What did jev-seo find on a real portfolio site?

38 orphan pages, 11 empty project pages, 42 slow pages with no compression, full agent readiness at 100/100, and a GEO score of 5 with verify. The whole sweep cost under a tenth of a cent in Jev spend.

What is an entity collision in brand search?

Two different people sharing one name. A RocketReach profile attributes a different Akash Priyadarshi's career to the portfolio owner's name. SameAs links and verbatim bios separate the identities for engines.

FIELD REPORT · 2026-09-24 · READ TIME 3 MIN

Real audit: portfolio site, 50 pages, zero spend

One full jev-seo sweep of akashpriyadarshi.vercel.app: crawl, agent readiness, sitemap, schema, GEO citation scoring, live SERP competitors, and rank tracking with provenance. Every number below came out of the binary on 24 September 2026, pre-fix, which is why the orphan pages and slow-page flags still read as open findings rather than victories. Total Jev spend for all seven commands: under a tenth of a cent.

Crawl: 0 broken, 38 orphans, 11 empty

jev-seo crawl https://akashpriyadarshi.vercel.app --max-pages 50

Zero broken links, zero redirects. The orphan rule earned its keep: 38 pages with no inbound links, including uses, privacy, and most blog posts. Eleven project pages return empty bodies. Forty-two of fifty pages load slower than 800ms with no compression signal, so gzip or brotli is the single highest-value fix. The crawl capped at 50 against a 60-URL sitemap.

Readiness: 100/100, then name three bots

jev-seo llms https://akashpriyadarshi.vercel.app

llms.txt present with six sections, sitemap advertised, robots present. Five of fifteen tracked bots named explicitly. The fix writes itself: name OAI-SearchBot, ChatGPT-User, and Claude-SearchBot to reach eight of fifteen.

Sitemap valid, lastmod thin

jev-seo sitemap https://akashpriyadarshi.vercel.app/sitemap.xml

Sixty URLs, all HTTPS, but only twenty-six carry lastmod. Adding it everywhere is a free freshness signal.

Schema valid, 100/100

jev-seo schema homepage.html

Person plus WebSite, both valid. Nothing to do, which is itself a result worth printing.

GEO 5 with verify, composite 7

jev-seo geo homepage.html --query "Akash Priyadarshi developer portfolio"

Query fit 0.95 and helpfulness 0.93 against directness 0.69 at zero confidence and no direct-answer opening. The prescription is one fact-dense definition block near the top of the homepage. Note what the tool does not do: it withholds the headline where confidence is thin instead of inventing certainty.

Brand search: position 1, with an entity collision

jev-seo query "Akash Priyadarshi developer" --provider tavily
jev-seo rank --domain akashpriyadarshi.vercel.app --query "Akash Priyadarshi"

The portfolio ranks first for the name query, and the tracker records fourth with Tavily provenance. Positions two, four, and five belong to a different Akash Priyadarshi, a CGI engineer in Bengaluru, whose career RocketReach attributes to this name. The fix is identity work, not keyword work: sameAs links to GitHub and X in the Person block, plus a verbatim bio quoted in third-party profiles.

What this run proves about the tool

Claim	Evidence in this run
Finds real issues	38 orphans, 11 empty pages, compression missing site-wide
Stays honest under uncertainty	GEO 5 with verify, directness at zero confidence shown, not hidden
Costs fractions of a cent	Full seven-command sweep under $0.001 of Jev spend
Tracks provenance	Rank row carries Tavily provider and engine, not a bare number

Frequently asked questions

Was anything fixed before publishing? No. These are pre-fix numbers, which is why the orphans and slow pages are still listed as findings, not victories.

Can I reproduce this? Yes. Every command above runs against the live site with the flags shown. Crawl caps at 50 pages by default; raise --max-pages for the full sixty.

NEXT: use cases → · benchmarks → · home