The Same-Employer Multi-Role Call, Scored to Pick One, Cluster, or Route

You've found three or more open roles at one company you want, and now you have to decide which to apply to, in what order, or whether to skip the portal and get a recruiter or referrer to place you. This is the scoring sheet for that decision. It scores the employer's setup on six dimensions and ends in a written verdict of Pick One, Cluster, or Route, with a send order and a one-line note that makes the pattern read as deliberate.

What do Pick One, Cluster, and Route mean?

Pick One means you apply to the single role you fit best and hold the rest. Cluster means you apply to a named set of two or three roles that make sense together, in a stated order, with a short note explaining why. Route means you apply to nothing cold, or to one role at most, and ask a recruiter or referrer to place you across the set.

The usual advice is a flat count: "two or three." That number comes from recruiter commentary, not data. A former recruiter writing a Monster column said she never hired anyone who applied to 20 different jobs at the same company, but did hire many people who applied to two or three. That's a useful outer limit, but it can't tell you anything about the employer in front of you. Three applications might give you three separate chances at a company that hires role by role, and nothing extra at one that runs everyone through a single shared interview loop.

Two things decide the verdict:

• Coherence. Do the roles tell one story about you?
• Access to a person. Can you name a recruiter or referrer who could place you?

The matrix gives you a first read. The six-dimension score below turns it into a decision you can defend, because it brings in the things the matrix can't see: caps, the hiring model, and lockout risk.

The six dimensions and how to score each one

Score visibility, hard caps, role coherence, level spread, recruiter overlap, and lockout risk from 0 to 2 each. A 2 favours sending a cluster and a 0 argues against it. The total out of 12, plus three override rules, gives the verdict.

A few terms first:

• An ATS (applicant tracking system) is the software behind the employer's apply form.
• A requisition is one approved opening with its own job ID.
• A pooled model is one where you go through one shared interview loop and get matched to a team afterward.
• A per-requisition model is one where each opening runs its own process.

What each signal proves, and when it lies

Every score is a claim about the employer. Here's what each one actually tells you, and the case where it misleads you.

Overrides, then totals

Apply these in order before you add anything up:

1. A referral is possible. Route first, before any cold application. Forum reports say applying cold first can block a later referral. That's unverified, but if it's true you lose the referral, and if it isn't, waiting costs you a few days.
2. The employer runs a pooled model. Pick One: the role closest to your level. Sibling roles all draw on the same evaluation, so you reach them through team matching, not extra applications.
3. A role scores 0 on coherence or level spread. Take it out of the set, then rescore the roles that are left.

Then read the total:

• 10 to 12: Cluster. Name the set and the send order. Never send more than the employer's published cap or more than three.
• 6 to 9: Route if you can name a recruiter or referrer. Otherwise Pick One.
• 0 to 5: Pick One.

Can recruiters see all my applications at the same company?

Usually yes, if you used the same email. Greenhouse, Lever, and Workable all tie one person's applications to one candidate record. Exactly what a given recruiter or hiring manager can open depends on the employer's own permission settings, and no vendor publishes a default for that.

What each system documents:

• Greenhouse keeps the person separate from the application: one candidate can have several applications, each tied to a specific job. It flags likely duplicates by email address, phone number, or LinkedIn profile URL by default, and name matching is optional. Automatic merging is a setting the employer has to switch on. Its Candidates page lists a person once per application, even when those applications sit on the same profile.
• Lever builds a cumulative profile, pulling information from every job you've been considered for. Your other active opportunities show up as a suitcase icon in the opportunity list.
• Workable links candidacies only when the email is identical. Your other candidacies are listed even to people who aren't on those hiring teams, but those people can't click into them.
• Zoho Recruit has a mode where a candidate can be attached to only one active job. If the portal refuses your second application, this is the likely reason, so score hard caps at 0.
• Workday has no primary documentation I could find. Treat its visibility as unknown and score it 1.

Meta's wait comes from a third-party interview guide, not from Meta itself.

The key point is that what joins your applications together is the identifier, not the vendor. Workable joins only on email, and Greenhouse joins on email, phone, or LinkedIn. If you use one consistent email, the recruiter sees your applications as a set, and that's what lets the set read as a deliberate choice. A second email turns it into what looks like scattered duplicates.

How many jobs can I apply to at one company?

Use the employer's own published limit if it has one, and stop at three if it doesn't. Google is the only large employer I could confirm that publishes a cap: up to three jobs in a rolling 30-day window, and 90 days before reapplying to the same job.

