P2 · One URL per story
Sites: Prajavani (PV, www.prajavani.net) and Deccan Herald (DH, www.deccanherald.com), both on Quintype · Priority: 2 of 6 · Prepared: 18 Nov 2026 by the Product, SEO and Data team · Internal refs: #43, #21, #35, #41, #37, #28, #38, #25, #24, #30
Summary
The server router should give every story exactly one URL. Any other path that ends in a story slug or ID must 301 to story.url, or return 404/410. Today both sites render the same story at several paths, each with a self-pointing canonical, and Google indexes the wrong copy.
Ship in two steps. Step 1 is one small release: a 301 rule in the story fallback loaders (items 1–3), the one-line "Also read" link fix (item 4) and the DH non-ASCII redirect-loop fix (item 5). Step 2 is the URL hygiene work: unknown tag slugs and /sites (items 6–7), the PV redirect-list cleanup (item 8, released together with the listing canonical work in P4), the two-hop http redirect (item 9) and the DH sponsored-content indexing rule (item 10, which does not depend on the rest).
Impact
- For Google: old DH story URLs are what Google picks as canonical. In Search Console Page indexing (data to 13 Nov 2026), DH has 80,379 pages in "Duplicate, Google chose different canonical than user". PV has 1,240. A 301 consolidates each pair into the current URL.
- For readers: DH stories with an accent or curly quote in the slug are unreachable (redirect loop). Item 5 fixes that.
- Crawl waste: DH "Other file type" requests (old image paths, empty 200) are 17–31K/day from 11 Nov 2026, about 5–9% of DH Googlebot requests (Crawl stats, 18 Aug – 14 Nov 2026).
- Not proven: none of these items has a measured traffic gain. Moving signals from old to current URLs is expected but not projected. The traffic the old URLs earn today (item 1) is exposure to the migration, not extra traffic.
Items
1. DH (and PV) serve migrated stories at old URLs with 200 and a self-canonical — High (DH), lower on PV · Status: fix proposed
Issue: Old pre-Quintype story URLs such as /content/267116/jds-leader-hacked-death.html return 200 with the full story and a canonical pointing to themselves. The same story is also live at its current URL (/india/karnataka/bengaluru/jds-leader-hacked-death-2350290). Two URLs each declare themselves canonical, and Google often picks the old one.
Evidence:
- Search Console Page indexing, DH, data to 13 Nov 2026 (export of 18 Nov 2026): "Duplicate, Google chose different canonical than user" is 80,379 pages (78,587 on 21 Aug 2026). A 1,000-URL sample: 60% current story URLs, 23%
/amp/content/…, plus/amp/amp/…gallery URLs. - URL Inspection, 6 sampled URLs, 18 Nov 2026: for 5 of 5 current story URLs, Google's canonical is the old
/content/<id>/<slug>.htmlURL (e.g.…/jds-leader-hacked-death-2350290→/content/267116/jds-leader-hacked-death.html). The 6th,/amp/story/content/615609/…, has canonical/content/615609/…. - Googlebot-UA fetch, 18 Nov 2026:
/content/267116/jds-leader-hacked-death.html→ 200, self-canonical, page typestory-page.route-data.json?path=/content/267116/…returns the story withhttpStatusCode: 200. - Search Console, 19 Oct – 15 Nov 2026: 1,910 DH
/content/URLs with 298,972 impressions and 7,199 clicks. Google ranks the old URLs, so their signals are not consolidated. - PV: Page indexing "Google chose different canonical" is 1,240 (47% old
.htmlstory URLs), so exposure is small.
Cause: tpml-dh app/server/data-loaders/story-page-data.js:105–115 (ee69e96a, 17 Nov 2026). When no story has the requested slug, fallbackForOldSlug() looks the story up by story-attributes.oldslug and renders it at the old path. The request arrives through the section fallback (section-page-data.js:56). The story-redirect middleware in app/server/app.js:356–390 compares only the current slug (getStoryBySlug), so it never redirects old slugs. PV has the same fallback (tpml 0b3f7227, app/server/data-loaders/story-page-data.js:107).
Fix: In the old-slug fallback, 301 to /${newMigratedSlug} (the story's current URL) instead of rendering. Either return a redirect from the loader or do the oldslug lookup in the app.js story-redirect middleware. Do the same for /amp/content/… and /amp/story/content/… (to the current AMP or article URL). Apply on PV in the same release. Keep it narrow: redirect only when the oldslug lookup finds a story. Unknown paths still 404.
Impact: Expected to consolidate the old and current URLs over time. Evidence limit: the size of the traffic effect is not estimated.
