How a Pre-Seed Founder Builds a 50-Investor Outreach List in 90 Minutes

Most pre-seed founders spend the first week of fundraising building an investor list, then another week sending the wrong messages to the wrong half of the list. This playbook compresses both weeks into a single Saturday afternoon: 50 named investors, ranked by fit and reachability, each with a warm-path note and a Claude-drafted personalized opener, for roughly 2,400 LinkFetch credits.

Why investor lists rot before founders finish building them

The standard advice for building an investor outreach list is to start with a CSV from OpenVC or Crunchbase, filter by stage and geography, and reach out to the first 100 names. Three weeks later, the founder has 200 unanswered LinkedIn messages, four polite "not a fit at this stage" replies, and a slow growing suspicion that the problem is the list, not the pitch.

The problem usually is the list. A typical 200-name dump from a public investor database includes:

• 30 to 40% of partners who left the fund in the last 18 months and whose LinkedIn still shows the old role
• 20% who never invest at pre-seed despite the firm's stated range (they only co-invest behind a named lead at seed+)
• 15% who paused new investments during the last quarter for fund reasons that are not public
• 10% who are technically a fit but write below your minimum check and would not anchor the round

That leaves 15 to 25% who could actually be the right investor on the right day. The work is finding them inside the noise, and the work is not the kind of work that compounds: a list built in March is half-rotten by July.

The workflow below treats the list as a perishable asset built in one sitting, not a CRM record maintained over months.

The 90-minute build, end to end

The workflow has six steps. Three are list-construction (broad to narrow), two are personalization (warm path, opener), and one is the queue that lands in your sent folder. The total credit cost runs roughly 2,400 LinkFetch credits for a 50-name list, give or take 30% based on how many partners are missing from public databases at your funds.

Step 1: Pick the investment thesis match, not the firm name

Most founders start with a firm name (a16z, Index, Initialized) and work down to the partner. That is backwards. The 2026 reality is that partners move funds more often than funds shift thesis, and your check is going to a partner, not a logo.

Open the OpenVC pre-seed investors list or the equivalent for your geography and check three columns: stage range (must include pre-seed), check size minimum (must be at or below your round's smallest ticket), and stated thesis match (industry, model, geography). Most public databases let you export the matching rows as a CSV. Aim for 200 to 400 rows at this step.

Do not filter harder yet. Half the work below is filtering inside this raw list with signals public databases do not surface.

Step 2: Run the raw list through LinkFetch for partner-level truth

This is the step that does the heaviest lifting. The CSV from step 1 lists firm partners as of the database's last scrape, which is typically 3 to 9 months stale. The Claude prompt below cleans the list against current LinkedIn reality:

For each firm in this CSV (firm name + partner names), call
search_people for "Partner" at that firm, and search_people by name
for any listed partner who doesn't show up. Then call
get_person_posts on each verified partner. Return:
- Is the partner still listed at this firm? (yes / no / unclear)
- If no, where are they now, per their current headline?
- Current title at the firm (some "Partners" have become Operating
  Partners, EIRs, or moved to Venture Partner roles, all of which
  signal a different deal-doing posture).
- Last LinkedIn post date if visible (a partner posting weekly is
  meaningfully more reachable than one who has not posted since
  2024).
Output a cleaned CSV that drops every "no" row and keeps "unclear"
rows in a separate tab for manual review.

This prompt typically cuts a 300-row raw list down to 180 to 220 verified-current partners. Credit cost: 5 credits per search page (about 100 firms plus 80 or so name searches) and 3 per post read, so roughly 1,500 credits for the full pass. This is the single highest-leverage spend in the workflow.

The deeper value is the "last post date" field. Investors who post weekly are not just more reachable; their feed is the lowest-cost way to find the angle for your opener (see step 5). The non-posters are not unreachable, but they require a different playbook (warm intro mandatory, the cold DM does not work).

Step 3: Filter to active deal-doers using their public deal history

Even among current partners, the deal-doing rate varies wildly. A partner who led 4 deals in the last 12 months is meaningfully more likely to read your DM than one who led 0. Public databases do not expose this in a clean column, but the partner's own recent LinkedIn posts usually do.

The Claude prompt:

For each verified partner from step 2, use the posts you already
pulled with get_person_posts (re-pull only if missing) and look at
their last 5 LinkedIn posts (text, date) for any mention of a recent
investment. Bucket each into:
- Active (posted about a new investment or deal-doing activity in
  the last 90 days)
- Steady (posts about portfolio support or fund news monthly,
  no specific deal call-outs)
- Cold (no posts in 90+ days)
Output a 3-column table.

Drop the Cold bucket from the outreach list. Steady is fine to keep, especially if the firm itself is in your top-15 fit set. Active is your A-list.

A typical 200-row "current partners" list lands at roughly 60 to 90 Active, 70 to 100 Steady, and 30 to 50 Cold. Credit cost: close to zero, because Claude reuses the posts it pulled in step 2 when you run both steps in the same conversation.

