On 15 July 2026 SSRN retired its public Rankings. Authors kept the download counter and lost everything to read it against. If your paper shows 7 downloads, nothing on the platform, and nothing in the literature, tells you whether that is bad, ordinary, or good.
This tool answers that question.
$ python3 ssrn_benchmark.py --downloads 7 --months 2 --field "Social Sciences"
Your figure : 7 downloads
Paper age : 2 months -> reference class "first year"
Compared to : Social Sciences (n=24)
Percentile : ~25% (higher than 25% of these papers)
That band : 25%=7 median=14 75%=30 90%=82 95%=86
Verdict : below typical for a paper of this age and field
No installation, no dependencies, no account. Python 3 and two files.
| Paper age | n | 25% | median | 75% | 90% | 95% |
|---|---|---|---|---|---|---|
| posting year | 203 | 5 | 12 | 21 | 51 | 111 |
| 1–2 years | 253 | 15 | 25 | 44 | 108 | 165 |
| 2–5 years | 375 | 28 | 52 | 101 | 208 | 332 |
| 5–10 years | 188 | 72 | 130 | 376 | 914 | 1,358 |
| over 10 years | 341 | 96 | 172 | 336 | 851 | 1,441 |
The median for a paper in its posting year is 12, far lower than most authors assume. Growth is concentrated in the first three or four years: within a field, median downloads roughly double per year over that stretch, and after five years the differences between age bands are not statistically significant (Mann-Whitney p = 0.23 to 0.82 for the three fields with enough old papers). This is a cross-section, so it shows levels by age, not the trajectory of any one paper.
Read the column downwards with care. The mix of fields in this sample changes with age (Engineering is 24% of the young papers and 3% of the old ones), so part of the rise down the column is composition rather than age.
Among papers of the same age, median downloads differ by roughly a factor of 3 across subject fields. Holding age constant in a regression on log downloads, the multiplier against Economics runs from 0.98 for Business to 0.30 for Materials Science. The raw spread within a single age band reaches 4.7, but that is the ratio of the highest to the lowest of eight medians, some resting on 23 papers; its bootstrap 95% interval is 3.2 to 7.4.
| Field | n | median downloads, age 1–3 years |
|---|---|---|
| Business; Management and Accounting | 35 | 109 |
| Social Sciences | 61 | 71 |
| Economics; Econometrics and Finance | 53 | 65 |
| Computer Science | 34 | 46 |
| Agricultural and Biological Sciences | 23 | 29 |
| Environmental Science | 42 | 28 |
| Engineering | 132 | 25 |
| Materials Science | 39 | 23 |
A percentile computed without the field can therefore mislead by a wide margin.
Pass --field whenever you know it; --fields lists all 25.
An archival cross-section of 1,360 SSRN papers, published 1996–2026 across 25 fields. Their lifetime download and abstract-view counters were read from Internet Archive captures, not from the live site.
This snapshot cannot be re-collected. SSRN now serves HTTP 403 to every automated client, and the public Rankings that gave these counts their context are gone. The window closed.
- The sample over-represents visible pages. A paper had to have a surviving, parseable archive capture to be included. Trust the shape of the distribution, not platform-representative absolute levels.
- Counters include machines. COUNTER's Best Practice on Generative and Agentic AI usage metrics (30 June 2026) requires excluding malicious bots and counting AI systems separately. Raw platform counters do neither, so some share of any count is not a human reader.
- Downloads are not citations. They are only weakly related, and this tool says nothing about citation prospects.
- Posting dates carry only a year. 1,358 of the 1,360 records are stamped 1 January, so there is no month in the data. Age is resolved to whole years and no shorter band is offered. An earlier version of this tool reported an "under 4 months" band and a claim that counts do not climb steadily over the first year; both were artefacts of that stamp (72 of the 76 captures in that band fell in a single April) and have been withdrawn.
| File | |
|---|---|
ssrn_benchmark.py |
the tool, standard library only |
study1-wayback-downloads.csv |
the data, one row per paper (n=1,360) |
Columns: aid (SSRN abstract ID), field (OpenAlex field), pub_date,
snap_ts (archive capture), downloads, abstract_views.
Chernets, V. (2026). An archival cross-section of SSRN download counts (n = 1,360). Zenodo. https://doi.org/10.5281/zenodo.22285871
The analysis behind it: What Does 7 Downloads Mean? Metric Blindness and the Cold Start of Scholarly Attention After the SSRN Rankings Sunset, https://doi.org/10.2139/ssrn.7296358
Code: MIT (LICENSE). Data (study1-wayback-downloads.csv): CC BY 4.0 (LICENSE-DATA).