Scaling Talent: The L4/L5/L6 Framework

Why most founders get this wrong

The standard error is deferring the levelling exercise out of cultural concern. Founders who built early teams on egalitarian flat structures believe that introducing levels will damage the culture; they hold the line until comp inversions and senior departures force the issue. The deferred exercise is then conducted under crisis conditions with worse outcomes than a planned exercise would have produced.

The second error is introducing levels without the supporting infrastructure. A levelling framework requires comp bands per level, scope expectations per level, promotion criteria per level, and a calibration process that prevents inflation. Founders who introduce the level names without the infrastructure produce a system that has the appearance of structure without the discipline, and the comp inversions persist regardless.

The third error is over-engineering the framework on first introduction. Some companies introduce L1 through L9 from day one, complete with junior management tracks, principal IC tracks, and detailed sub-level definitions. The framework is too complex for the team size, too rigid for the operating reality, and gets ignored within two cycles. The minimum viable framework is L4/L5/L6 — three levels covering the IC track, with management as a separate parallel track introduced when the team has multiple managers.

The signals that the levelling moment has arrived

Comp inversion is visible. Engineers hired in the past 6 months are earning more than engineers hired 18 months ago at the same titular role. The gap is 10 percent or more. The pattern is structural, not anomalous.

Senior engineers are leaving. Engineers with 4+ years tenure are departing for roles at other scaleups with clearer levelling. Exit interviews mention growth path or career trajectory. The pattern is consistent across multiple departures.

Promotion conversations are ad hoc. Engineers ask "what does it take to get promoted" and there is no documented answer. Managers grant or deny promotions based on judgement rather than calibrated criteria. The pattern produces resentment regardless of outcome.

How to apply it to your round

Series B partners diligence engineering organisation through the talent-and-org slide of the IC memo. The questions are direct: how is your engineering team structured, what is your levelling framework, what is your retention rate among engineers with two-plus years tenure. A founder who can answer with the L4/L5/L6 framework, the scope rubrics, and the calibration process presents an institutional-grade engineering function. A founder who can describe the team only in terms of "we hire great engineers" presents a structural risk that affects every product-roadmap claim.

The implementation sequence:

Audit current state. Identify whether comp inversions exist (compare hire-date to current comp by titular role); identify whether senior engineers have left in the past 12 months and the exit-interview themes; identify whether promotion conversations have a documented basis. If two of three signals are present, the levelling moment has arrived.

Adopt the minimum viable framework. L4/L5/L6 with documented scope rubrics, comp bands, promotion criteria, and a calibration process. Do not over-engineer with sub-levels or principal tracks until the team has six months of experience operating the basic framework.

Conduct the levelling exercise as a project. 6 to 8 weeks, deliberately scoped, calibrated across teams, communicated individually with comp adjustments. The exercise is the operating-model intervention; without it, the framework names exist but the discipline does not.

Cross-link reading: RevOps at 100 people for the parallel three-hire operating-model decision in the revenue function; scaling the team from 80 to 200 for the Series B team-scaling narrative dimension.

Note: US scaleups typically use the same L4/L5/L6 nomenclature with materially higher comp bands and a more aggressive equity component. Companies operating across UK and US markets need parallel comp bands per geography, calibrated against the local market for each — not a single band converted at exchange rate. The geographic comp dimension interlocks with the international expansion choices covered in the international cluster.

The management track that runs alongside

The L4/L5/L6 framework covers the IC track. Most scaleups also need a parallel management track from the moment the first engineering manager is hired — typically EM (engineering manager, the manager of a single team of 4 to 8 engineers), Senior EM (manager of 2 to 3 teams or a particularly complex single team), and Director of Engineering (multi-team leadership at the function level). The management track is parallel to the IC track, not a promotion from it; the worst engineering organisations promote strong L5 ICs to EM as the only growth path, and lose senior IC depth in the process. A separate management track makes management a deliberate choice, not the default career trajectory.

The pay-equity question between IC and management tracks should be addressed explicitly. A strong L6 staff IC and a strong Senior EM should be paid at comparable bands; making management materially better-paid than senior IC creates the wrong incentives and accelerates the IC-to-management drift that hollows out engineering depth. The bands should reward both tracks equivalently up to L6/Senior EM, with divergence only at the most senior levels (Distinguished Engineer / Director of Engineering and above).

The remote and hybrid dimension

Scaleups with distributed engineering teams face additional complexity in the levelling exercise. Engineers in different geographies have different external comp benchmarks, different cost-of-living adjustments, and different visibility into the calibration process. The two structurally sound approaches are: location-based bands (each geography has its own band per level, calibrated against local market) and location-independent bands (single band per level, applied globally, accepting that this overpays in lower-cost geographies). The location-based approach is more economically efficient but harder to administer; the location-independent approach is simpler but more expensive. Companies under 200 engineers usually find location-based bands manageable; above 200, the administrative overhead becomes significant and the choice should be re-evaluated.