Verify: curl -sI -A "Googlebot" https://www.deccanherald.com/content/267116/jds-leader-hacked-death.html returns 301 to /india/karnataka/bengaluru/jds-leader-hacked-death-2350290, and the target returns 200. After 4–8 weeks, watch Page indexing "Duplicate, Google chose different canonical than user" fall, and Search Console clicks move from /content/ URLs to current URLs (page filter "contains /content/").
2. PV serves every article at the site root (/<story-slug>) — Medium · PV · Status: fix proposed
Issue: https://www.prajavani.net/<story-slug> (last path segment only) returns 200 with the full article, a self-canonical, the home page title and no H1. DH answers the same pattern with a 301 to the real URL.
Evidence:
- Example, live 11 Nov 2026:
/jammu-and-kashmir-crpf-troop-terrorist-attack-soldiers-injured-4387960(real URL/news/india-news/…-4387960). - Search Console, 28 days before 11 Nov 2026: one such URL had impressions (17,153 impressions, 70 clicks; a
smartomorrows-…story). Small today, but every article has a second indexable copy whenever something links to it. - Code:
tpmlapp/server/routes.js:135(/:collectionSlugcatch-all) →loadSectionPageDatafinds no collection or section and falls back toloadStoryPageDataby slug (section-page-data.js). The page type stayssection-page, so SEO tags come from the section path, not the story.
Fix: In that fallback, 301 to story.url, as DH does. Optional: PV returns 200 for uppercase article URLs (canonical correct) while DH 301s; align if cheap.
Impact: Removes a second copy of every article. Measured traffic on it is small. Same rule as items 1 and 3.
Verify: curl -sI -A "Googlebot" https://www.prajavani.net/jammu-and-kashmir-crpf-troop-terrorist-attack-soldiers-injured-4387960 returns 301 to the real article URL.
3. PV serves a story under any wrong prefix (/zzz/<slug>) — Low · PV · Status: fix proposed (ships with item 2)
Issue: /zzz/<real-slug> returns 200 with the correct canonical on PV. DH 301s (tpml-dh app.js:354-394). Related: ?app=true copies are index, follow on PV (DH sets noindex); the canonical is correct, so this is crawl waste only.
Evidence: Live check 16 Nov 2026; source audit PV tpml upstream/master 70d7bebf. Quintype serves any /<anything>/<slug>-<story id> by the story ID (even /karnataka-news/x-4387960 returned the story with 200, checked 11 Nov 2026).
Fix: 301 any story URL whose path differs from story.url. Same rule as item 2. Optionally noindex ?app=true on PV as DH does.
Impact: Crawl waste and duplicate-URL risk only. No traffic effect is shown.
Verify: curl -sI https://www.prajavani.net/zzz/<real-slug> → 301 to story.url.
A related smaller finding from the same audit: a junk URL costs about 6 backend calls before the 404 (route order; the old-slug lookup fetches the public site with no timeout, story-page-data.js:106-112). Add a timeout on that fetch when touching the loader.
4. PV "Also read" card links to /https://www.prajavani.net/… — Medium · PV (DH for consistency) · Status: fix proposed
Issue: The desktop "Also read" (ಇದನ್ನೂ ಓದಿ) card in article bodies links with the host repeated in the path, e.g. href="/https://www.prajavani.net/entertainment/cinema/yash-ramayana-ravana-cinema-history-4165600". The link fix itself belongs to package P1 (Article server HTML). It is repeated here because the existing doubled URLs return 200 and must 301 under the rule in items 2–3, so ship them together.
Evidence:
- Page indexing samples (1,000 URLs each, export of 18 Nov 2026): doubled-host URLs are 81 of "Crawled – currently not indexed", 15 of "Alternate page with proper canonical", 90 of "Page with redirect". Story numbers reach 4392920 (stories from Nov 2026), so the bug is current.
- URL Inspection, 18 Nov 2026: two doubled URLs (
…-4383444,…-4385118) are "Submitted and indexed" with Google's canonical being the doubled URL itself. The real articles are also indexed, so Google holds two copies. - Search Console, 19 Oct – 15 Nov 2026: 47 doubled-host URLs with impressions, 2,572 impressions, 70 clicks (page filter "contains /https:"). DH: 0.
