dare-images — image naming audit · 2026-05-09
Bucket: dare-images · Objects: 494 · Mean score: 88/100 · WORST: 0 · MAINTAIN: 190 · BEST: 304
TL;DR
- This is a mute issue. Fewer than 10% of names score below 60 — the existing CDN is already mostly well-named. Spend effort elsewhere.
- Total objects: 494, mean score: 88/100.
- Distribution: WORST 0 (0%) · MAINTAIN 190 (38%) · BEST 304 (62%).
Top issues by frequency
| Issue | Count | % of objects |
|---|---|---|
| too generic (1 word) — single-word names rank poorly | 178 | 36.0% |
| trailing -N counter (likely duplicate suffix) | 49 | 9.9% |
| stem is 6 chars (<8 — likely too cryptic) | 29 | 5.9% |
| stem is 5 chars (<8 — likely too cryptic) | 27 | 5.5% |
| stem is 7 chars (<8 — likely too cryptic) | 23 | 4.7% |
| stem is 4 chars (<8 — likely too cryptic) | 21 | 4.3% |
| stem is 3 chars (<8 — likely too cryptic) | 10 | 2.0% |
| contains stop words: in | 8 | 1.6% |
| contains stop words: of | 5 | 1.0% |
| contains stop words: to | 4 | 0.8% |
| contains stop words: for | 3 | 0.6% |
| contains stop words: in, the | 2 | 0.4% |
| stem is 2 chars (<8 — likely too cryptic) | 2 | 0.4% |
| contains stop words: the | 2 | 0.4% |
| contains stop words: on, the | 2 | 0.4% |
WORST — uplift candidates
These names score below 60. The ‘Issues’ column captures specific bad patterns; the suggested rename has to come from a human (the script can’t see what’s in the image). Listing the worst 30:
| # | Filename | Score | Issues |
|---|---|---|---|
BEST — exemplars worth studying
Top-scoring names in the bucket. Use these as the template when picking new names.
| # | Filename | Score |
|---|---|---|
| 1 | 19th-century-english-fashion-scene.jpg |
100 |
| 2 | 2001-kubrick.jpg |
100 |
| 3 | afternoon-tea.jpg |
100 |
| 4 | aileron-diagram.jpg |
100 |
| 5 | all-matchbooks-small1.jpg |
100 |
| 6 | anti-war.jpg |
100 |
| 7 | apple-5th-ave.jpeg |
100 |
| 8 | aston-martin-dbr2-550pix.jpg |
100 |
| 9 | autopia-europia.jpg |
100 |
| 10 | bad-ass-bike-long.jpeg |
100 |
| 11 | bad-ass-bike.jpeg |
100 |
| 12 | bangkok-mash.jpg |
100 |
| 13 | bbc-world-advert.jpg |
100 |
| 14 | better-together.jpeg |
100 |
| 15 | big-rock.jpg |
100 |
What this tells us
- Only 0% of the bucket is genuinely bad. The CDN naming is already in healthy shape. Time spent on a cleanup-purge here is probably better spent elsewhere.
- The BEST exemplars above show what ‘ranks well’ looks like for this domain.
Watch items
- Edge caching: renaming on R2 changes the URL — old URLs cache aggressively at Cloudflare’s edge. Plan rewrites with redirect rules or accept short-term 404s.
- Inbound link rot: any rename breaks external links pointing at the old URL (backlinks, social shares, archived pages). Worth checking GSC referring URLs before bulk-renaming.
- Image-required article rules: dare’s featured-strip + 404-audit toolkits scan article HTML for
<img src=>; the article HTML needs updating in lock-step with the rename. - Term-of-art detection is heuristic: the script awards a small bonus for recognised culinary / design / typography terms but can’t catch domain-specific terms it doesn’t know. The BEST list is a starting point, not gospel.
Recommendations
- Keep this audit as a periodic spot-check (quarterly?) but don’t run a cleanup pass — the existing names are fine.
- Apply the 5-rule convention to new uploads going forward; the existing cohort’s name distribution will improve naturally as new content is added.
Audit generated 2026-05-09 via dare_image_naming_audit.py. Convention reference: feedback_seo_image_naming_convention.md.