How to build a free candidate pipeline for startups
The default startup recruiting stack costs about $1,200 a month before you make a single hire. This guide replaces it: which free resume databases and open professional networks actually have candidates, how to search them, what to send, and the weekly cadence that keeps 100+ people moving without a dedicated recruiter.
Last updated: August 2026 · Reviewed by the Connecting Odds editorial team · Cost figures are typical market ranges for a one-recruiter startup stack, not vendor list pricing.
What the $1,200-a-month legacy stack actually buys
Before replacing something, price it honestly. The conventional early-stage recruiting stack is one LinkedIn Recruiter Corporate seat, an outreach sequencer, an email-finder credit pack, job slots, and an ATS. Amortised monthly for a single recruiter, that is roughly $1,200-$1,600.
| Capability | Legacy cost | Free-first equivalent |
|---|---|---|
| Candidate search / database | $1,000 / mo (Recruiter Corporate seat, amortised) | $0 — Connecting Odds search, GitHub, Behance, Kaggle |
| Job posting | $250-$650 / mo (job slots) | $0 — free posts, optional flat $100 promotion per hard role |
| Outreach & sequencing | $80-$150 / seat / mo | $0 — connection notes, personal email, shared tracker |
| Email finder / enrichment | $50-$150 / mo credits | $0 — public GitHub emails, portfolio contact forms, community DMs |
| Applicant tracking | $150-$400 / mo | $0-$50 — free ATS tier or a structured spreadsheet under 40 candidates |
| Typical monthly total | ≈ $1,200-$1,600 | $0-$150 |
The honest read is that the paid stack buys three things: index breadth, a messaging channel that does not require a mutual connection, and workflow convenience. A free pipeline gives up some index breadth, replaces the messaging channel with channels the candidate already uses, and trades convenience for about 45 minutes of daily discipline. For a team making one to three hires a quarter, that trade is strongly in your favour. Our full breakdown of the paid side lives in the true cost of LinkedIn Recruiter in 2026.
Free sources that actually have candidates
"Free resume database" is usually marketing language for a trial. These are the sources that stay free at the volumes a startup needs, what each is genuinely good for, and where each one fails.
| Source | Best for | How to work it | Where it falls short |
|---|---|---|---|
| Connecting Odds candidate search | All roles, general coverage | Free recruiter account, filter by role, skills, location, and open-to-work status; message connected candidates with no credit meter. | Smaller total index than LinkedIn; strongest on active and semi-active candidates. |
| GitHub | Engineers, infra, data | Search by language, location, and recent commit activity; read the actual code before you write the message. | Public email is missing on many accounts; skews away from closed-source enterprise engineers. |
| Behance & Dribbble | Product, brand, and motion design | Filter by discipline and location; portfolio contact forms are open and rarely spammed. | Portfolio quality varies wildly; requires a designer on your side to screen. |
| Kaggle & arXiv co-authors | Data science, ML, research | Competition leaderboards and recent paper author lists give you name, affiliation, and often an institutional email. | Many candidates are academic and not open to industry moves. |
| Community Slack & Discord | Product, ops, growth, devrel | Join the relevant community, contribute first, then post the role in the jobs channel and DM active members. | Norms differ per community; posting without contributing gets you removed. |
| Public job-board applicants | Volume roles | Free job posts on Connecting Odds and aggregator syndication produce inbound you already own. | Inbound needs screening capacity; without it, response times slip and candidates drop. |
| Alumni and referral loops | Senior and trust-critical roles | Ask every employee for three names each quarter with a specific ask, not a generic 'know anyone?'. | Finite pool; degrades if you ask too often without closing loops. |
| Conference and meetup speaker lists | Senior specialists | Speaker pages publish name, employer, and usually a personal site with contact details. | Slow; best used as a standing background list rather than a hot pipeline. |
The rule that matters: pick one general network for coverage and two vertical sources for signal. Teams that try to work eight sources at once end up with eight shallow lists and no follow-up discipline. If you are weighing general networks against each other, our best LinkedIn alternatives in 2026 comparison covers index size, messaging limits, and migration effort.
