Traffic Dropped on Mobile but Not Desktop? Split It by Page

By Ugur Saritepe · October 15, 2026

Traffic dropped on mobile but not desktop? Before you open the mobile template, split the drop by page. On game-scout.app the 28-day device tab read mobile down 651 clicks and desktop down 40, and the first read on 2026-10-14 was that the loss was mobile-only. The page split said four of the five biggest losers had lost impressions on both devices.

Where this came from

game-scout.app is my other site, a catalogue of curated PlayStation lists with about 2,200 Google clicks in the 28 days to 2026-10-15. Weekly clicks peaked at 850 in the first week of September and have been 560 to 645 a week since. On 2026-10-14 I asked the assistant I work with to explain that, and the question I typed was about one page:

I want to understand first. Is this always like this? Why detective is not in serp now? Did our head queries started to have different response? What happened? why we were in rank 2 in no time and out of nowhere we are out of first page. I don't understand why it got there in the first place

The answer decomposed the loss into six pages published before July. One line in it stuck: the assistant had pulled the device compare and written that the loss was mobile-only, mobile 1,672 to 1,202 clicks against desktop 316 to 302. I had typed nothing about devices. The next morning I re-pulled the split on fixed windows to see whether that line held.

What the device tab said

sc-domain:game-scout.app, dimensions ["device"], 28 days to 2026-10-15 against the 28 before

device    clicks          impressions        ctr            position
mobile    2,371 > 1,720   100,272 > 95,471   2.36 > 1.80%   8.11 > 7.94
desktop     459 >   419    39,261 > 34,190   1.17 > 1.23%  11.79 > 9.70
tablet       47 >    38     2,236 >  2,515   2.10 > 1.51%   7.66 > 7.53

Mobile lost 651 clicks and desktop lost 40. Read as a difference, that is a mobile problem. Read as a share, it is the site: 82% of the clicks in the earlier window came from phones, so any loss lands there. The percentages sit closer than the differences do: mobile down 27%, desktop down 9%. That gap is real and I come back to it below. But 651 against 40 is mostly the mix, and the only way to find out what the rest was is to split by page.

The same drop, split by page and device

Same two windows, dimensions page and device, filtered to the list pages. Fourteen lists had at least 300 mobile impressions in both windows. The five that lost the most mobile clicks:

list        mobile clicks   mobile impr      desktop clicks   desktop impr
detective     197 >  37     2,636 >   931       22 >  8        500 > 198
trucking       96 >  64     2,207 > 2,436       12 >  6        422 > 339
fishing        80 >  40     3,624 > 2,061       10 > 10        691 > 412
space          66 >  31     2,221 > 1,256       10 > 15        626 > 406
police         45 >  22     1,834 > 1,331        5 >  6        463 > 307

Desktop clicks on these pages are 5 to 22 and cannot carry a percentage, so the desktop column to read is impressions, and impressions fell on both devices on four of the five. Detective lost 65% of its mobile impressions and 60% of its desktop ones. Fishing 43% and 40%. Space 43% and 35%. Police 27% and 34%. Those are page losses that landed on phones because that is where the pages' readers were. Trucking is the exception: mobile impressions rose 10%, desktop impressions fell 20%, and the clicks fell on both.

One page, detective, is 160 of the 651 mobile clicks lost, a quarter of the site's mobile drop. The five together are 290, nearly half. The earlier read of the detective page found its three biggest queries had each slipped two to four places, with nothing about phones in it, and nothing about phones has appeared since.

What survives the split

After the page split, one thing in the data is still device-shaped. On 10 of the 14 lists, mobile CTR fell while position barely moved. Trucking is the clean case: mobile impressions up 10%, position 6.55 to 6.68, CTR 4.35% to 2.63%, clicks 96 to 64. The animal games list: impressions 704 to 686, position 6.44 to 6.71, CTR 3.27% to 1.17%. Site-wide, mobile CTR went from 2.36% to 1.80% while desktop CTR went from 1.17% to 1.23%.

Google's own traffic-drops page files "impressions flat, clicks down" under the title link, the snippet, or a competitor gaining a rich result. On phones there is a fourth: more things on the results page between the searcher and the tenth link. I could not test it. My SERP tool returns mobile results only, so there is no desktop page to hold against them, and nothing I run stores a results page, so there is no mobile page from six weeks ago to compare with. What I can show is today's mobile top ten for the head query of each page:

"trucking games ps5", US, mobile

organic results: 9
Reddit, PlayStation Store, YouTube, Facebook, Amazon, game-scout (6),
YouTube, GameFAQs, Walmart
people also ask: yes   featured snippet: no

Fishing: eight organic results, two of them stores and two videos, game-scout at 5. Space and detective: ten results each, game-scout absent from both. Nine and eight slots instead of ten is a fact about today, not evidence of a change. I am recording it so the next read has something to compare against.

