Analysis | Science Mission Impact

Introduction

One of the stated goals of the current NASA administration is maximizing science per dollar.¹ Another is faster time to science.² For decades, NASA has struggled to reduce science mission costs and increase project cadence.³ Despite cost overruns and schedule challenges, the National Academies have consistently highlighted the unique value of NASA’s large strategic missions.⁴ However, the National Academies’ decadal survey⁵ and congressional statute⁶ also state that frequent, low-cost science missions are a cornerstone of a balanced program portfolio.

Programmatic experiments aimed at reducing cost and development time with low-cost missions are underway. NASA’s CubeSat, SIMPLEx, and Commercial Lunar Payload Services (CLPS) programs have produced dozens of low-cost missions across all science divisions since the late 2000s. Ride shares, commercial hardware, reusable components, and in-kind contributions have further driven down costs.^7,8

Recent amendments to NASA’s competitive mission selection process have the stated aim of decreasing development and selection time.² NASA originally aimed for two New Frontiers competitions per decade and Discovery mission launches every two years, but budget uncertainty has stalled announcements, producing a 10-year gap for New Frontiers (next selection expected 2027) and a gap of more than seven years for Discovery. Meanwhile, development of new directed science missions, such as the Aeolus and Skyfall missions to Mars and PROMISE Rover to the Moon, is proceeding quickly.

The recent uptick in the number of directed and low-cost missions, paired with the changes for competed missions, reflects a reprioritization. To prepare for precisely this type of restructuring, the National Academies’ 2017 report, Powering Science, recommended NASA develop a publicly accessible database that might be used to assess the value of different mission classes.⁴ Nearly a decade later, NASA has not released such a database.

To inform current discussions in the mission science community with quantitative evidence, we compiled a comprehensive database covering 141 science missions launched since 1980. We focused on science impact, cost, and time to impactful science. The overview gives the short version; this page sets out the analysis.

Measuring impact

Research community sizes and publication cultures differ across NASA’s five science divisions: Earth Science, Planetary Science, Heliophysics, Astrophysics, and Biological and Physical Sciences. Thus, we define science impact relative to the body of work retrieved for each division. We define “field-defining” publications as those in the top 1% of citation counts for their respective division. “High-impact” works are in the top 10%. “Above-average” works are in the top 50%. Each paper’s citations count over the four years from its publication; the Methods page explains how we gathered the papers, citations and costs.

Citation thresholds are a common approach to quantifying impact, used by NSF, universities, and others. We use them two ways.

High impact thresholds

i_c=papers with citations > c

Division	Papers	Top 50%	Top 10%	Top 1%
Astrophysics	49,745	13	46	146
Biological & Physical Sciences	7	7	16	16
Earth Science	35,996	7	26	75
Heliophysics	14,417	8	25	61
Planetary Science	10,853	9	31	98

Weighted output

N sub c equals papers plus citations divided by c

Division	c	Top 50%	Top 10%	Top 1%
Astrophysics	13	2	4.5	12.2
Biological & Physical Sciences	7	2	3.3	3.3
Earth Science	7	2	4.7	11.7
Heliophysics	8	2	4.1	8.6
Planetary Science	9	2	4.4	11.9

Time to science

Results from past missions show that the relationship between cost and “time to science,” measured from start of formulation to the publication date of a mission’s first high-impact paper, is not linear. (Here each paper is ranked on all its citations to date, and failed missions are left out.) Very low-cost missions can take years to publish a highly cited result, largely negating their rapid development advantage: of the 26 missions of $50M or less that didn’t fail, only six have one, after a median 5.8 years. Very expensive missions return high-impact results quickly once operations begin (a median 0.2 years for those costing more than $1.55B, against 2.3 for missions of $50M or less; Figure 1), but face protracted development times (11.5 years from formulation).

Low-cost ($50M to $150M), but not CubeSat-class, missions have historically provided the minimum time to science, with a median of 5.5 years from formulation to first high-impact paper, though missions of $150M to $350M were close behind at 5.7. However, no single mission in our evaluation set returned a high-impact result in less than 2.7 years from formulation (QuikSCAT, a rapid reflight of the NASA Scatterometer mission, cut short in 1997 when the Japanese ADEOS satellite carrying it failed).