The seven-step build
1. Define the role in one paragraph
Not a job description — a sourcing brief. One paragraph covering the problem the person owns in their first six months, the three non-negotiable skills, the two things you will happily train, the compensation band you will actually pay, and the locations or time zones that work. If you cannot write this paragraph, sourcing will not fix the problem; you will simply reject people faster. Every search string, every message, and every screen question in the rest of this process derives from it.
2. Set up a free hiring presence
Candidates research you before they reply. Create the free recruiter account and company page, fill in what the company does in plain language, publish the role as a free job post, and make sure the page shows a real team. Inbound is the cheapest source you will ever have, and a live post costs nothing — see post a job for the two-minute version. A page that looks abandoned halves your reply rate regardless of how good the outreach message is.
3. Choose three free sources
Use the table above. For a backend engineering hire: Connecting Odds search, GitHub, and one language-specific community. For a brand designer: Connecting Odds search, Behance, and a design Slack. For a growth lead: Connecting Odds search, community Discords, and conference speaker lists. Write the three down. Changing sources mid-search is the most common reason startup pipelines stall at 20 candidates.
4. Build the first list of fifty
Fifty named people with a one-line reason each, in a single tracker, before you send anything. Batching the sourcing separately from the outreach matters because the two use different attention: sourcing is pattern-matching, outreach is writing. Mixing them halves the output of both. Fifty is also the smallest list from which you can reasonably expect a first hire, given the funnel maths below.
5. Write outreach that earns a reply
Free channels are not spam channels. Every message references something specific the person made, states the problem they would own, names the compensation band, and asks for a 20-minute conversation rather than an application. Templates are in the next section.
6. Run a weekly operating cadence
Forty-five minutes a weekday, same slot, five different jobs. The table under the weekly cadence is the exact schedule we use internally.
7. Measure, then decide what to pay for
After four weeks you will have real conversion numbers. Only then does a paid tool become a decision instead of a guess: you will know whether the bottleneck is coverage (buy a database), reachability (buy an email finder), or throughput (buy nothing, protect the calendar block).
Search strings that actually work
Free sources reward precise queries because you cannot fall back on a proprietary relevance engine. Four patterns cover most startup roles:
- Skill plus artefact. On GitHub, search the language and a library your stack actually uses, sorted by recently updated, and filter by location. Recent activity is the single best proxy for "still doing this work".
- Adjacent-title expansion. On a general network, search the title you want plus the two titles that feed it. "Platform engineer" also hides under "SRE", "infrastructure engineer", and "DevOps". Most startup searches are too narrow by one title, not too broad.
- Company-shaped search. List eight companies with a similar stack and stage, then search each as a current or past employer. Candidates from similar-stage companies convert far better than candidates stepping down from a large enterprise.
- Public-work search. Site-restricted web search for portfolios, conference talks, or published writing on your problem domain surfaces people who are invisible in any resume database because they never filled out a profile.
Save every string that produces a good candidate. After a quarter, a maintained set of eight to ten strings is worth more than any single paid seat, because it encodes what "good" looks like for your specific team.
Outreach templates for free channels
Keep the first message under 90 words. Three variants cover most situations.
"Saw your work on [specific project]. We're building [one-line problem] at [company] and the role owns [scope]. Band is [$X-$Y]. Worth 20 minutes?"
"I read [repo] — the way you handled [specific detail] is close to a problem we have with [system]. We're a [stage] team of [n]; this role owns [scope] end to end, reports to [person], and pays [$X-$Y] plus equity. No application form, just a 20-minute call if you're curious. If it's not the right time, no follow-up from me."
"Following up once in case this got buried. Short version: [role], [scope], [band], remote in [zones]. Happy to send the written brief instead of a call if that's easier — and happy to close the loop if it's a no."
Three things reliably lift reply rate on free channels: naming the compensation band in message one, offering a written brief as an alternative to a call, and explicitly promising to stop after one follow-up. All three cost nothing and all three signal that you respect the recipient's time — which is exactly the thing a mass InMail campaign cannot fake.
