What Changed in Google’s Mobile Organic Results in May?
Same search. Nearby location. Different organic results. Across 8.16 million location specific SERP snapshots, Yext found a sharp shift in mobile organic results beginning in mid May 2026.
Research question raised by Joy Hawkins of Sterling Sky · Data source: Places Scout · Analysis: Yext Research
8.16 million valid location specific organic SERP snapshots from January 1 through August 9, 2026
mobile organic · geographic volatility · local search · classic 3 pack · Maps
Something changed in Google’s search results in May 2026.
Schieler De’Land first flagged it to Joy Hawkins after seeing people run the same search at the same time and get different organic results. Around the same time, similar reports started appearing in the Local Search Forum. Joy was seeing it too in Places Scout ranking grids, especially on searches where the city was already included in the keyword.
The question was whether this was just random ranking noise or something much bigger.
So Joy asked Mark Kabana at Yext to dig into the historical data.
What we found was not subtle.
The important change is not that rankings stopped mattering. It is that one observed ranking became a weaker description of an entire market. Clear winners still exist, but identifying them now requires looking at the pattern across many observations rather than treating one search as ground truth.
This study measures geographic variation, not conventional day to day ranking volatility. We compared the ordered top 10 organic results across nearby locations during the same report run.
The grid that made the pattern visible
Joy’s Places Scout grids gave us a concrete example to test. In an earlier scan, the tracked site’s organic ranking was relatively consistent across the market. By May 19, the same report looked far more scattered.
That May 19 scan was collected two days before Google’s May core update began. The timing made the pattern even more interesting, but one grid could not tell us whether Google had changed more broadly.
The two scans show what Joy saw. The analysis below tests whether that visual pattern held across the full dataset.
Open full-resolution image
What changed in the frame. The May scan shows the tracked website appearing across a more scattered set of grid locations than the February scan.
For each keyword, we compared the ordered top 10 organic results between adjacent points on a search grid. A higher score means nearby locations received more different results.
The index sets each report and keyword’s earlier baseline to 100. A value of 125 means geographic volatility was 25% above that baseline.
We tested the pattern across 8.16 million SERP snapshots
Was this a handful of strange grids, or a broad change in Google’s mobile organic results?
The study covered 1,272 reports and 9,371 report runs. Each organic snapshot represented one keyword, one location, one run, and one device. Each snapshot contained up to 10 traditional organic results.
Organic, the classic 3 pack, and Maps were scored separately. Each report received equal weight after its query series were summarized, so large reports could not dominate the finding.
| Measure | Final count |
|---|---|
| Reports with usable output | 1,272 |
| Report runs | 9,371 |
| Organic keyword and run observations | 132,512 |
| Local keyword, run, and surface observations | 263,829 |
| Valid organic SERP snapshots | 8,162,307 |
| Approximate individual organic listings | 73.85 million |
| Organic valid coverage | 99.07% |
| Report-level read failures | 2 |
The break begins the week of May 18
From January through early May, the mobile organic index stayed within its normal range. It rose during the week of May 18, climbed again during the week of May 25, and stayed high through June and July.
That timing overlaps Google’s May core update, which began May 21. The study measures the overlap but does not prove the update caused the change.
The baseline equals 100. The mobile organic line moves above its earlier range in mid May, while searches that already named a location rise the most.
All three series moved above 100 after mid May. Explicit location queries reached the largest peaks, even though those queries already name a location.
Mobile organic breaks away after mid May. The classic 3 pack and Maps remain close to their historical baselines.
The median paired report change was 7.20 points. 702 of 718 reports increased and 16 decreased. The 95% interval for the median report change ran from 6.82 to 7.49 points.
View the published values
| Measure | Value |
|---|---|
| Reports measured in both periods | 718 |
| Comparable query series | 10,524 |
| Baseline score | 18.86 |
| Later score | 28.04 |
| Relative increase | 48.7% |
| Reports that increased | 97.8% |
| Median paired report change | 7.20 points |
The biggest surprise came from searches that already named a location. Their matched geographic-volatility score rose from 8.55 to 15.08.
The biggest surprise: searches that already named a location
High-confidence explicit location queries rose from 8.55 to 15.08, a 76.3% increase. Near me searches rose from 26.04 to 33.91, a 30.2% increase.
The two organic query groups show the widest gaps. Mobile classic 3 pack and Maps rise by less than one raw point.