Years to a first top-10% paper

Years to each mission's first top-10% paper, by cost tier (failures left out)
Cost tier	Missions	With a top-10% paper	Without one yet	Median years from formulation start	Fewest years from formulation start	Median years from the start of science	Fewest years from the start of science
≤$50M	26	6	20	5.8	3.4	2.3	1.0
$50M–$150M	8	8	0	5.5	3.3	1.1	0.3
$150M–$350M	19	19	0	5.7	2.7	0.8	0.0
$350M–$750M	25	25	0	6.6	4.2	0.5	−2.1
$750M–$1.55B	27	27	0	7.5	4.1	0.3	−7.3
>$1.55B	16	16	0	11.5	7.3	0.2	−4.3

Figure 1. Median time from start of formulation, and from the start of science operations, to a mission’s first top-10% most-cited paper, binned by cost.

Science impact

The dataset indicates that a portfolio weighted more heavily toward missions costing $150M or less would reduce the volume of high-impact publications (Figure 2). The median mission costing $150M or less returned 0.18 N₅₀ per $1M; for missions of $150M to $1B the median was 1.59; for missions of more than $1B the median was 1.81.

N₅₀ per $1M, by cost

N50 per $1M (log scale): each box's missions, zeros and quartiles
Group	Missions	At 0	Lower quartile	Median	Upper quartile
≤$150M	46	20	0	0.18	1.31
$150M–$1B	60	7	0.47	1.59	4.32
>$1B	35	1	0.74	1.81	4.11

Figure 2. Citation-weighted publications (N₅₀: publications + citations ÷ the division’s 50th-percentile citation count) per million dollars spent for small, medium, and large missions. Each dot is a mission; those with no output, most of them failures, sit below the break.

Of the 46 analyzed missions costing $150M or less, 14 (or 30%) have failed or fallen short of their goals due to a technical issue, while only 8.4% of all other missions did. This means that under historical risk postures, in order for low-cost missions to sustain the publication output of larger missions, the average successful low-cost mission would have to be about 32% more productive for the cost. In fact, the $150M-or-less group averages 3.2 high-impact papers per mission, compared with 115 for larger missions (in the four divisions we compare). Normalized by cost, they are less cost-efficient than larger missions at drawing citations in three of those four divisions.

Mission successes and failures

Mission outcomes by cost tier and division: succeeded, partial failure or partial success, failed
Cost tier	Astrophysics	Biological & Physical Sciences	Earth Science	Heliophysics	Planetary Science	All divisions	Failed or fell short
≤$50M	3 succeeded, 0 partial, 0 failed	4 succeeded, 1 partial, 0 failed	11 succeeded, 0 partial, 2 failed	6 succeeded, 0 partial, 6 failed	1 succeeded, 0 partial, 3 failed	37 missions: 25 succeeded, 1 partial, 11 failed	32% (12 of 37)
$50M–$150M	1 succeeded, 1 partial, 0 failed	0 succeeded, 0 partial, 0 failed	0 succeeded, 0 partial, 0 failed	4 succeeded, 0 partial, 0 failed	2 succeeded, 0 partial, 1 failed	9 missions: 7 succeeded, 1 partial, 1 failed	22% (2 of 9)
$150M–$350M	5 succeeded, 0 partial, 1 failed	0 succeeded, 0 partial, 0 failed	5 succeeded, 0 partial, 0 failed	8 succeeded, 0 partial, 0 failed	1 succeeded, 0 partial, 3 failed	23 missions: 19 succeeded, 0 partial, 4 failed	17% (4 of 23)
$350M–$750M	6 succeeded, 0 partial, 0 failed	0 succeeded, 0 partial, 0 failed	6 succeeded, 0 partial, 3 failed	3 succeeded, 0 partial, 0 failed	10 succeeded, 0 partial, 0 failed	28 missions: 25 succeeded, 0 partial, 3 failed	11% (3 of 28)
$750M–$1.55B	4 succeeded, 0 partial, 0 failed	0 succeeded, 0 partial, 0 failed	8 succeeded, 0 partial, 0 failed	5 succeeded, 0 partial, 0 failed	10 succeeded, 0 partial, 0 failed	27 missions: 27 succeeded, 0 partial, 0 failed	0% (0 of 27)
>$1.55B	4 succeeded, 0 partial, 0 failed	0 succeeded, 0 partial, 0 failed	6 succeeded, 0 partial, 0 failed	1 succeeded, 0 partial, 0 failed	5 succeeded, 0 partial, 1 failed	17 missions: 16 succeeded, 0 partial, 1 failed	5.9% (1 of 17)

Figure 3. Mission successes and failures by NASA science division and cost. A square with one slash is a partial failure or partial success; an X is a failure. Below each tier, the share that failed or fell short.

