Confidential Successor Sourcing: Leak-Risk Map to Graded Slate

This guide is for an in-house recruiter, sourcer, or founder who has to build a slate to replace a leader who is still in the seat and does not know about it. It follows one confidential replacement search, for a sitting Chief Revenue Officer, from the first list of people who know to a ranked slate. Along the way it shows the queries, the counts Refolk's index returned, and the two places the search went wrong. By the end you can map the incumbent's network, sort the market into rings by leak risk, set a contact order with a blind spec for each ring, and hand over a slate where every name has an exposure grade and a reason.

The worked example: replacing a sitting CRO

The example search replaces a Chief Revenue Officer at a venture-backed B2B software company that hires across the US and UK. The incumbent does not know, and the board wants a slate before it tells them. I have removed identifying details from the client side. The market counts are real counts from Refolk's index.

Terms used throughout:

• Incumbent: the leader currently in the seat who is being replaced.
• Knower: anyone who has been told the search exists. The knower list is the written record of who they are.
• Leak-risk map: a sourced list of every tie between the incumbent and other people or companies.
• Ring: a leak-risk tier (outer, middle, or inner) that decides when, or whether, you approach someone.
• Blind spec: a job description that leaves out the company name.
• Exposure grade: a label on each slate candidate stating how close they sit to the incumbent's network and why.

The leak this process guards against usually starts inside the client, not at the search firm. A search firm in Oregon puts it this way: "most confidential searches are exposed from the inside, not from the search firm". It offers no data for that. Still, it matches how these searches are built. Most of the risk sits in who you look at, who you message, and what the message gives away.

How long a confidential search can stay quiet

Assume the search will run longer than it can safely stay secret. Typical retained timelines already sit at or past the 90-day mark that Boston Search Group flags as a leak factor. The table sets published timelines against that mark.

The 99 to 144 row is my own derivation: KORE1's 90 to 120 days multiplied by the 10 to 20% premium one firm claims for confidential searches. That firm gives no methodology. For planning, Talentfoot uses 3.5 to 4 months for a VP and about 4 months for a C-suite search. KORE1 describes three to five assessed finalists over 90 to 120 days. Excelon reports a qualified slate of 5 candidates delivered in 3 to 5 weeks, across 88 searches in a year. A short timeline is possible, but plan around the longer figures.

The worked search therefore had a simple constraint: speed controls more risk than NDAs do. Every step below has a day budget, and the leak-risk map is rebuilt if the search passes 90 days.

What the incumbent's network sees when you view a profile

What the incumbent's network sees depends on your LinkedIn viewing mode. Semi-private mode is the riskiest choice for an in-house team, because it shows your employer's name. Only full private mode reduces you to "LinkedIn Member".

LinkedIn's own example for semi-private mode is "Someone at Accenture". In an agency search that only reveals the agency. In an in-house search, it shows the incumbent's own employer looking at a rival CRO, or at a former direct report of the incumbent. Free accounts see the five most recent viewers in the last 90 days. That is a short list, so a single named view stands out on it.

In an in-house search, your employer's name is the secret, and semi-private mode prints it.

Two points are unsettled, and you should check both locally rather than trust a blog post:

• Recruiter seats. One source says Recruiter users browse anonymously by default. Another reports an "Anonymous Recruiter" label that prospects recognize. A third says LinkedIn has not published a full breakdown. To check, have a knower view a test profile from each seat type you use and report exactly what the owner sees.
• Retroactivity. A third-party source says privacy settings are not retroactive, so earlier named views keep showing your details. LinkedIn has not confirmed this. Treat it as true.

The first fork in this search. The procedure I started from put viewing hygiene after the leak-risk map. But building the map means looking at the incumbent's former colleagues and direct reports, which are the profiles most likely to notice a visit. I moved viewing hygiene ahead of the map, and the procedure below reflects that order.

Building the leak-risk map of the incumbent

A leak-risk map lists every tie between the incumbent and another person or company, and each tie has a named source. If a tie has no source, it is a guess, and you mark it as one. The map covers five tie types: former employers with dates, direct reports past and present, board seats, investors, and alumni or other personal cohorts.

For US public companies, Item 401 of Regulation S-K requires disclosure of all public company directorships held by directors and nominees during the past five years, plus five years of business experience. SEC rules adopted December 16, 2009 extended that lookback to five years. A proxy statement's "Former Directorships" table, like the one in AMD's 2025 DEF 14A, is the cleanest board evidence you will find. It is also incomplete. There is no SEC requirement to disclose private or nonprofit board seats, and in venture-backed companies the private board is where founders and revenue leaders tend to overlap. In a startup search, a proxy filing showing no shared board tells you almost nothing.

No one has published how reliably any of these ties predicts that two people actually talk. So the map does not score ties by likelihood. It records two judgements for each tie: how well documented it is, and how close it puts the person to the incumbent.