- Cause:
tpmlapp/isomorphic/components/story-templates/story-elements/also-read/index.js:51(master0b3f7227) uses<Link href={/${linkedStorySlug}}>, butlinkedStorySlug(fromgetLinkedStory(),app/api/utils.js:65) is the fullurl. The mobile card (same file, line 14) and the AMP card (app/server/amp/override/components/rows/story-elements/also-read/index.js:36) are correct. - DH has the same line (
tpml-dhee69e96a,also-read/index.js:50). Checked 18 Nov 2026: the card is not in DH's server HTML (the body is JavaScript-only), and the doubled path 301s to the real story. On DH it costs at most one redirect hop per link.
Fix: href={linkedStorySlug} on line 51 (rename the variable to linkedStoryUrl); same one-line change on DH. Existing PV doubled URLs are removed by the 301 rule in items 2–3. Two smaller defects in the same card: a nested <a> inside <a> (the Headline component renders its own link), and the card image src has w=undefined (cause not traced; srcset is correct).
Impact: Stops new doubled URLs and removes the existing 200 copies. No traffic projection.
Verify: A story with one "Also read" element has no /https: in its server HTML. After about 4–6 weeks, doubled URLs move to "Page with redirect" or drop out, and Search Console pages containing /https: reach 0.
5. DH story URLs with non-ASCII characters loop on 301 — Low (clear quick fix) · DH · Status: fix proposed
Issue: A story whose slug has a non-ASCII character redirects to its percent-encoded self forever. The article is unreachable for users and Google.
Evidence:
- Live check 16 Nov 2026:
…/el-ni%C3%B1o-…-4206975→ 301 → same URL (8 hops, then loop). 7 of 12,262 DH stories published 7 Sep – 5 Nov 2026 are affected. PV's 8 such URLs return 200. - Page indexing, DH, export of 18 Nov 2026: 111 URLs in "Redirect error" (98 on 21 Aug 2026). 80 of 111 contain non-ASCII characters (curly quotes
’40 and‘21,ñ13,é10, other accents). Most are old.htmlURLs. Several others contain a space or line break in the slug, probably the same mismatch. - Cause: the story-redirect middleware in
tpml-dhapp/server/app.js(~354–394 atupstream/main47a7e85e) comparesreq._parsedUrl.pathname(percent-encoded) with/${story.slug}(decoded). They never match, so it always redirects.
Fix: Compare decodeURIComponent(storyPathName) with /${storySlug}, and redirect to encodeURI("/" + storySlug). Also ask the desk or slug generator to transliterate slugs to ASCII.
Impact: Makes the affected stories reachable. Re-check the 111 count about 4 weeks after release.
Verify: curl -sI on …/weak-monsoon-el-ni%C3%B1o-…-4205618 returns 200 (no loop).
6. Unknown DH tag slugs return 200 (soft 404) — Low · DH (PV not yet checked) · Status: needs investigation
Issue: https://www.deccanherald.com/tags/zzqq-no-such-tag-123 returns 200, the home page title and a self-canonical instead of 404. Any misspelt or deleted tag link becomes an indexable near-empty page.
Evidence: Live check, 12 Nov 2026.
Fix: Find where the tag-page loader handles "tag not found" (tpml-dh tag data loader) and return 404 there. Run the same check on PV first.
Impact: Index hygiene only. No traffic effect is shown.
Verify: An unknown tag slug returns 404 on both sites, and real tags still return 200.
7. DH legacy Drupal image URLs (/sites/dh/files/…) get an empty 200 — Medium · DH · Status: fix proposed
Issue: /sites and /sites/<anything> return 200 with an empty body (no content type, no cache header), so every request reaches the origin and Google keeps re-crawling. (The same route is also listed under item 6 as a soft 404.)
Evidence:
- Crawl stats "Other file type", 18 Aug – 14 Nov 2026: all 367 sampled fetches are old Drupal image paths (e.g.
/sites/dh/files/article_images/2018/07/15/file710eaag883m1279z05d0153164.jpg, andstyles/…?itok=URLs). Live check 16 Nov 2026 (Googlebot UA): 200, 0 bytes. - Request volume: 17–31K/day from 11 Nov 2026, about 5–9% of DH Googlebot requests. Before 11 Nov the category also held the AMP 301s (63–164K/day).
- Cause:
tpml-dhapp/server/app.js:321-323at47a7e85e:app.use(["/sites", "/sites/:slug"], … res.status(200).send()).
Fix: Return 410 Gone for /sites/*. Where a mapping to the current image on media.assettype.com exists and the old image URLs still earn image-search traffic, 301 to it instead. Before choosing 410, check Search Console, search type Image, page contains /sites/dh/files/.
Impact: Removes origin load and wasted crawl. Image-search traffic on these URLs has not been measured.
