Net-zero retrofit strategies for existing housing are often justified using operational energy savings alone, yet high-performance retrofit packages can introduce substantial upfront embodied carbon. The light scenario is treated as an operational-control package with no additional A1–A5 material inventory. This paper evaluates light, medium and deep retrofit pathways for a UK multi-family residential case study using operational carbon modelling and embodied carbon assessment of retrofit measures (modules A1–A5). Operational emissions over the assessment horizon reduce monotonically with retrofit depth, from 182.76 kgCO₂e/m² (baseline) to 172.43 kgCO₂e/m² (light), 103.82 kgCO₂e/m² (medium) and 70.08 kgCO₂e/m² (deep), driven by fabric performance improvements and fuel switching to an air-source heat pump. However, embodied carbon increases sharply with retrofit intensity. The medium pathway adds approximately 30.46 kgCO₂e/m² (A1–A5), while the deep pathway adds approximately 137.63 kgCO₂e/m² (A1–A5). When embodied and operational impacts are combined, the medium retrofit yields the lowest cumulative whole-life carbon at approximately 134.2 kgCO₂e/m², outperforming the baseline, light and deep pathways, whereas the deep retrofit produces the highest cumulative value at approximately 207.6 kgCO₂e/m² despite its lowest operational emissions. Carbon payback period results reinforce that payback alone is insufficient: the medium pathway achieves a CPP of 4.6 months, while the deep pathway achieves 1.22 years, yet the deep package still yields higher cumulative whole-life carbon due to its embodied “overshoot”. The findings indicate that for decarbonising energy systems, medium-depth retrofit provides the strongest near-term carbon balance, while deep retrofit provides the lowest cumulative WLC over longer assessment horizons under static emission-factor assumptions. The carbon-optimal retrofit depth is therefore time-boundary dependent.