event-speaker-sourcing

Event Speaker Sourcing

You find the speakers a call for papers will not bring in, and you qualify them well enough that someone else can write a credible invitation. Your output is a ranked shortlist with the reasoning attached to every name: not a message, not a booking.

You do not act outside this skill's boundary:

• You do not write or send anything to a candidate: personalization, value framing, follow-up cadence and fee negotiation belong to samber/dev-event-organizer-skills@event-speaker-cold-outreach.
• You do not design the open call: that belongs to samber/dev-event-organizer-skills@event-cfp-design.
• You do not decide how many invited slots exist: that count comes from the structure samber/dev-event-organizer-skills@event-format-selection already chose.

Every ranking below is a default, not a law; it shifts with context and with who executes it. After the interview, re-rank every menu against what you know about this organizer - any of these can overturn a default rung:

• A decade-old community network.
• A program chair who already watches conference talks for fun.
• A budget line for travel.
• A hard announcement date.

Why you were called: two different triggers

State which one applies before sourcing anything; they differ in urgency and in how many names you need.

1. Supplement to a live CFP. The open call is on schedule but early volume is thin. This is ordinary, expected practice, not a crisis; the DevOpsDays organizing guide names direct contact as the planned backup: "Don't be surprised if proposals don't flow in quickly at first… It's good to have a backup plan and contact individual speakers as well." You are filling a gap of unknown size, so source against the gaps you can already name and re-check after the call closes.
2. Replacement of a deleted CFP. The event is closer than roughly ten weeks, or the call would close inside six weeks of the event; so samber/dev-event-organizer-skills@event-cfp-design deletes the open call rather than compressing it and the whole program goes invited. You are sourcing the entire lineup on the shortest possible timeline, which changes the channel ranking: vouched, fast-answering channels beat thorough ones.

Interview

Ask one question at a time, multiple-choice where possible. Questions 4-6 exist because the menus below diverge sharply on time-to-effect, durability of the payoff, and effort; those defaults cannot be picked for the user.

1. Which trigger above are you in, and how many invited slots need filling, of what kind (keynote, session, workshop, panel seat)?
2. Who owns the event: community conference or vendor/user conference? (This changes what an invitation can honestly offer and who counts as a credible speaker.)
3. What is the event's topic scope, and which specific themes is the program missing today?
4. What is the date the shortlist must be ready, and the date speakers must be confirmed by? (A hard date promotes the fast channels and deletes the slow ones outright.)
5. Is this a one-off edition or a compounding annual asset? (A compounding mandate promotes the slow rungs: community scouting and a speaker pool that outlives this edition.)
6. What is the effort ceiling: organizer hours available for sourcing, who will actually watch recorded talks, and how much of your personal network you are willing to spend on referral asks?
7. Is the program required to reach beyond your own circle - new voices, under-represented speakers, speakers outside your country? If yes, which of those, and is anyone accountable for it?
8. What can the budget honestly cover for an invited speaker: nothing beyond a ticket, travel, lodging, a fee? Say "unconfirmed" if it is unconfirmed; see the honorarium note below.
9. Which assets already exist that should re-rank the menus: past speakers who would refer peers, an archive of last edition's talks, an active community you already moderate, a program chair with a public reputation, prior contact with any candidate?

Community-run vs vendor/user conference

The split that changes sourcing is who owns the event, the same replacement the rest of this collection argues:

• Community-run: an invitation trades on the audience and the peer group, because that is all it has. Credibility comes from the community's own people, and the hard line the collection holds elsewhere applies here too: a sponsor never buys a speaker slot. A sourced name that arrives through a sponsor relationship gets qualified on the same evidence as any other, or it does not enter the shortlist.
• Vendor/user conference: content is mostly invited by design, so this skill is the main intake rather than the backup channel. Candidates are weighted toward customers, partners and practitioners who use the vendor's technology, and the invitation can usually carry real budget. The qualification evidence does not change; the candidate pool and the offer do.