What employers document themselves:

• Google. Three applications per rolling 30 days, and a 90-day wait to reapply to the same job. It also says people reapplying for technical roles do best when they've added 12 to 18 months more experience.
• Amazon. Its Philippines FAQ says you can reapply 3 months after your last application. Its Colombia FAQ says applications removed for inactivity can reapply after 90 days. I found no published per-candidate cap.
• Meta. No cap. It says you can apply to more than one role, but should make sure you meet the minimum qualifications for each. It says its process usually takes about 2 to 3 months.

What practitioners say about counts

None of these are measured thresholds. Use them as outer limits, not targets:

• The former recruiter's line: many hires among people who applied to two or three roles, none among people who applied to 20.
• One staffing firm's rule: five or more roles, or unrelated roles, sent in one burst is a red flag.
• Nicole Kaiser, quoted by CNBC, says applying to several roles is "not unusual" at large enterprises.
• Maribel Valencia, in the same piece, says hiring managers at small companies are more likely to notice when candidates apply across multiple teams.

Pooled versus per-requisition hiring

This one question changes the verdict more than any count. Meta is the documented pooled example. According to interviewing.io, because the process is centralized, you can't interview with multiple teams at once, and you get one shot roughly every 3 to 6 months. Team matching now happens before the offer and involves anywhere from a handful to over 10 managers, and both you and the hiring manager have to opt in. My inference: in a pooled model, a second application to a sibling role adds nothing new, because the same loop gets evaluated either way.

No employer formally describes itself as per-requisition. The evidence is commentary. One practitioner described running full interview loops at Microsoft for three different positions with three different recruiters at the same time. If your research turns up that pattern, score recruiter overlap at 2.

Who you route to: the recruiter supply behind the Route verdict

In the US, Route usually means the employer's own recruiter, not an agency. In Refolk's index of professional profiles, three in four US Technical Recruiters work somewhere other than a staffing or recruiting firm.

All counts are from Refolk's index. The ratios and shares are derived.

What this means for scoring:

• Recruiter overlap is market-dependent. The UK pool is roughly 30 times smaller (716 against 21,415). My inference is that one UK recruiter is more likely to cover several of your target roles, so lean toward scoring overlap at 1 there and treat Route as the stronger play.
• Small employers get the same treatment. At a small company, treat recruiter overlap as close to certain. That's Valencia's warning put into the score.
• The person you find may not own the role. There are 0.35 coordinators and sourcers per Technical Recruiter, so the first name you find may be in a support role. Ask them who owns the roles, and route to that person.

Refolk's index has a small sample of top employers for these titles: Monzo and Meta in the UK, and Google, Snowflake, and Blue Origin in the US. It's 25 profiles, which illustrates who employs in-house technical recruiters but proves nothing about any one company's structure.

Step-by-step: scoring one employer's open roles

Run this once per employer, with the list of matching roles open beside it. It takes about an hour and ends with one written verdict.

A worked example

Say a mid-size employer's apply links all point to lever.co. It publishes no cap, and you match three roles: two senior backend roles on sister teams, and one engineering manager role.

• Visibility: 2. Lever builds one cumulative profile, and you'll apply with one email.
• Hard caps: 2. None published.
• Role coherence: 1 across all three, because the manager role breaks the sentence.
• Level spread: 0 for the manager role, so override 3 removes it.
• Recruiter overlap: 1. Unknown.
• Lockout risk: 2. Lever archives one opportunity at a time.

Rescored on the two backend roles, coherence and level spread both go to 2. The total is 11, so the verdict is Cluster: two roles, the stronger fit sent first, with a note.

To rank which backend role is the stronger fit, Refolk scores how well your actual history matches each posting. That turns "strongest fit first" into a comparison rather than a gut call.

Send order, timing, and the one-line note

Send your strongest fit first. Send the rest of the cluster together unless a published cap forces you to space them out. Attach one sentence that answers "why these roles" before anyone has to ask.

Sources disagree on timing:

• Together. Kaiser and one application-advice site say to submit related roles together.
• Spaced. Another advice site says to submit over several weeks.

My default is together, and here's why. When the applications land on one profile within a day with the same note, the recruiter sees one decision. When they trickle in over weeks, each new one makes the recruiter wonder whether the previous one failed. Space them only when a cap forces it. At Google, for example, a fourth role has to wait until your oldest application ages out of the rolling 30-day window.

The send order itself:

1. Strongest fit, with the note.
2. The next role in the cluster, with the same note.
3. A third role only if it passed the one-sentence test on its own and fits inside any published cap.