Verify: curl -sI -A "Googlebot" https://www.deccanherald.com/sites/dh/files/… returns 410. Crawl stats "Other file type" for DH should fall toward zero over the following weeks.
8. PV hand-written redirect list needs cleanup — Medium · PV · Status: fix proposed
Issue: PV holds 303 hand-written 301s in tpml app/server/handlers/redirect-urls.js (REDIRECT_URLS.static), all from the April 2023 migration and unchanged since. Many never fire, end on a soft 404, or add a second hop. Every entry was requested live on 11 Nov 2026 (source and target, hop by hop; list at master fc684344).
Evidence (how the list is applied): app/server/app.js ~line 276: if (REDIRECT_URLS.static[req.path]) res.redirect(301, target). req.path never contains a query string, so keys with ? can never match (69 entries). In a JavaScript object a repeated key keeps only the last value, so earlier copies of /bagalakote, /ityadi and /dailyhoroscope.html are dead.
| Action | Entries | What |
|---|---|---|
| Delete: ends on search | 84 | 83 old Drupal date-sitemap pages (/sitemap/detail/days/…, /sitemap/detail/year/…) and /archives, all sent to /search. Google treats a redirect to search as a soft 404. Remove them and let them 404. |
| Delete: never fires | 69 | Keys with a query string (67 cricket scorecard URLs, /prajamatha?fbclid…, others). /sports/cricket/detailed (entry 65) already covers the cricket ones. |
| Delete: overridden or broken key | 4 | 3 overridden duplicate keys; taxonomy/term/5 (no leading /, duplicate of the working /taxonomy/term/5). |
| Change target | 35 | 22 point at a short collection URL (e.g. /national → /india-news): point at the section URL (/news/india-news, /news/cartoons, /district/mysuru). 11 point at a URL that redirects again (/pravasa → /leisure/travel → /miscellaneous/travel): point at the final URL. 2 have a relative target: add the leading /. |
| Review | 18 | End on the home page (soft 404): budget 2020/2021, /elections-2021, 7 Lok Sabha 2019 result pages, /prajamatha, /stories, /amp, /homepage, /newpv, /ys-testing, /users/…, /cartoon/24052018. Map each to the closest tag or section (e.g. /tags/lok-sabha-elections-2019), else delete and let it 404. |
| Fix target | 4 | Target returns 404: /contact-us → /contact; three /topics/karnataka-… entries point at /tags/… pages that do not exist. |
| Keep for now | 9 | Target is a hand-made collection: the horoscope pages. Repoint once the horoscope URL is decided. |
| Keep | 80 | One hop to a live page. |
After cleanup the list drops from 303 to about 80–100 entries.
Rule-based redirects (SectioinRedirects, app.js ~line 325): /national/<x> → /india-news/<x>, /stateregional/<x> → /karnataka-news/<x>, /topic/<x> → /tags/<x>. They work but land on non-canonical URLs (a second hop, not a duplicate; the canonical is correct). Point them at the final paths (/news/india-news/<x>, /news/karnataka-news/<x>). Ideally the story rule in items 1–3 also catches any story URL whose path differs from story.url.
Fix: Apply the table above. Release it together with the PV section/listing canonical rule (#19, package P4), so section targets are stable. Quintype's CMS may hold its own redirect list, which we cannot see.
Impact: Removes soft 404s and extra hops. No traffic effect is shown.
Verify: Re-run a live check of every remaining entry: each should be one 301 to a 200 page.
9. Two-hop http:// non-www redirect — Low · PV and DH · Status: fix proposed
Issue: http:// and non-www both 301 to the canonical host, but http://prajavani.net/ and http://deccanherald.com/ take two hops (to https:// non-www, then to www).
Fix: Make it one hop at the CDN. For reference, these already work: story trailing slash → 301 on both sites, and utm_ parameters keep a clean canonical.
Impact: Minor crawl and latency saving. No traffic effect is shown.
Verify: curl -sIL http://deccanherald.com/ shows a single 301 to https://www.deccanherald.com/.
10. DH sponsored stories are indexable — Medium · DH · Status: fix proposed
Issue: sponsored stories on DH (brandspot-client-story template) are indexable on several paths. The sponsored noindex rule in app.js (~655, customMetaTags) only matches paths containing /brandspot, /dhbrandspot-pr, /dhbrandpr, /brandpr, /dhsponsored-health, so /dhbrandspot/ and /connect/ stories stay index, follow. Stories under /dhbrandpr/ are blocked in robots.txt, so Google never sees their noindex and keeps the bare URLs indexed.