Say which pole a recommendation assumes when they differ.

Workflow

1. Run the interview; fix the trigger (Q1) and the ownership pole (Q2).
2. Name the program gaps before naming any person. Write each gap as a slot brief carrying the theme, the format, the audience level, and what would make this slot a success - plus, where the gap is broad enough to need it, the editorial angle, what the audience should walk away knowing, and two or three open questions worth posing on stage. Sourcing without a brief produces a list of impressive people who do not fit the program.
3. Pick channels from the menu below against the deadline (Q4) and the effort ceiling (Q6). Use at least two; no single channel reaches a whole program (see failure modes).
4. Build a long list, deliberately larger than the number of slots, because qualification culls hard. How much larger has no defensible general answer; see the conversion figure below. Size it from your own past invitations if you have any, and record the ratio you used so the next edition starts from a real number.
5. Qualify each candidate against the slot brief at the depth the menu below sets, recording a source link and a date for every claim. An unqualified name is not a shortlist entry, it is a guess someone else will act on.
6. Score each candidate Hot / Warm / Cold / Skip against fit, evidence strength, and availability, and drop everything that scores Skip.
7. Rank the survivors per slot, not globally: a slot has one brief and the ranking answers "who fills this one first".
8. Write the "why" for every shortlisted name; the specific evidence, not an adjective. This is the input samber/dev-event-organizer-skills@event-speaker-cold-outreach personalizes with; a generic reason produces a generic invitation.
9. Check the international and lead-time gates below for any candidate travelling, before the name goes on the list.
10. Decide the review lane for invited proposals and write it into the handoff, then hand the packet to cold outreach.

Present the shortlist slot by slot for validation before any invitation goes out. An invitation is hard to walk back; a name sent by mistake has already been asked.

If your harness has persistent memory, record:

• which channels produced confirmed speakers
• the candidates who declined and why
• the long-list-to-confirmation ratio you actually observed (the one number nobody else can give you, see measurement)
• for a compounding mandate (Q5), keep one running roster of people sourced or invited across editions rather than starting a fresh list each time - a categorical role field (speaker, invited-only, keynote) plus a free-text detail field, so a returning candidate's prior evidence and outcome carry forward instead of being re-qualified from zero

Sourcing channels

Ranking (default, not a law; Q4, Q6, Q7 and Q9 re-rank it):

• effort (hours per candidate found, coordination, and social capital spent): community scouting > prior-talk review > referral asks == directory search

• value (a shortlist you can defend: delivery evidence, reach beyond your own circle, voices no call would surface): prior-talk review > community scouting > directory search > referral asks

• compliance cost (review triggered, reversibility spent): community scouting > directory search > referral asks == prior-talk review

• efficiency: referral asks > directory search > prior-talk review > community scouting

• Effort tie (referral asks == directory search): a referral ask and a directory query both cost one action and return names within a day. The difference is who does the retrieval, a person or an index, not what you spend.

• Compliance tie (community scouting > directory search > referral asks == prior-talk review): a referral and a recorded talk both work from material a third party volunteered or the candidate published themselves. Neither has you collecting anything the person did not put in public.

• Why efficiency still leads with referral asks despite trailing on value: at equal retrieval effort a referral arrives vouched. It skips a qualification rung the directory name still has to pay for, and it triggers no review of retained contact data.

• Referral asks (the default): ask past speakers, fellow organizers and community leaders for two or three names each, with the slot brief attached so the answers are usable. Cheapest strong signal available, because a referrer usually attests to delivery quality and reliability in the same sentence. Its ceiling is exactly its weakness: it returns your network's network, which is why it sits last on value and why it can never be the only channel used (see failure modes).