I've applied to [Role A] and [Role B] at [Company] because both centre on [the one shared problem or skill], which is where my last [N] years have been; [Role A] is the closest fit, and I'd welcome being considered wherever that work sits.

A note like this only works if each application is also tailored to its own posting. Writing two or three versions of the same history by hand is where drift creeps in. Refolk tailors the resume to each posting from one record of your history and drafts the cover letter, so the note and the resumes tell the same story.

The routing script

Route means asking a person to place you across the set. Practitioner advice is to mention the multiple roles early, frame them as focused interest in the company, and follow up about a week after the first contact. One recruiter on Blind claimed most companies will only consider you for one role at a time, which is one more reason to let the recruiter choose.

Hi [Name],
I'm interested in [Company] and found [N] open roles that match my background: [Role A, ID], [Role B, ID], and [Role C, ID].
They share one thread: [the one-sentence reason]. I'd rather you point me to the best fit than apply to all of them cold.
My strongest match is likely [Role A], based on [one specific piece of evidence].
If it's easier, I'm happy to apply to the one you suggest and mention you. Could I follow up next [weekday]?
Thanks,
[Your name]

The count was never the problem. A scattered set of roles is.

How the multi-role call goes wrong

The call usually fails because of an assumption that was never checked: that your profiles merged, that a blog's cap applies to this employer, that "Archived" means rejected everywhere, that similar titles mean coherent roles, or that you can apply first and get referred later. Each one below comes with the check that catches it.

Assuming the profiles merged

Workable links applications only on an identical email, and Greenhouse's automatic merge is optional. If you apply under a second email, you create a split record that looks like hidden duplicates, which is the opposite of a deliberate set. Check: use one email for every application at the employer. If you've already split, mention both applications in the note so the recruiter can join them.

Counting an assumed cap as published

The "three per 30 days" rule is Google's. Repeating it for another employer leaves you either under-applying or confident about a rule that doesn't exist. Check: find the cap on the employer's own FAQ page, or record "none published".

Reading "Archived" as a global rejection

Lever archives one opportunity at a time. When you have active opportunities for several roles, only the specified one is archived and the others stay in the active pipeline. Check: look at the status of each role separately in the candidate portal before you decide the employer has said no.

The coherence false positive

Titles look alike but sit at different levels. Kaiser's logic about applying to sister teams holds only within one level band. A senior and a junior version of the same title tells the recruiter you don't know your own level. Check: score level spread separately from coherence, and treat a 0 on either as a removal, not a deduction.

The referral you blocked by applying first

Forum reports say a cold application can stop a later referral from counting. This is unverified, but the downside is asymmetric. Check: ask the referrer how their employer handles referrals for existing applicants before you submit anything.

Hiding concurrent loops

The small-employer penalty

At a small company, recruiter overlap is close to certain, and Valencia's point is that hiring managers there are more likely to notice applications across teams. A cluster that scores 10 at a large employer can come across as unfocused at a small one. Check: at small employers, score recruiter overlap no higher than 1, and prefer Route whenever you can name a person.

Treating forum lockouts as policy, or ignoring them

No employer documents a rejection on one role carrying over to others. Reported Amazon cross-role holds, said to be 6 months after a failed screen and 2 years after a failed loop, come only from forum posts. Check: score lockout risk at 1 when the only evidence is forum reports, not 0 and not 2, and write the source next to the score so you know how much to trust it.

Keeping the call current across a long search

A verdict is a snapshot of one employer at one moment. Rescore when a role closes or a new one opens, when a recruiter replies, or when a cap window rolls over. Keep the verdict, its date, and any cap clock in one row per employer.

What changes the score mid-search:

• A role closes. Rescore the remaining set. A two-role cluster becomes a Pick One automatically.
• A new matching role opens. Run it through the one-sentence test first. If it passes and you're under the cap, add it with the same note. If not, leave it.
• A recruiter replies. Recruiter overlap is now known, so update the score. If they offer to steer you, the verdict becomes Route, whatever the total was.
• A cap window rolls over. At Google, track the date of your oldest application in the 30-day window, and the 90-day same-job date for any role you were turned down for.
• The employer's FAQ changes. Caps and reapply waits are employer policy and can change without notice. Re-read the page before each new batch rather than trusting your earlier notes.

When a new role does join the set, rerun the fit score against the actual posting rather than assuming the old ranking still holds. Refolk scores each new posting against your history, so you can see whether it should go first in the send order or not at all.