?
agent Read outbound/weekly/2026-09-28.md
Comparison written below the table.
you For the campaign with the most interested replies, how many of
them have no answer yet?
agent list_replies
Unanswered Interested replies listed.Three habits make the review useful over time:
- Save the prompt. Put it in a file in the repo, for example
outbound/weekly-review.md. The review is then the same each week, and a colleague can run it. - Keep one file per week. The connector returns the numbers as they are now. It does not return last week. Your history is the files that you keep.
- Store counts, not only rates. With numerators and denominators in the file, you can add campaigns together later. You cannot add percentages.
What do the numbers mean?
Read the results as a chain: each rate tells you about one step, and the step with the weakest number is where to work next. Do not compare your rates with a number from the internet before you have checked the denominators.
| What you see | Where to look | What to ask the agent |
|---|---|---|
| Few connection requests accepted | The audience, or the sender profile | “Start a prospect preview and show me who matches.” |
| Requests accepted, but few replies | The first message and the value offer | “Show me the message sequence of this campaign.” |
| Replies, but few positive ones | The offer, or an audience that does not have the pain | “List the replies labeled Not interested and group the reasons.” |
| A rate that jumps from week to week | A small denominator | “Give me the denominators for both weeks.” |
Two cautions. First, a young campaign has small denominators, and a rate on a small base moves a lot with one event. Compare campaigns of a similar age. Second, an interested reply is a label, and the label is as good as the judgment behind it. Spot-check the reply labels before you report the number upward.
For reference ranges, see our articles on LinkedIn outreach benchmarks and outbound KPIs. To connect the counts to cost, use the method in how to measure outbound ROI.
A worked example: one campaign, one review
The company in this example sells contract review software. The GTM engineer owns the weekly outbound numbers. The names below are invented.
The audience sentence. “General counsel and heads of legal at technology companies with 50 to 300 employees in the United Kingdom and Ireland.”
What the criteria check said. The check passed every criterion. The agent asked two questions of its own. Should “legal counsel” without “head” or “general” in the title be included, since at a company of 60 people that person is often the whole legal team? And should a company with its head office in another country and an office in Dublin count? The answers: yes to the first, no to the second.
Two previewed prospects.
- Aoife M., General Counsel at a payments software company with about 180 employees in Dublin. Marked Good fit: right role, right size, head office in Ireland.
- Daniel P., Legal Counsel at the 40-person London office of a company with its head office in the US. Marked Not a fit: the buying decision is made in another country.
One reply draft. A lead wrote: “We use outside counsel for anything complex. What would this replace?” The agent read the conversation with get_conversation, followed the campaign playbook and wrote, for review:
“It does not replace outside counsel for the complex work. It takes the first read of standard agreements, such as NDAs and order forms, so that fewer of them go outside at all. Would it help to see it on one of your own NDAs?”
The engineer showed the draft to the sales lead, then approved it in the session, and the agent sent the approved text.
The results question. “Give me the results for the UK and Ireland legal campaign. Print each rate with its numerator, denominator and definition, and tell me the as-of time. Then append the row to this week’s file.”
What the review found. The acceptance step looked healthy against the team’s own earlier campaigns, and the interested reply rate rested on a small denominator. The engineer wrote “too early to judge” next to it. That note is the value of having the base next to the rate.
Five prompts for results in the terminal
- “Get the results for the UK and Ireland legal campaign and print every rate with its numerator, denominator and definition.”
- “Run the weekly review from
outbound/weekly-review.mdand write the table to a new file named with today’s date.” - “Which of my active campaigns has the oldest as-of time? Ask for its results again and show me if anything changed.”
- “Add the numerators and denominators of the three legal campaigns and give me one combined acceptance rate. Show the sum.”
- “Create a campaign draft for general counsel and heads of legal at technology companies with 50 to 300 employees in the United Kingdom and Ireland, so that I have a second legal campaign to compare. Stop after the criteria check.”
Three objections, and the honest limits
“I want this in my warehouse, not in a markdown file.” The connector is an interface for an agent in a session. It is not an export feed. There are no webhooks and no API beyond the connector on offer here. The agent can write the results to a file or to a table that it has access to in your environment, each time you ask. If you need a pipeline that runs with nobody present, this does not give you one.
“I want a scheduled report every Monday.” GTM Bud does not send one. The numbers refresh when someone asks, and they can be up to about 15 minutes old. If a refresh is still running, the answer says so, and the as-of time tells you the age of what you see. The Monday review is a saved prompt and a person who runs it.
“Per campaign is not enough. I need the account level.” The tool returns one campaign per call. An account view is a loop over list_campaigns and a sum of numerators and denominators, which the agent can do in one request. What you do not get is a history. Last week exists only if you saved last week.
Frequently asked questions about outbound results in the terminal
Can I get LinkedIn campaign results without opening a dashboard?
Yes. With the GTM Bud MCP server in a coding agent, you ask for the results of a campaign and the agent calls get_campaign_results. The answer comes back in the terminal, and the agent can write it to a file in your repo. The setup steps are on GTM Bud for coding agents.
How fresh are the numbers from get_campaign_results?
They carry an as-of time, which is when they were calculated. The numbers refresh when someone asks, and they can be up to about 15 minutes old. Read the as-of time before you compare two campaigns or two weeks.
Can I get one result set for all campaigns in one call?
No. Results are per campaign. For a view across campaigns, the agent calls list_campaigns and then get_campaign_results once for each campaign, and builds the table itself.
Can GTM Bud send the weekly numbers to me on a schedule?
No. GTM Bud does not send a scheduled report. The weekly review is a prompt that you run on a fixed day. Save the prompt in your repo so that the review is the same each week.
Can I see which leads are behind a number?
Partly. list_campaign_leads returns the leads of one campaign by lead status, and list_replies returns the replies with their labels. Use them to look at the people behind a count. They are separate calls from the results call.
Read your LinkedIn outbound results where you already work
To get LinkedIn outbound results in your terminal, add the server, save one review prompt, and keep one file per week with counts, definitions and the as-of time. Setup is one command, claude mcp add --transport http gtm-bud https://gtmbud.com/mcp, and the steps for Cursor are on GTM Bud for coding agents. When one campaign stands out, clone it for five segments from the same session. The page on AI LinkedIn outreach for GTM engineers shows the rest of what the agent can do with GTM Bud.