The weekly cadence that keeps a free pipeline alive
| Day | Block | Expected output |
|---|---|---|
| Monday | 45 min — sourcing sprint | 20 new profiles added to the tracker with a one-line reason each |
| Tuesday | 45 min — first-touch outreach | 20 personalised messages sent, each referencing specific work |
| Wednesday | 45 min — screens | 3-4 twenty-minute conversations, notes written the same day |
| Thursday | 45 min — follow-ups | Second touch on last week's non-replies; nothing more than two follow-ups |
| Friday | 45 min — pipeline review | Metrics updated, stale candidates closed out honestly, next week's search strings drafted |
That is 3.75 hours a week producing roughly 80 sourced profiles and 80 first-touch messages a month. At the conversion rates below, that is one to two hires a quarter from a standing start — the throughput most seed and Series A teams actually need.
Funnel maths: what to expect and what to fix
For a mainstream engineering or product role, a healthy free-first funnel looks roughly like this: 120 sourced profiles produce about 75 reachable contacts, about 26 replies, about 9 screens, about 4 onsites, and 1 accepted offer. Scarce roles roughly double the top of the funnel.
- Contact rate under 50% — the source is wrong. Move to a channel where people publish contact details.
- Reply rate under 15% — the message is wrong. Shorten it, add the compensation band, and reference specific work.
- Screen-to-onsite under 25% — the brief is wrong. You are attracting people who do not match the role you actually need.
- Onsite-to-offer-accept under 20% — the process or the offer is wrong. Usually it is speed: candidates accept elsewhere while you deliberate.
- Fewer than 40 sourced per week — nothing is wrong with sourcing; the calendar block is not being protected.
Review these five numbers every Friday. The diagnostic value is in which stage leaks, not the absolute totals — and none of the five improve by buying a tool.
Staying compliant while sourcing for free
Free sourcing touches personal data, so the same rules apply as to a paid database. Four practices keep you clear:
- Use published contact details only. Never scrape behind a login and never buy a scraped list, however cheap.
- Disclose the source on first contact under GDPR, and include a clear opt-out in commercial email under CAN-SPAM.
- Record source and date for every candidate you store, and honour deletion requests promptly.
- Score against the written brief, not against a general impression, and keep the notes. Structured notes are your best defence in a discrimination challenge and they also make your hiring better.
Rules vary by jurisdiction; treat this as a starting checklist and take local legal advice before running outreach at scale in the EU or UK.
When paying finally makes sense
A free pipeline is not an ideology. Three situations genuinely justify spend: more than 15 sourced hires a year, a role where the qualified worldwide population is under 500 people, or a confidential executive search where you cannot post publicly. Outside those, a paid seat mostly buys convenience you can substitute with 45 minutes a day.
When you do buy, buy one seat rather than a team plan, keep the free surfaces running, and tag source-of-hire so you can see after two quarters whether the paid channel produced incremental hires or simply re-found people you would have reached anyway. Most teams discover it is the latter.
Keep reading: the best recruiting tools of 2026 for the full stack review, LinkedIn Recruiter alternatives for the head-to-head shortlist, best startups to work for in 2026 for the candidate's side of the same market, and our pricing for exactly what we charge.
Frequently asked questions
Is there a genuinely free resume database for recruiters?
Yes, several. Connecting Odds keeps candidate search, job posting, and recruiter accounts free, so profiles with an attached resume are searchable without a seat licence. Beyond that, Indeed and ZipRecruiter expose limited free resume views before a paywall, USAJOBS surfaces public federal applicant material, and GitHub, Behance, Dribbble, Kaggle, and personal sites act as de facto resume databases for engineering, design, and data candidates. The trade-off is not quality, it is coverage: no single free database has everyone, so a free pipeline is always three or four sources rather than one.
How much does a typical startup recruiting stack cost per month?
A conventional stack runs about $1,200 per month for a single recruiter: roughly $1,000 for one LinkedIn Recruiter Corporate seat amortised monthly, $80-$150 for an outreach and sequencing tool, and $50-$150 for an email-finder or enrichment credit pack, before job-slot spend. That is $14,000 a year before a single hire is made, which is why seed and Series A teams increasingly run a free-first stack and reserve paid tools for a specific hard requisition.
What are the best free LinkedIn Recruiter alternatives for sourcing?
