Leveling a Candidate: Reading Real Seniority From Title and Company

You have a candidate's title and employer on the screen, and you need to decide what level they actually operate at, not what the title claims. This guide is the scoring rubric for that call: how to calibrate a title against company headcount, stage, team size, and scope evidence so you can assign a normalized level and defend it. It is written for in-house recruiters, sourcers, talent leaders, and founders leveling one live profile at a time, not for building an internal comp framework or a data pipeline.

The problem is that a title alone lies in both directions. A "Head of Growth" who is the entire growth team and a "Head of Growth" leading a large org share a title but not much else. A genuinely Staff-scope engineer may carry only "Senior" at an engineer-led firm. The fix is not a better dictionary of titles. It is a repeatable read of the title against its company context, with scope as the tiebreaker.

Why the title alone cannot be trusted

A title is a claim about level, and the claim is cheap. A title upgrade costs nothing while a salary increase costs real money, so employers substitute titles for pay exactly where hiring is hardest and where you are looking most. That makes inflation systematic, not random.

The magnitude is measured. Datapeople found that a quarter of tech jobs considered junior in 2019 now carry senior titles, that twice as many junior jobs now include "Lead," and that 20% more mid-level jobs include "Senior." They define junior as under four years of experience, mid-level as four to 10, and senior as 10 or more. A secondary read of the same data reports a 38% increase in senior-sounding titles given to less-experienced people in a single year.

Inflation is asymmetric, too. It runs one way, upward, and it concentrates in the hot functions. That is why "normalize seniority" advice built on government taxonomies fails a sourcer. The federal ONET and SOC systems are occupational, not vertical: their five levels are major, minor, broad, detailed, and specific occupation groups, and their Job Zones group occupations by shared experience, education, and training. None of that gives you an IC-versus-manager rung. The 2019 ONET-SOC taxonomy carries 1,016 titles across more than 55,000 jobs, and not one of those rungs tells you whether a person leads anyone. Your normalized seniority scale has to come from industry ladders, and the calibration has to come from company context.

The dimensions that decide real level

Real level is set by four dimensions you can read from a public profile: company stage, engineering headcount, team size and reporting depth, and cross-team scope. The title word is a fifth input, and it is the weakest of the five.

Company stage and headcount are load-bearing because a formal IC-versus-manager track is typically introduced only when engineering reaches 30 to 50 people. Below that size there is no real ladder past Senior, so Staff, Principal, and Head titles are titles a founder chose, not rungs a system awarded. Above it, the same words start to mean what the ladder says they mean.

Team size and reporting depth calibrate any management-flavored title. One interviewer found a head of sales with 15 years of experience, more than 100 reports across five countries, and a multimillion-dollar B2B target, and another head of sales leading a team of two with a 200,000-euro target. Same title, three levels apart. The number of direct reports and the depth of the reporting chain separate them instantly.

Cross-team scope is what confirms a senior IC. On the standard IC ladder, IC3 Senior owns a service and drives its design, IC4 Staff leads multi-team work and sets patterns and mentors seniors, and IC5 Principal sets technical direction across a domain. The jump from IC3 to IC4 is a jump from one team to many. That jump is visible in a profile if you look for it, and invisible if you only read the label.

How the IC and manager tracks map onto one scale

To compare an IC and a manager on one scale, use the pay-anchored equivalence: Staff Engineer maps to Manager, Senior Staff to Senior Manager, and Principal to Director. This is not a metaphor. It is anchored in comp bands, and comp band is the only cross-track anchor that holds.

The pay evidence is concrete. At top tech companies, Staff Engineers earn roughly $250K to $450K and Principal Engineers $400K to $600K, matching or exceeding EM and Director compensation. Because the two tracks are paid the same at equivalent rungs, you can place a manager and an IC side by side without pretending one is more senior for having reports.

Two cautions about this table. First, the rung labels vary by company. Google runs eight levels in its standard engineering track and Microsoft runs thirteen, so the same word sits at different heights at different firms. Second, the equivalence collapses at very small companies, where no real IC ladder exists past Senior, so do not apply it below the 30-to-50-engineer threshold.

Pay band is the only cross-track anchor that holds; the title word bends under every company's policy.

What Refolk's index reveals about title-to-company fit

The clearest proof that employer size beats the title word comes from the raw distribution of who holds which title. In Refolk's index of professional profiles, filtered to the US, the IC-track pool dwarfs the "Head of" pool, and the two pools sit at visibly different kinds of company.

Two things fall out of this. First, "Head of Engineering" is scarce and clusters at small, early-stage firms, while Staff and Principal are common and cluster at Stripe, NVIDIA, Amazon, and Cloudflare. Employer size is a stronger level signal than the word "Head." Second, Principal is not rarer than Staff in the wild, with 33,356 Principal profiles against 16,828 Staff. Ladder theory says IC roles should get scarcer near the top, so this inversion is itself the inflation signal: title-granting policy varies by company, and the word alone no longer tracks rarity.

Geography moves the meaning, too. In Refolk's index, US "Head of Engineering" profiles number 2,233 and UK "Head of Engineering" profiles number 2,226, near-identical in absolute count despite the US labor market being several times larger. That means the same title is proportionally far more common in the UK, and therefore a weaker seniority signal there. When you level a UK "Head of," discount it harder than the same title in the US.