Divisional cost-sensitivity

The data also reflect a strong sensitivity of low-cost mission output to division (Figure 4). Heliophysics missions have had more high-impact publication productivity in the $150M-or-less classes than any other division. Astrophysics missions costing $150M or less have produced less than one high-impact paper between them, but missions in the $150M to $350M range have produced the greatest return per dollar spent. Likewise, in Earth Science, missions costing more than $1.55B, and in Planetary Science, missions in the $750M to $1.55B range, have produced the greatest return per dollar spent.

These differences reflect the varying technical requirements of different science goals. For example, a $200M mission in low Earth orbit and a $200M mission landed on another planet do not belong to a common class that can predict scientific return. These distinctions indicate that a fixed cost cap does not carry the same meaning across all of NASA space science.

Efficiency ratio by division and cost

	≤$50M	$50–150M	$150–350M	$350–750M	$750M–$1.55B	>$1.55B
AstrophysicsAstrophysics	0.2×: 3 missions, 0.02% of the division's N50 for 0.11% of its cost	0.1×: 2 missions, 0.08% of the division's N50 for 0.83% of its cost	2.6×: 6 missions, 11% of the division's N50 for 4.3% of its cost; the division's highest	1.7×: 6 missions, 15% of the division's N50 for 9.0% of its cost	1.2×: 4 missions, 16% of the division's N50 for 14% of its cost	0.8×: 4 missions, 57% of the division's N50 for 72% of its cost
Earth ScienceEarth	0.1×: 13 missions, 0.06% of the division's N50 for 0.43% of its cost	no missions	1.1×: 5 missions, 3.5% of the division's N50 for 3.2% of its cost	0.4×: 9 missions, 5.7% of the division's N50 for 15% of its cost	0.8×: 8 missions, 27% of the division's N50 for 36% of its cost	1.4×: 6 missions, 64% of the division's N50 for 46% of its cost; the division's highest
HeliophysicsHelio	0.7×: 12 missions, 0.42% of the division's N50 for 0.64% of its cost	1.5×: 4 missions, 6.5% of the division's N50 for 4.4% of its cost; the division's highest	1.2×: 8 missions, 21% of the division's N50 for 17% of its cost	1.1×: 3 missions, 16% of the division's N50 for 14% of its cost	1.0×: 5 missions, 49% of the division's N50 for 49% of its cost	0.5×: 1 mission, 7.2% of the division's N50 for 14% of its cost
Planetary SciencePlanetary	0.0×: 4 missions, under 0.01% of the division's N50 for 0.15% of its cost	0.7×: 3 missions, 0.59% of the division's N50 for 0.85% of its cost	0.0×: 4 missions, 0.16% of the division's N50 for 3.2% of its cost	1.4×: 10 missions, 19% of the division's N50 for 13% of its cost	1.5×: 10 missions, 39% of the division's N50 for 25% of its cost; the division's highest	0.7×: 6 missions, 42% of the division's N50 for 57% of its cost

Figure 4. Different divisions have different cost-to-science thresholds. Each tier’s share of its division’s N₅₀, divided by its share of the division’s spending: above 1×, a tier returned more than its share. The outlined cell is each division’s highest; grey cells have no missions.

Success at low cost

Considering divisional differences, Astrophysics, Heliophysics, and Planetary Science have each produced successful missions at the low end of their division’s cost range with a high ratio of high-impact publications to mission cost: HaloSat in Astrophysics; ELFIN, TRACE, MinXSS-1, and FAST in Heliophysics; and Lunar Prospector in Planetary Science. Earth Science has none by this measure.

To gain insight into common themes that could inform program strategy for low-cost programs, we interviewed PIs and managers from four Heliophysics missions, three Astrophysics missions, two Earth Science missions, and one Planetary Science mission. Those conversations revealed three common keys to success:

A focused guiding science objective driving development.
High PI/manager authority with streamlined institutional oversight.
Post-launch funding for scientific analysis and open data access.

N₅₀ per $1M against cost, by division

N50 per $1M (log scale)

Life-cycle cost (log scale)

Figure 5. Each mark is a mission. Rings cost $150M or less; the named ones beat the median of their division’s costlier missions (dashed line). Failures are marked ×; missions with no output sit below the break.