View the published values
| Group | Earlier score | Later score |
|---|---|---|
| All mobile organic | 18.86 | 28.04 |
| Explicit location queries | 8.55 | 15.08 |
| Near me | 26.04 | 33.91 |
| Mobile classic 3 pack | 23.28 | 23.84 |
| Mobile Maps top 10 | 18.16 | 18.66 |
The 3 pack and Maps did not follow
Mobile classic 3 pack volatility changed 2.4%, from 23.28 to 23.84. Mobile Maps changed 2.8%, from 18.16 to 18.66. Those shifts were tiny compared with mobile organic.
The change is concentrated in mobile organic. Explicit location queries rose 76.3%, mobile organic rose 48.7%, and near me rose 30.2%. The mobile local surfaces stayed below 3%.
View every surface value
| Surface | Relative change |
|---|---|
| Explicit location queries | 76.3% |
| Mobile organic | 48.7% |
| Near me | 30.2% |
| Desktop classic 3 pack | 12.0% |
| Desktop organic | 9.6% |
| Mobile Maps | 2.8% |
| Mobile classic 3 pack | 2.4% |
| Desktop Maps | 0.8% |
The change did not fade after May
The longer January baseline shows normal earlier spikes, but no earlier period stayed as high as mobile organic did from late May through July.
The monthly view shows that the mobile organic increase lasted through June and July instead of fading after one volatile week.
Mobile organic holds above its earlier range in June and July. The other three series stay close to 100, apart from smaller desktop movement.
Early August was noisy, but it was not a second jump
From August 1 through August 9, mobile organic geographic volatility was 15.6% lower than the unusually high late July comparison. The 3 pack was 6.4% lower and Maps was 2.6% lower.
August still sat above the longer mobile organic baseline. It simply did not create another upward step in geographic volatility.
Mobile organic eased from 36.9 to 31.2. The classic 3 pack moved from 34.6 to 32.4, and Maps moved from 25.6 to 25.0.
View the published values
| Surface | Jul 20 to Jul 31 | Aug 1 to Aug 9 |
|---|---|---|
| Mobile organic | 36.9 | 31.2 |
| Mobile classic 3 pack | 34.6 | 32.4 |
| Mobile Maps | 25.6 | 25.0 |
One website’s rank became less consistent too
The share of pins matching the center rank fell from 38.9% to 31.5%. The share within one position of the center fell from 70.6% to 56.1%.
Both center agreement measures declined. More pins ended at ranks that differed from the center pin.
Same location comparison. Mobile organic ranking change at the same location rose 7.5% for runs separated by 3 to 10 days, far less than the 48.7% change across nearby locations.
Fewer pins matched the center after the shift, either exactly or within one position.
Mobile organic rose from 28.1 to 30.2. The classic 3 pack, Maps, and desktop organic declined in the same location group measured in both periods.
View the published values
| Surface | Earlier score | Later score |
|---|---|---|
| Mobile organic | 28.1 | 30.2 |
| Mobile classic 3 pack | 14.5 | 12.1 |
| Mobile Maps | 11.6 | 10.0 |
| Desktop organic | 24.2 | 23.2 |
Reports with the same result depth also became more volatile, so shorter result lists do not explain the change.
The result survived every major cross check
The finding remained when we compared domains, landing page URLs, center overlap, grid sizes, and reports with the same result depth before and after.
Nearby result sets were not only reordered. They also shared fewer domains, and the number of unique SERP variations increased.
| Supporting measure | Baseline | Later | Relative change |
|---|---|---|---|
| Neighbor URL volatility | 22.04 | 32.15 | +45.9% |
| Center domain volatility | 26.94 | 34.49 | +28.0% |
| Center URL volatility | 31.85 | 40.74 | +27.9% |
| Center domain Jaccard overlap | 64.03% | 55.23% | -13.8% |
| Unique ordered domain SERPs | 30.88 | 34.50 | +11.7% |
| Average organic result count | 9.35 | 9.06 | -3.0% |
Three volatility measures rise by 27.9% to 45.9%. Center domain overlap falls 13.8%, which points to fewer shared domains across nearby pins.
Grid-size sensitivity. The mobile organic increase appeared across 49-, 81-, and 121-location grids rather than depending on one layout.
