31. PV article volume: 79% print copies, and they make up 96% of articles Google never shows — High (PV), analysis / decision
Added 16 Nov 2026. Source: checks/traffic_funnel/ (stories published 7 Sep – 5 Nov 2026 from the Quintype API, Search Console per URL 7 Sep – 13 Nov 2026, so each story had ≥7 days), files report_pv_2026-11-13.md, joined_pv_2026-11-13.json; DH comparison report_dh_2026-11-13.md.
Terminology: PV's print copies are published automatically by Varta Setu, the print-to-web system live from late March 2026; in Quintype they use the syndicated-print template (before Varta Setu, print copies were published as syndicated). "Print copies" below = Varta Setu / syndicated-print stories.
Decision waits on #1 (confirmed 19 Nov 2026), and why:
- Cause: a short hyperlocal story can only rank on words in its body (village, person, institution names). Google's stored copy lacks the body for ~76% of PV stories (#2) and for 9 of 10 never-shown print copies, so today print copies rank on headlines alone. #1 puts the body in the server HTML.
- What #1 can't settle by itself: the body is missing as often for shown stories as for never-shown ones, so part of the gap may be low demand for hyperlocal items. Only a before/after comparison separates the two.
- Attribution: any other change to print copies in the same window (removing,
noindex, merging, sitemap rules, publishing times) would mix with #1's effect.
- Baseline (before): stories published 7 Sep – 5 Nov 2026 (table below;
checks/traffic_funnel/report_pv_2026-11-13.md). - After: about 4 weeks after PV's #1 release, run
checks/traffic_funnel/collect.py --site pv+report.pyfor a window of stories published after the release; compare the print-copy "any web impression" share (63.7%), web clicks/story (8.5) and Discover clicks/story (121), plus the never-shown share (30% of all PV stories). The #2 body re-test runs at the same time. - Held until then: no
noindexor removal of print copies, no length rules or round-ups, no news-sitemap change for short items, and no change to the 06:00–09:00 publishing burst (it changes crawl at the same time as #1). - Doesn't wait: nothing in this issue; it is measurement first.
What: PV's "good articles get no traction" is true for the bulk of its output, not for its web-desk journalism.
| Stories published 7 Sep – 5 Nov 2026 | Stories | Any web impression | Web clicks / story | Any Discover impression | Discover clicks / story |
|---|---|---|---|---|---|
PV syndicated-print (print copies) |
30,177 (79%) | 63.7% | 8.5 | 48.7% | 121 |
PV text (web desk) |
4,749 | 95.6% | 61.9 | 86.1% | 2,234 |
PV syndicated (wire) |
2,450 | 89.1% | 39.7 | 76.0% | 943 |
PV pv-exclusive |
150 | 96.0% | 53.7 | 96.0% | 2,390 |
| DH, all templates | 12,262 | 95.4% | 103.4 | 71.1% | 1,156 |
- 11,454 PV stories (30%) had no web impression at all after ≥7 days; 96% of them are print copies. DH: 4.6%.
- New PV articles earn 8% of PV's web clicks (896,806 of 10.9M; the rest is home, cartoons, e-paper, listings; 57% of PV's web query clicks are branded, e.g. "prajavani paper", "prajavani cartoon"). DH: 40%, branded 21%.
- Print copies are published in one burst: 27,408 of 38,193 PV stories go out 06:00–09:00 IST.
- Print copies are short: 64% under 150 words. Within print copies, yield rises with length (Discover clicks/story: <150 words 74, 150–299 170, 300+ 359).
- They are not duplicates of web stories: only 218 of 30,177 (0.7%) share most of their headline words with a same-day or previous-day web story. Most are unique hyperlocal (district) news.
- Indexing is not the gap. URL Inspection API sample, 16 Nov 2026 (
checks/traffic_funnel/inspect_pv_2026-11-13.md, 300 random stories from the window): 150/150 zero-impression stories and 150/150 stories with impressions are "Submitted and indexed", fetched successfully as mobile, Google canonical = ours in all 300. Google has them and does not rank them, which fits headline-only indexing (#2) and low demand. Crawled HTML (Search Console "View crawled page", checked 16 Nov 2026,checks/traffic_funnel/crawled_html_check_pv.md): the body phrase was missing from Google's stored HTML for 9 of 10 never-shown print copies (controls with impressions: 1 of 2 missing). Small sample, but it ties #31 to #2: these stories are indexed headline-only. DH comparison: body found for 8 of 8 never-shown DH editorial stories (crawled_html_check_dh.md), so DH has no equivalent group. - Ruled out: hero image width (≥1200 vs <1200 performs the same within a template) and weekday.
Update 16 Nov 2026 (#2, 663-story test): the body is missing for ~76% of PV stories whether Google shows them or not (never-shown print copies 77%, shown print copies 76%, shown text 71%). So missing body explains why PV can rank only on headlines, but not why these particular stories get nothing: with the body equally missing everywhere, the difference is whether the headline alone matches search demand or Discover interest.
Likely causes (in order): (1) #1/#2: a short hyperlocal story can only rank on body terms (village, person, institution names), and on PV the body is missing from Google's copy for ~3 in 4 stories, so it ranks on its headline only; (2) low search demand for many hyperlocal items; (3) the 06–09 burst competes for PV's limited crawl (#3).
Not recommended now: noindex or removing print copies. They are unique content, and #1 has not shipped yet.
Plan
- Ship #1. Then compare print-copy "any impression" share and clicks/story for the 4 weeks before and after: a clean before/after test. Re-run
collect.py+report.pywith a window that starts after the release. - If print copies stay low after #1: editorial decision on minimum length / merging small items into district round-ups, and whether short items go into the news sitemap.
- Spread the print publishing burst over the morning if the CMS allows (crawl discovery), only after the #1 before/after measurement is done (held, 19 Nov 2026).