What the ranked pages say

I read the pages that rank for this question on 2026-10-15. The one written for the exact phrasing lists six device-specific causes, content missing from the mobile template first and more results-page features on mobile last, and tells you to segment by device in Search Console. It carries no named site and no number. A diagnosis guide dated 2026-10-01 does say to compare mobile with desktop and to sort the Pages table, and states that it contains no figures from any client or site. Google's traffic-drops page, updated 2025-12-10, says to check every tab including devices and to sort Pages by clicks difference to see whether a drop is site-wide, a group, or one page. Google's mobile-first indexing page, same date, is where the template cause comes from: a mobile page with less content than the desktop page can expect some traffic loss.

None of them runs the two tabs against each other, which is the whole method here. It only works because of something I could rule out in a minute: game-scout serves one HTML to every device. No m-dot site, no server-side device check, one page per URL. On a site like that the first five device causes cannot apply, and the device tab can only be reporting mix or layout.

What we have not done

The six pages that explain the loss have now been read three times and not worked. The changes decided on 2026-10-03 for them, a harvest pass on the co-op RPG list, a retitle on the flight list and a switch in the comparison routine, have not shipped. The detective list was refreshed on 2026-10-11, which voided the leave-it-alone test the prediction on it depended on, and Google had not recrawled it by 2026-10-14. And the two things that would let the mobile CTR question be answered, a desktop read of the same results page and a stored copy of the mobile one, are not built in this product.

How to run the split in ten minutes

1. Device tab, two equal windows, percentages not differences. Good: both devices fell by about the same share. Bad: one device fell by a much larger share. Either way, keep going.
2. Pages table with the device filter on mobile, sorted by clicks difference. Note the five biggest losers and the share of the mobile loss they carry. On game-scout five pages carried 45%.
3. For each of the five, switch the device filter to desktop and read impressions. Not clicks. Desktop clicks on a small site are single digits and move at random.
4. Branch on what you see. Impressions down on both devices: the page or its queries lost, and the phone had nothing to do with it. Read impressions dropped but position improved for the daily-series check that dates it. Mobile impressions down with desktop impressions flat, on a site that serves different content to phones: Google's mobile-first page and the template.
5. Mobile CTR down at flat position and flat impressions. Read the live mobile results page for the head query and count the organic slots and the features around them. Keep that read. It is the baseline I did not have this time.

The free triage report checks a URL from the outside and does not read Search Console, so it cannot run this split; it does tell you whether the page itself is indexable and serving. The rest of these measured reads are in the AI-assistant SEO series.

FAQ

Why did my traffic drop on mobile but not desktop?

Check the share first. If most of your clicks come from phones, any loss will show up as a mobile loss. On game-scout.app 82% of clicks were mobile, so a 700-click drop read as mobile 651 and desktop 40. The page split showed four of the five biggest losers had lost impressions on both devices. The device tab was reporting the mix, not a mobile cause.

How do I check whether a drop is really mobile-only?

Open the Performance report, compare two equal windows, and read the percentage change per device rather than the click difference. Then open the Pages table with the device filter on mobile, sort by clicks difference, and for each of the top losers switch the filter to desktop and read impressions. Impressions down on both devices means the page or its queries lost, not the phone version.

What counts as a device-specific cause?

A difference between what Google indexes from the mobile version and what the desktop version shows: content left out of the mobile template, an interstitial on entry, a mobile speed failure, a separate m-dot site out of sync, or more features on the mobile results page. Google documents the first on its mobile-first indexing page. If your site serves one HTML to every device, only the last one can apply.

Why did mobile CTR fall when position did not move?

On our site mobile CTR fell on 10 of 14 lists at a near-flat position, one of them with impressions up 10%. That pattern fits a change in what sits around the result on phones, such as a video carousel or store listings, but Search Console does not store the results page and I have no copy of the mobile page from six weeks ago. It stays unexplained.

In your AI assistant

Ask your AI about your own site

TrafficTriage plugs into Claude, Cursor, and ChatGPT. Your assistant runs the 8 checks, reads your Search Console data, and can even check your site every morning. Free account, no API key, nothing to install.

Sign up freeSee how it works