The increase measured 29.0% for 49-location grids, 56.1% for 81-location grids, and 57.8% for 121-location grids.
It was not limited to one service category
Mobile organic volatility rose across all five broad service groups in the sensitivity check. The increases ranged from 21.2% to 70.6%.
These are broad groupings, not a formal industry model. Their purpose is to show that one category did not create the overall result.
The relative increase ranged from 21.2% to 70.6%. The share of reports that increased ranged from 95.2% to 100.0%.
View the published values
| Broad service group | Relative increase | Reports increased |
|---|---|---|
| Other | 70.6% | 98.4% |
| Legal | 68.6% | 100.0% |
| Senior and home care | 29.9% | 95.2% |
| Home and property services | 23.8% | 96.7% |
| Healthcare and dental | 21.2% | 100.0% |
A messy rank grid does not always mean the full SERP changed
The Toronto grid looks chaotic for one tracked website, but the full organic SERP changed only 8.6%. The Vaughan example shows a much larger full SERP change of 55.9%.
The heatmaps show one tracked website. The percentages below measure the full ordered organic SERP.
Open full-resolution image
How to read the examples. The Toronto heatmap looks more scattered for one tracked website, but its full organic spatial score increased 8.6%. Vaughan's full organic score increased 55.9%.
Maps and the classic 3 pack moved differently from mobile organic across these examples. The surface contrast is why each system should be monitored separately.
What might explain the timing?
We know when the break happened. We do not know exactly what Google changed. But three explanations now look worth testing.
1. The May core update may be part of the story. Our first clear rise begins in the week of May 18. Google says the May 2026 core update began May 21 and ran through June 2. The timing lines up. But Joy’s May 19 grid already showed the new pattern, so we should treat the core update as a possible part of the change, not a proven cause.
2. Google may be using more context when it chooses organic results. On May 19, Google expanded Personal Intelligence in AI Mode. That feature is about AI Mode, not traditional organic results, so it does not prove anything about blue-link rankings. Still, it shows where parts of Search are heading: more context, more personalization, and fewer one-size-fits-all answers. In the Local Search Forum discussion, Darren Shaw of Whitespark said different people and tools can get different result sets while one person’s own results may stay fairly stable. Schieler De’Land also found different logged-in users seeing different organic winners for the same query. That makes a broader context-based change in how Google serves organic results a reasonable theory.
3. Different search environments may be getting different SERPs. Other practitioners have reported large gaps between browser searches and rank-tracking environments, including in Reddit discussions. One theory is that Google may treat those request types differently. That could make the effect look stronger in some environments. But it cannot explain everything: Joy says she is seeing the behavior with human users too, and Schieler’s study used logged-in human searches. So this may be part of the story without being the whole story.
A sustained shift in mobile organic geographic volatility began in mid-May 2026. The timing overlaps the May core update and Google’s broader move toward more contextual Search experiences.
What we cannot say: that Personal Intelligence, memory, bot detection, or the core update alone caused the change. This study was not designed to isolate those mechanisms.
Position #1 is not literally gone. A rank is still a real observation. The narrower conclusion is stronger: one observed rank is less representative of the wider mobile market than it used to be.
What this changes about SEO measurement
Rankings still matter. What changed is how much confidence we should place in one observed rank.
A single mobile search is better treated as one sample from a market than as a description of the entire market. In our data, nearby locations increasingly returned different ordered organic results after mid May, even when the query itself already named a location. A single scan can therefore be accurate and still give an incomplete picture of visibility.
For market level reporting, rank should be treated as a distribution rather than one number. Measure how often a brand appears across a defined market, how frequently it reaches prominent positions, how much visibility it holds relative to competitors, and how those patterns change over time. Keep location, device, and sampling methodology consistent when comparing periods. An average alone can also conceal meaningful differences across a market.
Rank should then be connected to business outcomes. Calls, clicks, leads, and revenue tell you whether search visibility is producing value. A volatile ranking report does not automatically mean business performance deteriorated, just as one first place ranking does not prove a market has been won.
Search surfaces should also be measured separately. Mobile organic changed dramatically in this study while the classic 3 pack and Maps barely moved. Combining them into one ranking story would hide the very difference this research uncovered.
