Stop Posting From Your Company Page. Build an AI-Run Employee Creator Program.
By Jamie Pagan, Director of Demand at Leadfeeder ·
Your LinkedIn company page reaches fewer people every quarter, and no amount of daily posting fixes it. The algorithm now rewards individuals, not logos. So Leadfeeder stopped fighting it. Instead of pouring effort into a branded page nobody sees, Jamie Pagan and his team built an internal Creator Program that turns 15 employees into the company's reach engine, with AI doing most of the heavy lifting behind the scenes. In this session, Jamie walked through the entire system, the results after two months, and what he would change. Here is how it works and what the numbers look like.
Why build a creator program instead of posting from the company page?
The starting problem is one every B2B marketer recognizes. Around 95% of the market is not in buying mode at any given time, and 80% of buyers go with the first vendor they seriously evaluate. That makes brand familiarity during the "not buying yet" window a real competitive advantage. The catch is that the old channel for building it, the company page, has quietly stopped working. Organic reach and engagement on company pages are the lowest they have ever been, because LinkedIn's algorithm now promotes the people behind a business rather than the business itself.
Jamie's framing was blunt: your website, pricing, and positioning can all be copied in weeks, especially now that AI accelerates everything. Reputation cannot. The way to build reputation at scale is to lean on the employees who already carry the company's expertise, and give them the tools to post consistently. Leadfeeder built the whole thing in-house and ran it as a pilot for roughly three months. The results below come from the first two full months of data.
What are the four pillars of the system?
The program is organized into four pillars, with AI woven through every one of them rather than treated as a separate feature.
| Pillar | What it does |
|---|---|
| Creator Hub | The central workspace holding templates, a knowledge base, tools, and the creator roster |
| One-to-one coaching | Kickoffs, bi-weekly syncs, and weekly walkthroughs run with each creator |
| Rewards and leaderboards | Points, skill levels, and a swag store to drive consistency |
| AI assist | Content generation, coaching notes, reporting, and analysis end to end |
Everything lives in Notion. Leadfeeder moved off Asana and Confluence at the turn of the year, and while the migration was painful at first, it let them build the custom workspaces the program depends on.
Who is in the program and how did it start?
The roster is around 15 people. At least half sit in marketing, since marketers are Leadfeeder's core ICP, with the rest spread across sales, R&D, people and culture, partnerships, and the CEO. The program was scoped in one quarter, built in the next, and launched in the one after.
Onboarding is deliberately slow. It started with a shortlist of people already active on LinkedIn or keen to be, plus the low-hanging fruit who needed the least enablement. One or two new creators are added per month, paced entirely by the program owner's coaching capacity since the one-to-one work is all done by hand. Recruitment is topped up by a call to action at each company all-hands: here is the latest data, reach out if you want in.
How does the Creator Hub help people actually post?
The Creator Hub is built like software onboarding, with a getting-started section, quick navigation, and several databases. Two parts do the most work.
Templates. An inline database where every entry follows the same structure: an overview, an example of a strong post, copy-and-paste frameworks for LinkedIn, tips, and then customization prompts for Claude, Gemini, and ChatGPT. Because the structure is fixed, the content is almost entirely AI-written once a human sets the post style in the header. The first 20 to 25 templates were built in about 60 minutes. New templates are seeded by studying high-performing posts, for example pulling 10 posts about a comparable product launch, running them through Claude, and asking it to derive five template styles.
Tools. These automate or accelerate content creation. Two stood out:
- Revenue Call Ghostwriter. Every 30 minutes it scans for the latest call transcript from a specified rep. When it finds one, it runs the transcript through topic clusters and keywords, drafts a LinkedIn post, and links it back to the original Salesforce account ID. People in sales, CS, and partnerships get two or three posts drafted for them each week this way.
- LinkedIn co-pilot. A companion built as a skill and a set of scheduled tasks inside Claude's Cowork environment. Each person's is custom.
How is the LinkedIn co-pilot set up for each person?
The co-pilot starts with a roughly 15-minute onboarding questionnaire in Claude. You feed it your LinkedIn history, a set of "peer voices" to learn from (people whose writing you admire, usually 10 to 12 of them), and any other material that captures how you sound. In Jamie's case that meant 20 episodes of a podcast he hosted, plus blogs. From there the co-pilot can analyze the last 12 months of your LinkedIn performance, draft posts and comments, and run a weekly review against your goals.
The weekly review is where the coaching becomes concrete. It surfaces your top and weakest post, checks your balance across four chosen topic pillars, and enforces personal rules. One of Jamie's rules: the first five words of any post must start from a personal point of view (I, me, we, our). It then gives about three recommendations for the week ahead, such as "break the demand strategy drought" or "cap the personal origin stories at one." The advice is genuinely good, though the honest guidance is to treat it as a menu and take the one or two tips that fit rather than following all of it.
On cost, this kind of co-pilot is not credit-hungry. Most of the load is front-loaded when you build the context, and after that it stays light. The one thing to avoid is a scheduled task that uses the LinkedIn Chrome extension to auto-reply to comments. That proved extremely intensive, running up around $200 in a single day of testing.
How does the one-to-one coaching work?
Each creator gets a kickoff, then a bi-weekly sync during onboarding that later drops to monthly or less. Between syncs, a weekly walkthrough is auto-published every Monday. It bundles a reminder, relevant templates, and outputs from the creator's own tools, then poses roughly eight questions that produce three posts for the week.
The coaching is as much about confidence and ideas as tactics. A common blocker is "I don't have time" or "I have nothing to say," and the fix is usually reframing what already happened. One creator mentioned being slammed with 18 demos after an event and not having posted; that is two posts right there. Every month, each creator also receives a teardown built from AuthoredUp data run through Claude: two or three things that worked, two or three that did not, and what to try next month. An overarching teardown across all 80 posts feeds the tips shared in the creator channel.