The practical lesson: read the employer before you read the title, and read the country before you trust the word.

When you are sourcing at volume, the friction is pulling the company context alongside every title fast enough to apply this rubric. Refolk returns the profile with its current employer and stage attached, so you can see the "Head of at a sub-30-engineer startup" signal in the same view as the title instead of leaving the tool to look it up.

The procedure: leveling one candidate

Run these seven steps on the profile in front of you. Steps two through five take under five minutes each once you have the practice; the whole read is under 20 minutes for a genuinely ambiguous case and under five for a clear one.

One order disagreement worth naming. Some ladder sources put scope verification before any title parsing, arguing that scope definitions are more authoritative than title labels. Others parse the title first for triage speed. Both are defensible. Parse first if you are moving fast, but let scope win every tie, because the scope evidence is what you will actually cite when a hiring manager pushes back.

How this goes wrong: false positives and the reverse error

The most valuable part of this rubric is the list of ways it misfires. Eight patterns recur. Six are false positives, where a title reads more senior than the person operates. Two are structural traps. One, the last, is the reverse error that costs you real candidates.

Taking "Head of" as director-level. A "Head of X" with zero direct reports at a 15-person startup is not a director. A Head of Marketing at a 20-person company may cover everything solo with no team to delegate to. What proves it: reporting depth and headcount. When it lies: an early-stage founder handed the title to a strong individual contributor.

Reading a finance "VP" as an executive. The banking VP is the canonical structural false positive. At Goldman Sachs, roughly 12,000 of 36,000 employees held the VP title, over 25% of the workforce, and the firm promotes about 1,500 people to VP each year against only 500 to managing director. In 2025 the firm's CEO conceded it had hired too many VPs. What proves it: the industry and the workforce ratio. When it lies: any bank or professional-services firm where VP is a mid-career rung, not a leadership one.

Matching "Tech Lead" to Staff level. It is common to perform the Tech Lead role without having the impact expected of a Staff engineer. A Tech Lead titled as Staff, but whose work never crosses one team, is not Staff-scope. What proves it: cross-team impact. When it lies: a company that titles every team's technical anchor "Staff."

Trusting "Senior" as a fixed rung. Two "Senior" candidates can sit a full level apart. A single "Senior" title spans IC3 to IC4 rather than mapping cleanly, and IC3-to-IC4 is the external-hire sweet spot where the most data and hires concentrate. What proves it: years of experience and whether they own a service end to end. When it lies: any firm that stops its ladder at Senior and lets it absorb everyone above.

Assuming a big-company ladder exists at a startup. Crediting a Staff or Principal title at a sub-30-engineer company assumes a ladder that is not there. The equivalence breaks down at very small firms with no real IC track past Senior. What proves it: engineering headcount against the 30-to-50 threshold. When it lies: an early hire who wrote the title into their own offer letter.

Solo-founder VP. A self-assigned "VP" or "COO" at a company the person founded is a title with no org beneath it. The same caution that warns against over-inflating a junior hire to COO flags founder self-titles. What proves it: whether the title predates any real team. When it lies: a founder who genuinely built and ran the function they named themselves for.

Confusing Job Zone with seniority. Treating an O*NET or SOC code as a seniority level is a category error. Job Zones group occupations by experience, education, and training, not by IC or management rung. What proves it: remembering the taxonomy is occupational. When it lies: any pipeline that maps codes to "levels" and hands you the output.

The undertitled Staff IC (the reverse error). This one costs you hires. A genuinely Staff-scope engineer may carry only "Senior" or plain "Engineer" at an engineer-led firm, where there is a CEO, sometimes a CTO, and everyone else is "just" an Engineer. What proves it: cross-team design ownership under a flat title. When you skip this check, you filter out your strongest candidates for looking junior on paper.

Verify before you commit the level

Before you record a normalized level and move the candidate forward, confirm you have done the work the rubric requires. A level you cannot defend is worse than no level, because it will anchor everyone downstream.

Keeping the read current

A leveling read has a shelf life, because both the person and the market move. Re-run the rubric when the candidate changes employers, because a title carries its old company's stage with it and means something different at the new one. The query "Former Goldman Sachs or Morgan Stanley Vice Presidents now at fintech startups" is a live test of exactly this: the VP that was structural inflation at the bank has to be re-leveled against the startup's headcount and scope.

Watch the calibration inputs, not fixed cutoffs. The 30-to-50-engineer ladder threshold is a mechanism, not a constant; a company that was sub-threshold two funding rounds ago may now have a real ladder, which upgrades its old Staff and Head titles from unverified to plausible. The title-inflation rates Datapeople measured will drift with the labor market, so re-read the direction, not the exact percentage. The one anchor that stays stable is pay parity: Staff tracks EM and Principal tracks Director because comp bands hold them together, so when a title read and a comp read disagree, trust the comp.

Finally, when the title and the scope genuinely conflict and you cannot resolve them from public signals, say so in the record. Mark the level as provisional, name the missing evidence, and pass the specific question to the hiring manager for the screen. A stated limit is what makes the whole read defensible.