For the worked search, the map was built before any candidate list. It covered every employer on the incumbent's record with tenure dates and locations, the incumbent's current direct reports, public board seats from proxy filings, the client's own investors, and the incumbent's schools. Each line had a source column. Several lines were marked "inferred" because no document backed them. Those went in the top-left quadrant and were treated as inner ring until checked.

The time estimate is 2 to 4 days. That figure is unverified, and no practitioner source documents the mechanics of this step. Track how long it actually takes you.

Sorting the market into outer, middle, and inner rings

A ring sort tags every source company and candidate by how close they sit to the incumbent's network. The tag sets both the contact order and how much the message can reveal. The outer ring has no documented tie and is approached first. The middle ring has distant or non-overlapping ties and is approached once outer-ring NDAs are back. The inner ring has documented close ties and is approached last or not at all.

The market queries and their counts

The first query was plain: Chief Revenue Officers across the United States and United Kingdom. It returned 6,244 people in Refolk's index. Before sorting anything, I sampled 25 profiles to find out what kind of market this was.

In the sample, 11 profiles were in US regions and 3 in UK regions. The ten top employers each appeared once. They included Paxos, Newsela, Afresh, Hatch, and Suralink. The New York figure combines the metro, the city, and Brooklyn. With 25 profiles, all of this is indicative only.

What this changed in the ring sort: at least in this sample, the CRO market is fragmented by company and clustered by city. Shared employers will rarely be what puts a candidate in the inner ring, because few candidates share an employer with anyone. Shared metros and personal ties are the likelier channels, and Boston Search Group lists recruiting in the incumbent's own location as a leak factor too. In the worked search, candidates in the incumbent's own metro started one ring further in than their employer history alone would put them.

Wrong turn: the empty industry filter

The next query added the industry filter "Computer Software" so the market would be limited to software companies. It returned 0 for both the US and the UK. That is a label mismatch between the filter value and how profiles are tagged, not an empty market. A sourcer working quickly could easily report "no software CROs available" to a board at this point. The fix is to rerun without the filter and narrow by description or company list instead.

The US-only versus UK-only comparison I planned also failed. It depended on the same filter, and the query cap was reached before I could rerun it. Run both countries separately, without the industry filter, before reading anything into the 11-to-3 regional split above.

Narrowing with exclusions instead of labels

The off-limits list from scoping and the leak-risk map both turn into exclusions. Writing those as boolean strings across several tools is where mistakes get in. Refolk takes the exclusion in plain English and returns named people, so the off-limits list can go into the query as written:

A second outer-ring source is people who are not currently inside any employer's internal channels. "Former CROs now in advisory or interim roles, US, enterprise software" is a query worth running early, because those candidates can be approached without the message landing in a current colleague's inbox.

Contact order and the blind spec for each ring

Approach outward-in. Outer ring first, with a blind spec covering sector, size, and geography only. Middle ring after outer-ring NDAs are signed. Inner ring last, if at all. Each ring gets its own message, and the company name comes out at a stage you set before outreach starts.

Published practice differs on when to reveal the client. In Exec Capital's sequence, the first approach covers sector, size, and geography, the first substantive conversation adds the business model, and the second conversation discloses the company name under an NDA. Sprounix shares anonymized scope after the candidate signs an NDA and reveals identity after mutual interest, "often at finalist stage". No one has published leak rates by stage. In the worked search I used Exec Capital's faster order for the outer ring, because time is the bigger risk there. I used Sprounix's later disclosure for the middle and inner rings, because those candidates have a path back to the incumbent.

Subject: Revenue leadership role, confidential

Hi [first name],

I'm working on a confidential search for a revenue leader at a [B2B software] company in [broad region, e.g. "the US"]. It's at the [growth] stage, with a team in the [range] band.

The role owns [new business, expansion, or both] across [segment]. Your background in [one specific thing from their record] is why I'm writing.

I can't name the company in a first note. If it's worth a 20-minute conversation, I'll walk you through the business model and scope. The company name comes under an NDA once we both think it's worth continuing.

Would that be useful?
[name]

Thanks for signing. Here is the scope without the name:

- Reports to: [CEO or founder]
- Owns: [functions], with [range] in revenue headcount
- Stage and model: [stage], [sales motion]
- Why now: [a growth reason that does not describe the incumbent]

If this still fits after our next conversation, the company name is the next step, under the same NDA. Until then, please keep this between us.

Spreading approaches across units

Spread approaches across many companies and many units. That lowers leak risk and legal risk at the same time. The general rule is that businesses are free to solicit a competitor's at-will employees. Liability can arise from wrongful means, intent to injure, or leaving the target employer or one of its divisions "bereft of personnel". In California, Ixchel Pharma v. Biogen (2020) requires an independently wrongful act to interfere with an at-will contract. Leaders under contract fall under the stricter interference-with-contract tort. Guild Mortgage v. CrossCountry (2026) revived raiding claims, and Guild recovered nearly $11 million in a parallel arbitration. In one New Jersey case, liability followed after a competitor hired all of the target company's sales force. No court has set a percentage threshold. Keep a count of approaches per business unit and take anything concentrated to counsel. One practitioner case shows the outer-ring pattern working: "All six candidates approached were employed at non-competing companies."