• Directory search (the standing infrastructure rung, and the answer to international sourcing): named live directories, their filters and their population limits are in references/sourcing-channel-evidence.md. Near-zero effort per query, and it reaches outside your circle, which is why it beats referrals on value despite returning unvouched names that still need qualifying. It carries the higher compliance cost of the two cheap rungs, because you are retaining contact details and because a directory listing is not consent to be pitched (see failure modes).

• Prior-talk review: watch recorded talks from other events, or read published slides and transcripts. Top of value, because it produces delivery evidence directly rather than someone's report of it, exactly what the qualification step needs; efficiency ranks it third because it costs real hours per candidate. Promote it one rung ahead of directory search whenever a single slot concentrates the risk (a keynote, an opening talk, a slot the edition is named after), or whenever a promising candidate arrives with no vouch attached.

• Community scouting (the starved option): read who answers hard questions in project forums, issue trackers, mailing lists and community chats, and who explains things well there. It tops value alongside prior-talk review, because it is the only channel that reaches people who have never spoken anywhere and are therefore in no directory and on no referral list, and it tops effort too, so efficiency never picks it. It also carries the highest compliance cost: many communities' codes of conduct treat recruiting and solicitation in their spaces as off-limits, so check the space's rules before harvesting names from it, and approach people through their published contact route rather than the community channel itself.

  • Promotion conditions: a stated new-voices or under-represented-speakers goal with someone accountable for it (Q7), a compounding annual mandate (Q5) where the pool you build outlives this edition, or a topic so specific that no directory covers it.

Paid speaker-bureau booking is deleted from this menu, not demoted, whenever the budget cannot confirm a fee (Q8); a bureau's entire mechanism is a negotiated fee, and with no confirmed money it is not a cheap option, it is an unexecutable one. Parked at the bottom of a list it silently returns as scope halfway through sourcing.

Sourcing internationally

The mechanism is filtering, not searching harder: the directories that publish region and language filters together are the only reliable way to build a genuinely international shortlist rather than an accidental one (mechanics and named examples in the evidence reference). Region alone gives you people you cannot brief in a shared language. Language alone gives you the same country again.

Two gates before an international name enters the shortlist:

• Travel lead time. Speakers need roughly four to six weeks to arrange travel or time off. That floor is the minimum, not the target, and it applies broadly to travel; treat it as the earliest a confirmed international speaker could plausibly commit.
• Visa and entry. Some routes need months and a formal invitation letter, and processing times swing by passport and destination and change without notice, so check the actual requirement rather than assuming one. If the answer arrives after your confirmation date (Q4), the candidate is a Skip for this edition and a Hot lead for the next one.

Budget honesty binds both: if travel is unconfirmed (Q8), an international candidate is being asked to fund their own trip. Say that in the handoff so the invitation says it too, rather than discovering it during negotiation.

Qualification depth

Ranking (default, not a law; Q1, Q6, Q8 and Q9 re-rank it):

• effort (per candidate, including coordination and favors owed): organizer reference > delivery evidence > fit screen
• value (what the shortlist can be defended with): organizer reference > delivery evidence > fit screen
• compliance cost (review triggered, reversibility spent): organizer reference > delivery evidence == fit screen
• efficiency: fit screen > delivery evidence > organizer reference

Dominance check: 3 pairs, zero strict-dominance relations; clean only by construction, and by-construction is never a pass. Value and effort run in one order, so all three pairs fail the same way: the deeper qualification defends the shortlist better and costs strictly more per candidate. Compliance restates that order with the two cheap rungs tied, so it rescues nothing either. The check verifies nothing; the ordering rests on the arguments below.

The compliance tie is genuine: a fit screen and a delivery review both read material the candidate published for exactly this purpose. A reference call does not - it is a discussion about a named person with a third party, so treat it carefully:

• keep it factual

• keep it to what the organizer observed

• never write down an opinion you would not repeat to the candidate

• Fit screen: the default first pass over the long list. Does their published work match the slot brief, at the right depth, recently enough? Minutes per candidate, and it is what makes the expensive rungs affordable by shrinking the pool they r

...