Selection strategies

Most low-cost missions in our set (40 of the 45 with a recorded selection method), including nine of the 10 captured in interviews, were competed (Figure 6). Among mid-sized missions ($150M to $1B), where both types are well represented, competed missions have produced more impactful science output per dollar: a median 1.85 N₅₀ per $1M, against 1.30 for directed missions.

Competed and directed missions

DirectedCompeted

N50 per $1M (log scale): each box's missions, zeros and quartiles
Group	Missions	At 0	Lower quartile	Median	Upper quartile
≤$150M, Directed	5	2	0	0.09	0.75
≤$150M, Competed	40	17	0	0.31	1.60
$150M–$1B, Directed	24	4	0.25	1.30	2.38
$150M–$1B, Competed	35	2	1.11	1.85	6.21
>$1B, Directed	31	1	0.99	1.86	4.11
>$1B, Competed	4	0	0.66	0.79	2.74

Figure 6. Comparison of competed vs. directed missions. Citation-weighted publications (N₅₀) per million dollars spent for small, medium, and large missions. Two missions with no recorded selection method are left out.

CLPS missions are too immature to include in the main analysis, but as of Oct. 28, 2026, they have collectively produced fewer publications, in total and per dollar, than any other successful surface mission at the same point after the start of prime operations (Figure 7).

CLPS and other planetary surface missions

Papers per $1B, first 33 months

Life-cycle cost (log scale)

Papers per $1B: surface missions in their first 33 months, CLPS landers to date
Mission	Selection	Life-cycle cost	Papers	Papers per $1B
DART	Directed	$400M	140	350
Deep Impact (EPOXI)	Competed	$624M	139	223
LCROSS	Competed	$139M	24	173
Mars Pathfinder	Directed	$613M	67	109
Spirit	Directed	$756M	100	132
Opportunity	Directed	$756M	108	143
Phoenix	Competed	$869M	105	121
InSight	Competed	$1.4B	118	86
Curiosity	Directed	$4.0B	221	55
Mars 2020	Directed	$3.6B	206	57
DS-2 (failed)	Directed	$58M	0	0
Mars Polar Lander (failed)	Directed	$306M	0	0
Peregrine (CLPS lander, to date)	—	$130M	3	23
IM-1 Odysseus (CLPS lander, to date)	—	$147M	6	41
Blue Ghost M1 (CLPS lander, to date)	—	$174M	4	23
IM-2 Athena (CLPS lander, to date)	—	$170M	3	18
CLPS, 4 landers combined	—	$621M	16	26

Figure 7. Publication volume 33 months after the start of prime mission divided by cost for planetary surface missions: refereed papers per $1B of life-cycle cost. CLPS counts all four landers from the first launch; each lander’s own mark is its papers to date.

Maximizing science per dollar

Our results do not support investing in more directed and low-cost missions purely as a means of reducing the cost of science. Among the most efficient missions in terms of high-impact publications per dollar have been competed mid-to-large missions (e.g., Discovery, New Frontiers, MIDEX, Earth System Science Pathfinder). Given the outsized success of these missions, accepting more risk in the mid-to-large tiers may yield cost savings.

Beyond science

The cases for missions of different sizes extend beyond cost alone. Small missions broaden participation, train early-career researchers on flight hardware and testing, build institutional capacity, and mature affordable technologies that other missions and industry later adopt. Large strategic missions develop radically new technologies, access domains once thought impossible to investigate, sustain entire fields for decades, and offer high-profile opportunities for international visibility and collaboration. These factors should each be weighed alongside science when gauging the value of a new program.

References

1. NASA (2026). A message from Administrator Jared Isaacman.
2. NASA Science Mission Directorate. Faster to science: a letter to the science community.
3. McCurdy, H. E. (2003). Faster, Better, Cheaper: Low-Cost Innovation in the U.S. Space Program.
4. National Academies Press (2017). Powering Science: NASA’s Large Strategic Science Missions.
5. National Academies Press (2023). Origins, Worlds, and Life.
6. CHIPS and Science Act, H.R. 4346, 117th Congress (2022).
7. Fiore, F., & Elvis, M. (2025). Space Policy, 101713.
8. Kim, M. J. (2025). Acta Astronautica, 232, 633–639.