Evidence:
- The
brandspot-client-storytemplate produced 327 stories in 7 Sep – 5 Nov 2026 (/dhbrandpr/243,/connect/67,/dhbrandspot/17); 18% got Google impressions./dhbrandspot/jio-ipo-use-of-proceeds-…-4184885isindex, follow. - URL Inspection, 16 Nov 2026: of 44 sampled sponsored stories with no impressions, 38 are "Indexed, though blocked by robots.txt" and 2 "Blocked by robots.txt"; all 40 are
/dhbrandpr/(Disallow: /dhbrandpr/*). The 4 allowed ones (/connect/,/dhbrandspot/) are indexed normally. - Search Console Manual actions, checked 16 Nov 2026 for DH and PV: none, and no security issues.
Fix:
- Make the sponsored rule cover every sponsored path by template (
brandspot-client-story), not by URL substring:noindex(at leastrel="sponsored"on outbound links and a clear label) for/dhbrandspot/,/dhbrandpr/,/connect/. - Remove the robots.txt
Disallowlines for sponsored paths (/dhbrandpr/*,/brandspot/*,/dhbrandspot-pr/*) until Google has recrawled and dropped the pages (Page indexing, "Excluded by 'noindex' tag").noindexworks only on pages Google may fetch.
Impact: keeps paid content out of Google's index, in line with Google's guidance on sponsored content. Not proven: any traffic effect.
Verify: a /dhbrandspot/ and a /connect/ story carry noindex. Sponsored URLs are crawlable and move to "Excluded by 'noindex' tag".
Release checklist
- Implement the story-fallback 301 (items 1–3) in
tpml-dhandtpml:oldsluglookup, root slug and wrong prefix →story.url; also/amp/content/…and/amp/story/content/…. Keep it narrow; unknown paths still 404. - Apply the "Also read" link fix (item 4) and the DH non-ASCII redirect fix (item 5) in the same release.
- Return 404 for unknown tag slugs (item 6) and 410 or a mapped 301 for
/sites/*(item 7), after the image-search check. - Ship the DH sponsored-content rule (item 10) independently. Make the robots.txt change in the same step as the
noindexchange. - Release the PV redirect-list cleanup (item 8) together with the PV listing canonical work (P4), then re-run the live check of every entry.
- Make the http non-www redirect one hop at the CDN (item 9).
- Watch for 4–8 weeks after step 1: Page indexing "Duplicate, Google chose different canonical" (DH 80,379 and PV 1,240 as of 13 Nov 2026), DH "Redirect error" (111), PV pages containing
/https:.
| Test URL | Expected |
|---|---|
https://www.deccanherald.com/content/267116/jds-leader-hacked-death.html |
301 → /india/karnataka/bengaluru/jds-leader-hacked-death-2350290 (200) |
https://www.prajavani.net/jammu-and-kashmir-crpf-troop-terrorist-attack-soldiers-injured-4387960 |
301 → /news/india-news/…-4387960 |
https://www.prajavani.net/zzz/<real-slug> |
301 → story.url |
https://www.prajavani.net/https://www.prajavani.net/entertainment/cinema/yash-ramayana-ravana-cinema-history-4165600 |
301 → real article URL |
https://www.deccanherald.com/https://www.deccanherald.com/india/karnataka/bengaluru/…-4156084 |
already 301; keep a single hop to the real story |
DH …/weak-monsoon-el-ni%C3%B1o-…-4205618 |
200 (no loop) |
https://www.deccanherald.com/tags/zzqq-no-such-tag-123 |
404 |
https://www.deccanherald.com/sites/dh/files/article_images/2018/07/15/file710eaag883m1279z05d0153164.jpg |
410 (or 301 if mapped) |
http://deccanherald.com/, http://prajavani.net/ |
one 301 to https://www… |
DH /dhbrandspot/jio-ipo-use-of-proceeds-…-4184885 |
noindex |
For every check use a Googlebot user agent, and read the final status, the Location header and the canonical.
Dependencies and related packages
- P1 Article server HTML: the "Also read" link fix (item 4) is owned there. DH's "Also read" card enters the server HTML only when DH's JavaScript-only article body is fixed, so correct the link in the same release.
- P4 Listing pages and internal links: item 8 releases together with the PV listing canonical consolidation (#19). Redirect targets must match the final section URLs.
- P3 AMP: the
/amp/content/…and/amp/story/content/…redirects in item 1 must match the AMP URL set. - P5 Structured data and head tags: PV schema findings from the same audit (organisation nodes, publisher, video templates) and the DH brandspot double robots meta tag are not in this package.
- P6 Page speed, ads and analytics: no dependency identified.