7.0%per year
The share of engineers who choose to leave in a year. This is the headline number, and it is internally audited. It is what makes a named-squad proposal still credible a month later.
Proof · Retention · Evidence
The stability you're
buying, in numbers.
The evidence you need to trust the promise: audited retention, named leadership and referenceable work. Where a figure is internally audited rather than externally validated, we say so plainly, so you can take it to your board.
Senior & vetted
Global reach
Productive in weeks
Built to deliver
Evidence systemEvery figure carries its method and validation status
The proof points
Four numbers, each with its source.
A figure without a method is an adjective. Each proof point below carries its definition and its current validation status. Nothing is presented as more settled than it is.
The evidence, at a glance
Voluntary attrition / year
Tech Teamz7.0%
Industry~18%
Time to a productive senior engineer
Tech TeamzWeeks
Hiring9–18 mo
Two comparisons carry the case. The table below sources every figure and states exactly what is audited and what is not yet externally validated.
Attrition internally audited; the ~18% is one client's measure, illustrative.
| Proof point | What it measures | Status |
|---|---|---|
| Voluntary attrition ≈ 7% | Engineers who choose to leave in a year, roughly half the level the large firms report. | Internally audited |
| Engineer retention ≈ 74% | Multi-year retention, all causes, including involuntary and performance-managed exits. | Methodology pending |
| Onboarding in weeks | Time to a productive squad, against nine to eighteen months to hire a senior engineer. | From live engagements |
| Customer satisfaction | Relationship quality, captured through a structured client survey. | Baseline in progress |
Attrition & retention
Two numbers, one story.
They invite conflation ("7% attrition but only 74% retention?"), so we never show one without the other, or without the line that defines each.
~74%all causes
Retention across several years counting every departure: voluntary, involuntary and performance-managed. A lower number than attrition alone implies, because it is a stricter measure. Published only with its methodology attached.
Why it matters commercially. Replacing an experienced engineer typically costs one to two times salary and, more importantly, disrupts delivery: productivity falls during the transition and a replacement takes months to reach full effectiveness. Stability is what a client is actually buying. One current client measures internal attrition near 18% against our 7%: longevity they can buy rather than build.
Case studies
Referenceable work, by sector and discipline.
Written from the founders' first-hand account of the work. Named, client-signed versions are in preparation; until each is signed off, the engagement is shown by sector and discipline, not by client name.
GLOBAL SEMICONDUCTOR LEADERSign-off pending
BSP integration & OEM release support
Board-support-package integration and OEM product-release support alongside the client's own platform teams. Customer Engineering at the standard the most demanding accounts expect.
BSP integrationOEM release support
SECURITY-IP COMPANYSign-off pending
Security-software testing & release verification
Independent security-software testing and release verification on security-critical products, with verification kept structurally separate from the development organisation.
Security testingRelease verification
SWISS ULTRA-LOW-POWER CHIPMAKERSign-off pending
QA, firmware verification & wireless certification
Software QA, firmware verification and Bluetooth certification for a specialist product company, with a specialist RF and Bluetooth-certification placement seeding the account.
Firmware verificationBluetooth certification
FUNDED SATELLITE-CONNECTIVITY STARTUPSign-off pending
Algorithm validation & RTL verification
MATLAB/C++ algorithm validation, RTL verification and mixed-signal SoC work for a recently funded deep-tech startup, with named profiles inside a week.
Algorithm validationRTL verification
The record also includes high-speed SerDes analog verification for an interface-IP company. Full, client-signed case studies (challenge, squad, way of working and outcome) are released as each client approves.
Clients
A roster that spans the field.
Global semiconductor leaders and sub-fifty-person funded startups alike, every one buying the same core capability. Named references and marks are shared with permission; some are available under NDA ahead of public sign-off.
Established semiconductor & systems
Global semiconductor leaders and larger IP and systems firms running multi-year silicon, firmware and embedded programmes.
Specialist product companies
Ultra-low-power, RF, imaging, agricultural and industrial electronics firms: precise technical matches to our disciplines.
Funded deep-tech startups
Recently funded companies scaling engineering faster than any hiring plan allows, reached through VC and founder networks.
On naming. Public client figures are sequenced behind the external-validation calendar and individual sign-offs. We would rather show a mark late than claim one we have not earned the right to use.
The retention report
A retention figure you can rely on.
An externally validated retention report has been commissioned. It will disclose the methodology in full, separating voluntary departures from involuntary and performance-managed exits, so both figures are unimpeachable at source before either is marketed.
Until it is published, we present attrition as internally audited, and say exactly that. When something is not yet verified, the brand says so.
Methodology disclosed
Voluntary vs. all-cause separated
External validation
REPORT STATUS
In review
Not yet externally validated
| Attrition figure | Internally audited |
| External validation | Commissioned |
| Methodology | Disclosed on release |
| Marketed | After validation |
The teal Hallmark marks only reviewed and signed artefacts. This report has not yet earned it, and will carry it when it does.
See the evidence in your own context