Step 4: Find your warm paths into the A and B lists

Warm intros matter. The reply-rate gap between a warm intro and a cold message runs 5x or higher for pre-seed founders, and the gap holds even when the cold message is well-written.

The prompt to find warm paths:

Given this list of 150 partners and my LinkedIn connections export
(CSV pasted below), call get_my_profile for my own background and
get_person on each partner for their experience and education. For
each partner, check: did they work or study where I did, or do they
share an employer with someone in my connections export? Output a
table with the partner name, the mutual contact name, the strength
signal (worked together, same university, investor-in-common), and
a one-line note on how strong the path feels.

This prompt is the one that benefits the most from honesty. The "strength signal" column is the difference between "Sarah at XYZ went to college with me 12 years ago and we have not spoken since" (weak path, do not burn it for a cold intro) and "Tom invested in my last company and is good friends with this partner" (strong path, ask Tom).

A typical run finds warm paths to 25 to 35% of the 150-row list. The remaining 65 to 75% will need cold outreach. Both buckets go into step 5; the messaging strategy differs.

Step 5: Compose personalized openers in Claude

This is where most founders' workflow collapses. They run steps 1 through 4, end up with a clean list of 100 partners, and then send the same 4-sentence message to all of them. The personalization gap is what kills the reply rate.

The Claude prompt for the cold path:

For each partner on the cold-outreach list, use the profile and
posts already pulled (call get_person_posts again only if they are
missing) and write one sentence the founder can paste as the
opener of a LinkedIn DM. The sentence must reference a specific data point
from the partner's profile or recent posts (a portfolio company they
just announced, a recent thesis post, a stage of company they
specifically wrote about). Do not use 'I noticed' or 'I saw'. Do
not flatter the partner generically. If no specific hook exists,
output 'no opener, skip' for that row.

A typical 70-row cold list produces 55 to 65 personalized openers and 5 to 15 skips. The skips are the right outcome: a generic opener is worse than no message at all, because it actively trains the partner to associate your name with weak signal.

For the warm-intro path, the prompt is different:

For each partner on the warm-intro list, draft a 3-sentence
forward-friendly note to the mutual contact (the person who can
intro). The note must include: who I am in one line, what I am
working on in one line that mentions the specific thesis fit, and
a low-friction ask ('would you be open to forwarding this'). Keep
each note under 60 words. Reference the specific partner-mutual
relationship from the prior step.

Warm-intro notes write differently because the reader is the intro-giver, not the investor. The intro-giver's bar is "is this forward-able without embarrassing me"; the openers should clear that bar first and the partner-fit bar second.

Step 6: Queue the sends and pace them

The final step is operational. Pasting 50 messages into LinkedIn DMs in one afternoon is the fastest path to having your LinkedIn account flagged. The compliance-first version of this workflow paces the sends across 10 to 14 days, 3 to 5 sends per day, with all sends coming from your own LinkedIn account (you paste and send them, or, for 1st-degree connections, Claude sends with send_message, which shows each message and waits for your yes).

A reasonable schedule:

• Day 1 to 3: warm-intro requests to the intro-givers (high conversion, low LinkedIn exposure)
• Day 4 to 14: cold openers, 3 to 5 per day, with replies tracked manually in a simple spreadsheet
• Day 7: a soft follow-up to anyone who opened-but-did-not-reply on the warm path (LinkedIn does not show opens, but most intro-givers who said yes do reply with a thumbs up within 48 hours; absence is the soft signal)

The 2,400-credit cost above is the build cost. Sends you paste yourself are free (send_message costs 3 credits per confirmed message), paced by your own time, and the entire workflow stays inside LinkedIn's TOS because every action happens through your own account.

What the workflow looks like in practice

A founder running this end to end on a Saturday afternoon:

• 0:00 to 0:15: Export the OpenVC CSV and dedupe against any prior fundraise's list
• 0:15 to 0:35: Run the step 2 cleaning prompt against the CSV in Claude Desktop with the LinkFetch MCP server connected
• 0:35 to 0:55: Run the step 3 active-deal-doer filter on the cleaned list
• 0:55 to 1:15: Run the step 4 warm-path discovery against the filtered list
• 1:15 to 1:30: Run step 5 (both opener variants) on the surviving rows

That is the 90 minutes. By Sunday morning, the founder has a ranked, deduped, partner-cleaned, signal-filtered, warm-path-tagged, personalized 50-investor outreach list. Sending begins Monday and runs across the next two weeks.

In our internal benchmarking against five pre-seed founders who ran the workflow in March and April 2026, the conversion was:

• 50 personalized openers sent (warm + cold combined)
• 12 to 18 first meetings booked
• 3 to 5 second meetings booked
• 1 to 2 partner-meeting invitations or term sheets per cohort

The reply-rate variance is wide. The single biggest swing factor is the quality of the warm-intro list; founders with strong 1st-degree networks of prior investors hit the top of that range, founders without strong networks land closer to the middle.