Questions readers ask
What changed in Google mobile organic results in May 2026?+
Did the classic local pack and Maps change too?+
Why did explicit location queries rise the most? +
Did the May 2026 core update cause the change?+
Could Google’s personalization changes be related? +
How is this different from conventional volatility tracking?+
Did early August create another geographic jump?+
How this was measured
- Source records
- Each source record represented one report run, one keyword, and one search location. Mobile and desktop organic, classic pack, and Maps were analyzed separately.
- Organic identity
- Traditional organic results used normalized domain and landing page identity. Query strings and fragments were removed before the ordered top 10 was compared.
- Local identity
- Local identity preferred Google's CID business identifier, followed by the record's internal local-page identifier, then a conservative name plus address or phone fallback. CID coverage ranged from about 99.8% to 100% across local surfaces.
- Geographic pairing
- The center pin was the valid point nearest the geographic centroid. Neighbor pairs used physical distance with longitude adjusted for latitude. Equally adjacent neighbors were included and duplicate unordered pairs were removed.
- Ranked lists
- Rank Biased Overlap with persistence 0.90 compared ordered lists and gave more weight to top positions. Jaccard overlap measured shared membership without rank order.
- Missing local output
- A successfully observed applicable surface with zero qualifying local entities was treated as empty. Missing records, unselected engines, and poor-quality runs were treated as missing or ineligible.
- Eligibility
- Spatial analysis required at least four eligible locations, at least one neighbor pair, 90% record coverage, 90% usable surface coverage, and no explicit incomplete run. Temporal analysis required at least 80% location matching, then used 3 through 10 day intervals.
- Comparison periods
- The primary comparison excluded the May core rollout. Each comparable report, query, and surface series received an earlier median and a later median. Series were summarized within reports, and reports received equal influence.
- Uncertainty
- Intervals came from repeatedly resampling whole reports with a fixed random seed. Two standard paired statistical checks, a Wilcoxon signed-rank test and a sign test, also tested whether the direction of change was consistent across reports.
- Data quality
- Two independent extractions reproduced identical spatial scores for 584 overlapping report runs. A known collection outage created a roughly two week gap in April, and no values were estimated across it. Two report level reads failed.
| Surface | Reports | Valid or usable locations | Coverage or CID quality |
|---|---|---|---|
| Mobile organic | 1,078 | 6,577,036 | 98.97% valid |
| Desktop organic | 243 | 1,585,271 | 99.48% valid |
| Mobile classic 3 pack | 1,078 | 6,638,360 | 99.77% CID |
| Mobile Maps top 10 | 1,061 | 6,561,514 | 99.97% CID |
| Desktop classic 3 pack | 244 | 1,590,916 | 100.00% CID |
| Desktop Maps top 10 | 206 | 1,543,014 | 99.81% CID |
Tests that can narrow the mechanism
Research and update records
- Google Search Status Dashboard. May 2026 core update.
- Hawkins, Joy. The Dynamic SERP Update: How the May 2026 Core Update Changed SEO. Sterling Sky, September 2, 2026.
- Google. Personal Intelligence in AI Mode in Search. January 22, 2026.
- Google. A New Era for AI Search. May 19, 2026.
- Hardawar, Devindra. Google Shoves More AI Into Search, Including a Dynamic Search Box and Agentic Features. Engadget, May 19, 2026.
- Google Search Help. Manage Personalized Recommendations in Search Services.
- Google Search Help. How AI Mode Uses Personal Intelligence and Memory.
- Google Search Status Dashboard. June 2026 spam update.
- Google Search Status Dashboard. Ranking history.
- Webber, William, Alistair Moffat, and Justin Zobel. A Similarity Measure for Indefinite Rankings. ACM Transactions on Information Systems, 2010.
- Schwartz, Barry. Google Search Ranking Volatility Hits August 1st and Heats Up Into August 3rd. Search Engine Roundtable, August 3, 2026.
- Schwartz, Barry. Google Search Ranking Volatility Heats Up Into August 5th and 6th. Search Engine Roundtable, August 6, 2026.
- Local Search Forum. Odd Rankings: Tools vs Human Results. Discussion started June 23, 2026.
- De’Land, Schieler. Same Keyword, Different Google – The Dynamic SERP Personalization Theory. Service Lifter, June 12, 2026.
- Ward, Christian J. Rank Is Too Volatile to Be the Whole Story. Be Datable, September 2, 2026.
- Hawkins, Joy. “Position #1 on Google is officially DEAD.” LinkedIn discussion, September 2026.