What content actually performs?
The 80-post analysis produced a few clear, and slightly surprising, patterns.
| Lesson | Detail |
|---|---|
| Text-only posts win | The company's CPTO had her top three posts of the month, including one at 70,000 impressions, all text-based. Don't wait for a perfect photo. |
| Start personal | The first five words should begin with I, we, me, or our. "Product release" as an opener was killed on the spot; "we've built something" works. |
| Write what only you can | Regurgitated go-to-market advice that anyone could pull from Google or an LLM underperforms. First-person specifics win. A post about the speaker's own LinkedIn campaigns hit 50,000 impressions. |
| No pitches | The lowest-reach posts across all 80 were product pitches, some barely clearing 35 impressions. |
| Video hasn't landed yet | So far video, mostly team-week clips, has not performed. |
Does it actually work?
The reach gap between creators and the company page is the headline. In month two, the creator roster generated roughly 20 to 21 times the reach of the company page. May brought 200,000 impressions across creators; June was tracking toward 300,000, even as the company posted on its page every working day. Website traffic from LinkedIn rose 13%.
The quality of that traffic held up too. Around 30% of LinkedIn visitors fit the Tier 1 ICP, and about 70% scored an intent of 8 or higher, based on pricing-page visits, dwell time, and scroll behavior. Both figures beat other channels, including organic search.
How does paid amplify the organic program?
The creator program is purely organic and stands on its own, but Leadfeeder built a paid layer on top of it. The model is borrowed from Fibbler, the LinkedIn ads attribution tool. They spend around $7,500 a month, split 70% cold and 30% retargeting, with roughly 60 to 70% of the cold budget going to thought leadership ads that put spend behind creators' best organic posts. Those posts fuel the paid strategy, so about two-thirds of the cold audience runs on content that would not exist without active creators.
Tracked in Fibbler, the thought leadership ads had influenced over $12,500 in pipeline and about $5,000 in closed-won revenue in the early window. A lift analysis, though still early, showed deal sizes roughly 50% larger for cohorts exposed to the LinkedIn ads during their buying cycle.
On the brand side, since the rebrand the team has seen a 2x increase in Tier 1 ICP visits to pricing and demo pages, rising branded search year over year, and higher direct traffic. There was also an eye-catching 800% jump in AI-referred visits that converted between two months, though this is measured ad hoc rather than through a structured process.
How is all the reporting built?
The tracking database sits in Notion, built almost entirely by Claude over a three-to-four-hour reporting session. Notion was chosen over Google Sheets for a practical reason: the team can only read from their Google Workspace connection, not write to it, whereas Notion allows full read and write. Each month, AuthoredUp and LinkedIn analytics are uploaded, Claude digests and processes them, and the results feed a dashboard surfaced across the business: posts, impressions, engagements, leaderboards, and per-creator and per-month breakdowns. A bonus of staying in Notion is that the leaderboards embed cleanly back into the Creator Hub.
What would he change?
The rewards system was the biggest learning. V1 was based purely on impressions (20,000, 30,000, or 40,000 for bronze, silver, gold), which unfairly favored people who had been posting for years over newcomers. V2 introduced beginner, intermediate, and advanced skill levels. Core rewards are identical across levels (30, 60, and 90 points for bronze, silver, gold); only the impression thresholds differ. Everyone must hit a floor of 8 posts a month to earn anything, which enforces consistency, and there is a bonus for moving up a level. Points are spent in a swag store run through Monday Merch. A future V3 will likely blend impressions, engagement, and reach into a single score so a high impression count alone does not dominate.
The other lesson was simply consistency. Even after two months, the compounding effect of getting the whole roster posting at least twice a week is clear. The system's structure, repeatability, and SOPs are what make it scalable without heavy ongoing workload.
Key Takeaways for GTM Teams
- Shift reach from the page to the people. Company-page organic reach has collapsed. Employee creators drove roughly 20x the reach of the company page in month two.
- Let AI do the structure, not the judgment. Fixed templates and a personal co-pilot let AI draft most content, while humans set style, edit, and coach.
- Build a personal voice model. Feed a co-pilot your post history, 10 to 12 peer voices, and anything that captures how you sound, so drafts read like you and not like generic AI.
- Reward consistency, not just impressions. Impression-only incentives punish beginners. Skill levels plus a posting floor keep everyone motivated.
- Personal beats polished. Text-only, first-person posts outperformed images, video, and anything that reads like it came off Google.
- Layer paid on top of organic. Put spend behind proven organic posts as thought leadership ads, and attribute it to pipeline.
- Choose tools you can write to. Notion won as the reporting database partly because the team could read and write to it, unlike their Google Workspace connection.
Want to Attend the Next Session?
The GTM Clawllective is a community of go-to-market leaders sharing how they build with AI. Sessions are live and not recorded, so the room is where the real detail lives. Next up: Elaine Zelby, CMO at Tofu, on how she built proactive sales agents that surface hidden opportunities. Join the community to get the invite.
About the Speaker
Jamie Pagan is Director of Demand at Leadfeeder, where he leads demand generation and demand capture. Leadfeeder is a website visitor identification tool that reveals the companies visiting your site at the company level and tiers them by ICP fit, so marketing and sales can act on real intent signals.
At Leadfeeder, Jamie built the company's employee Creator Program from scratch, an AI-assisted system for turning subject-matter experts into a consistent LinkedIn presence, and pairs it with a Fibbler-tracked thought leadership ad strategy.
- Jamie Pagan on LinkedIn
- Leadfeeder
- Fibbler (LinkedIn ads attribution)
- AuthoredUp (LinkedIn analytics)