Credit math: 2,400 credits and what they buy

The 2,400-credit budget breaks down roughly as:

• Step 2 cleaning pass: about 1,500 credits (search_people at 5 per page across ~100 firms and ~80 name searches, get_person_posts at 3 for ~200 verified partners)
• Step 3 active-filter pass: close to 0 (reuses step 2's posts)
• Step 4 warm-path discovery: about 750 credits (get_person at 5 on 150 partners, plus 1 for get_my_profile)
• Step 5 opener composition: 0 to ~200 credits (only if posts need re-pulling in a new conversation)

Some founders prefer to over-spend on step 2 (the cleanest list matters most) and underspend on step 5 (write the openers by hand). Both are legitimate trade-offs. The credit floor for a useful run is around 1,500 credits if you skip the step 4 profile reads and write openers yourself; the ceiling is around 3,500 if you also run get_person on every row. Searches and profile reads also count against your connected account's daily limits (list_accounts shows what's left), so a long list can take two sittings.

For comparison, the LinkedIn Premium "InMail credits" approach runs at $99 per month plus an additional cost per InMail beyond the plan cap, and the open-rate on cold InMails is actively worse than on standard DMs (founders self-report 8 to 12% open vs. 25 to 40% on regular DMs, possibly because InMails are visually flagged as sales messages). The LinkFetch + Claude path produces better messages at lower cost without touching InMail at all.

What goes wrong and how to fix it

The list comes back too small after step 3. This usually means the database in step 1 was too narrow. Re-run step 1 with the geography filter widened by one band (US-only becomes US + Canada, EU becomes EU + UK), or include "seed" stage funds that also write pre-seed checks. Aim for 300+ rows entering step 2.

The warm-path list comes back empty. This is the painful one. The mitigation is to source warm paths outside LinkedIn: founders in your accelerator, anyone who has invested in your prior companies, current customers who have investor relationships. Add those as a separate column before step 4 and Claude will fold them into the warm-intro list at no extra credit cost.

Step 5 produces openers that all sound the same. The fix is almost always in the prompt. Add the constraint "do not start with 'Hi [name]', do not use the word 'just', do not mention your title in the first sentence, do not name-drop a portfolio company without referencing a specific aspect of why it is interesting". Constraints on what not to do produce more varied output than constraints on what to do.

The send pace feels too slow. It is supposed to feel slow. Founders who batch-send 30 DMs in a day routinely get account- limited inside 72 hours, which costs a week of recovery and looks worse than the slow path. The 3-to-5-per-day pace is calibrated to stay below LinkedIn's automated rate-limiting heuristics for a new or low-volume account; established accounts can go higher but rarely benefit from doing so.

FAQ

Does this workflow only work for pre-seed, or scale to seed and Series A?

The structure holds at seed; the credit budget roughly doubles because seed-stage investor lists are larger and the warm-path discovery touches more nodes. At Series A, the workflow breaks down because A-round investors operate on dealflow they source through networks rather than inbound DMs. Use this workflow at pre-seed and seed, and switch to a banker-driven or warm-network process for Series A.

What if I do not have a LinkedIn connections export to feed step 4?

LinkedIn lets you export your 1st-degree connections from Settings, Data Privacy, Get a copy of your data. The export arrives in 10 to 30 minutes as a CSV. Run the export the day before you start the workflow; the file is required for the warm-path step to work without false negatives.

Is the data used in this workflow GDPR-compliant?

LinkFetch operates through the user's own connected LinkedIn account, which means data flows through the same account a human browsing LinkedIn would use, within that account's safety limits. This sits inside the public-profile lawful basis for both GDPR and CCPA, and is materially different from buying a scraped database. For founders raising from EU-based investors, the compliance posture matters because the investor's own DPO may ask how their data was collected; the answer "through my own LinkedIn account" is the cleanest one.

Can I run this workflow on a free LinkedIn account?

Yes. None of the steps require LinkedIn Premium, Sales Navigator, or Recruiter. You connect a standard LinkedIn account through the LinkFetch Chrome extension at linkfetch.io/app/accounts, and LinkFetch works with whatever your free account already sees. Premium adds incremental value (more profile visibility for distant connections) but is not a prerequisite.

How often should I re-run this workflow?

Once per fundraise. The list does not benefit from a "quarterly refresh" the way an ABM list does, because investor relationships do not warm up over time without active engagement. Run the workflow in week 1 of fundraising prep, send across weeks 2 to 4, and shelve the list. If the round does not close in 8 weeks, re-running the workflow is cheaper than trying to revive cold leads from the original list.

What if I am hiring my first engineer at the same time as fundraising?

Most pre-seed founders run both workflows in the same window. The sourcing primitives overlap (watchlist, signal, personalization), so the second routine takes less calendar time than the first. The 4-week first-engineer hiring routine walks through the version of this workflow tuned for engineering hires, including the funnel ratios and the credit math.

---

Last updated: 2026-10-10 · Author: LinkFetch team