There is a pattern every systems engineer recognizes. At SRR, the mass budget shows 30% margin. By PDR, it is 20%. By CDR, you are fighting for 10% and reclassifying items to make the numbers work. The spacecraft did not get heavier overnight. The estimates got more honest.
AIAA and NASA program data show mass growth from concept to launch averaging 25-35% for typical space missions. This is not a function of bad engineering — it is a function of optimism bias in early-phase estimates combined with discoveries during detailed design. Knowing the average mass growth, you can back-calculate the margin you actually need at SRR (closer to 35%) versus the margin you usually report (closer to 25%).
The problem is how mass budgets are typically managed. In a spreadsheet, each subsystem lead enters their best estimate. They round down because no one wants to be the reason the design does not close. Harness mass gets estimated at 5% of dry mass when the actual number is closer to 8%. Thermal blankets get a placeholder that never gets updated. The margin looks healthy because the estimates are systematically biased low.
There are five categories where mass estimates lie consistently. First, harness — typically estimated at 4-6% of dry mass and almost always closer to 7-9% by integration. Second, thermal blankets — usually a placeholder until thermal analysis is complete in late Phase B. Third, brackets and secondary structure — invisible during early design and accounting for 5-10% of dry mass at launch. Fourth, fasteners — never explicitly tracked but adding 1-2% to the total. Fifth, contamination control hardware (purge fittings, witness coupons, removable covers) that no one wants to own.
Margin policy by phase, drawn from NASA Goddard and JPL guidelines: at MCR, carry 35% mass margin against the dry mass current best estimate. At SRR, 30%. At PDR, 25-30%. At CDR, 15-20%. At PSR / FRR, 5-10%. These numbers feel high because they are higher than what most programs actually carry. But every program that carried less than these numbers and ran into mass issues during integration is a program where someone wished they had been more honest at PDR.
A connected mass budget fixes this by tying allocations to actual component selections, not estimates. When you select a specific reaction wheel, the mass budget pulls the datasheet mass, not an engineer's guess. When the harness routing is defined, the budget reflects measured lengths and wire gauges. The margin is real because the inputs are real.
Maturity levels matter as much as mass values. The standard NASA Mass Properties Control Board uses six maturity codes: estimated (E), calculated (C), measured (M), as-built (A), and so on. A budget that shows "100 kg dry, 25% margin" with no maturity column is opaque. A budget that shows "100 kg dry, 25% margin, 60% E / 30% C / 10% M" tells you that most of the mass is still guesswork and the real margin is much smaller than 25%.
The honest way to report margin is to report it against measured values, not estimates. Take your current best estimate and assign each line item a confidence level. Apply growth allowances by maturity (typically 20% for estimated, 10% for calculated, 5% for measured, 0% for as-built). The growth-adjusted total is your honest worst case. Margin against the honest worst case is the number that should appear on your phase gate slide. Margin against unadjusted estimates is the number that gets you in trouble at integration.
Power and link budgets follow the same pattern. Power consumption gets estimated low because no one wants to be the reason the array gets bigger. Link margin gets reported optimistically because the antenna gain is at boresight, not realistic pointing. The discipline is the same: track maturity, apply growth allowances, report against the worst case.
Hitt Hosting SE's mass budget tracks allocations at the component level with margins computed at each tier. When a subsystem allocation changes, the total margin updates immediately. You see the design close or not close in real time, not at the next monthly program review. The same applies to power, link, data, and thermal budgets — all live in one place, all maintained automatically, all visible to the review board on demand.