Wrong turn: the former direct report

The costliest mistake in the worked search was approaching a former direct report of the incumbent too early. On the market map, this candidate looked like outer ring: a different current employer, a different metro, no shared board. The leak-risk map had captured the incumbent's current direct reports, but it had not done the same for the incumbent's earlier employers. The overlap only showed up after the first blind message went out. When I compared tenure dates, this person had reported to the incumbent in the same unit, at a previous company.

What I did:

1. Stopped climbing the disclosure ladder. The candidate had seen only the outer-ring spec, so no name, stage detail, or reporting line was given.
2. Logged the contact against the knower list as "approached, not informed", so the sponsor knew the exposure.
3. Moved the candidate to the inner ring and graded them exposed on the slate.
4. Rebuilt the map to include direct reports at every prior employer where the tenures overlapped, and re-sorted the rings.

The lesson for the procedure: the leak-risk map has to cover the incumbent's whole career, not just the current company.

The procedure, step by step

This procedure runs one confidential replacement search from the first list of knowers to a graded slate. The day budgets are KORE1's retained stages where they exist and estimates where they do not. Viewing hygiene comes before the leak-risk map, for the reason explained in the viewing section.

How confidential sourcing goes wrong

Confidential searches rarely fail because someone breaks an NDA. More often, something that looked safe is quietly identifying the client. Each failure below is listed with what makes it look safe and the check that catches it.

1. Semi-private viewing. Looks safe: your name is hidden. Lies because: it shows "Someone at" your company, and for an in-house recruiter the company is the secret. Check: confirm full private mode on every seat.
2. Switching to private mid-search. Looks safe: the setting is correct now. Lies because: earlier named views are reported to persist. That is a third-party claim, so assume it is true. Check: set private before the first view.
3. "No shared board" from proxy filings. Looks safe: Item 401 lists directorships for five years. Lies because: it covers public boards only, and private and nonprofit seats are not disclosed. Check: look at nonprofit 990s and investor portfolio pages too.
4. Shared past employer. Looks risky: same company name on both profiles. Lies because: the dates may not overlap, or the two may have sat in different units of a large company. Check: compare tenure dates and locations before moving anyone inward. Getting this right keeps good candidates out of the inner ring when they don't belong there.
5. Leak map scoped to the current employer. Looks safe: current direct reports are covered. Lies because: former direct reports at earlier companies are the closest ties of all. This was the worked search's wrong turn. Check: map direct reports at every employer where tenures overlapped.
6. The blind spec. Looks safe: no company name. Lies because: size, geography, and stage can point to one company. Check: ask a knower outside the deal to guess the company from the spec.
7. Several approaches into one unit. Looks efficient: one strong team, several strong candidates. Lies because: it raises leak risk and the legal risk of leaving a division "bereft" together. Check: count approaches per business unit.
8. Leak-factor percentages quoted as odds. Looks rigorous: 75%, 60%, 85%. Lies because: Boston Search Group gives no method or definition. Check: present them as risk factors, never as probabilities.
9. A zero from an index filter. Looks like: an empty market. Lies because: the industry label didn't match, as with "Computer Software" returning 0 in the worked search. Check: rerun without the filter before reporting a thin market.

Grading the slate and keeping it current

The slate is 3 to 5 ranked names. Each has an exposure grade and a one-line reason that points back to a specific line on the leak-risk map. A grade without a reason is an opinion. A grade with a sourced reason is something the board can act on.

The worked search used three grades:

• Clear: outer ring, no documented tie after the full-career map, and a metro different from the incumbent's.
• Watch: middle ring. There is a documented but distant tie, such as a shared past employer with no date overlap, or the same metro with no personal tie found.
• Exposed: inner ring, or approached before a close tie was found. The board should only advance this name once it is ready for the incumbent to hear.

Rank | Candidate | Ring | Exposure grade | Reason (with source) | Disclosure stage reached | NDA signed
1 | [name] | Outer | Clear | No overlapping employer, board, or investor; different metro (leak map lines [n]) | Company name | Yes
2 | [name] | Middle | Watch | Shared past employer, tenures did not overlap (profile dates) | Anonymized scope | Yes
3 | [name] | Inner | Exposed | Reported to incumbent at prior employer, same unit (profile dates); approached at blind-spec stage only | First approach | No

Keeping the search current while it runs

A confidential search goes stale in three ways, and each one needs a scheduled re-check:

• The clock. If the search passes 90 days, rebuild the leak-risk map. The incumbent's network has had time to change, and Boston Search Group lists long searches as a leak factor.
• The platform. LinkedIn's viewing modes and Recruiter behavior can change without notice. At the start of each search, re-read LinkedIn's help pages on profile viewing and re-run the test-profile check with a knower.
• The market. Candidates move. Rerunning the market queries in Refolk before the slate goes out catches anyone who has since joined a company on the off-limits list or the leak-risk map.