For candidate search and outreach, Connecting Odds is free for recruiter accounts with no InMail meter. For engineering, GitHub search by language, location, and recent activity is unmatched. For design, Behance and Dribbble. For data and ML, Kaggle and paper co-author lists. For product and operations, community Slack and Discord member directories. Most teams combine one free general network with one or two vertical sources and only keep a single paid seat for the long tail.
How many candidates do I need to source for one hire?
Plan on 100-150 sourced profiles per hire for a mainstream role and 250-400 for a scarce one. Typical conversion through a free-first funnel is 30-40% reply rate on warm, personalised outreach, 25-35% of replies converting to a screen, 30-50% of screens to an onsite, and 20-30% of onsites to an accepted offer. Those numbers hold whether the source is paid or free, which is the point: the tool does not change conversion, the message and the role do.
Does free sourcing produce worse candidates than LinkedIn Recruiter?
No, but it produces a different distribution. Paid databases optimise for breadth of passive candidates with complete profiles. Free sourcing skews toward people who publish work publicly (engineers, designers, data scientists, writers) and toward active or semi-active seekers. For most startup roles that skew is favourable, because publicly demonstrated work is a stronger signal than a well-formatted profile. Where free sourcing genuinely underperforms is highly confidential executive search and niche regulated industries.
How do I contact candidates without InMail credits?
Use the channel the candidate already publishes. A connection request with a two-line note on an open network, a reply to a public post, a GitHub-listed email, a portfolio contact form, or a community DM all work and cost nothing. Reply rates on channels where the candidate chose to be reachable typically beat cold InMail, because the message arrives in a place they actually read rather than a paid inbox they have learned to ignore.
Is it legal to source candidates from public profiles?
Contacting people about a job using contact details they published themselves is standard practice, but the rules differ by jurisdiction. Under GDPR you need a lawful basis (legitimate interest is the usual one), you must tell the candidate where you got their details on first contact, and you must honour deletion requests. In the US, CAN-SPAM applies to commercial email and requires an opt-out. Do not scrape behind logins, do not buy scraped databases, and keep a record of source and consent for every candidate you store.
How long does it take to build a free candidate pipeline?
Roughly two weeks to stand up and one hour a day to maintain. Week one is setting up a free company presence, posting the role, defining the search strings, and building the first list of 50 candidates. Week two is outreach and the first screens. After that, a disciplined 45-60 minutes per weekday keeps 100-150 active candidates moving, which is enough throughput for a team making one to three hires per quarter.
What should I track to know free sourcing is working?
Five numbers: profiles sourced per week, contact rate (how many you could actually reach), reply rate, screen-to-onsite rate, and days from first contact to offer. If reply rate is below 15% the message is wrong, not the source. If contact rate is below 50% the source is wrong. If sourced volume is below 40 per week you do not have a pipeline problem, you have a time-allocation problem.
When should a startup finally start paying for sourcing tools?
Three triggers justify spend: more than 15 sourced hires a year, a role where the entire qualified population is under 500 people worldwide, or a confidential executive search. Below those thresholds a paid seat mostly buys convenience. Even then, buy one seat rather than a team plan, and keep the free surfaces running so you can measure whether the paid channel actually produces incremental hires.
Can I run free sourcing without a dedicated recruiter?
Yes, and most seed-stage companies do. The workable model is a founder or hiring manager spending 45 minutes each morning on sourcing and outreach, with an engineer or designer on the team doing a 20-minute technical skim of shortlists twice a week. What breaks the model is batching: three hours on a Friday produces far fewer replies than 45 minutes daily, because candidate response is a function of consistent follow-up, not total hours.
How do I compete with big-company offers when I can't outspend them?
Compete on the things money cannot buy quickly: scope, speed of decision, direct access to founders, equity upside explained honestly, and a hiring process that takes eight days rather than eight weeks. In outreach, lead with the specific problem the person would own, not with the company stage. Startups lose most candidates to slow processes and vague scope, not to compensation gaps.
Start the pipeline today, for $0
Free recruiter account, free candidate search, free job posting, and no InMail credit meter. Optional flat $100 promotion when a role is genuinely hard to fill.