Pass source google search for a weekly keyword series, or type Google Trends for the live breakout list. The two strings are not interchangeable on get_growth.
Live data as of 2026-09-15
A google search vs Google Trends comparison on Trends MCP is two different tools. Source google search on get_growth or get_time_series returns a weekly 0-100 series for one phrase, with an estimated volume field when the pipeline has it. Type Google Trends on get_top_trends returns ranked breakout queries and takes no keyword. From the September 15, 2026 board (as_of 2026-09-15T06:01:30 UTC), rank 1 was iphone duo. The matching google search close was index 100.0 on 2026-09-12 after 0.0 on 2026-09-05, so 7D printed 999999.0. Rank 17 costco closed at index 78.0 with estimated volume 36,500,000, down 3.7% over 7D. Mixing Google Trends into get_growth is a 400.
The September 15, 2026 board is the split that still matters next month: live rank is a breakout list stamped that morning, while source google search is a weekly index that can sit near zero for years, then print 100.0 on one close. Omitting limit returned limit 25, count 25, offset 0, as_of_ts 2026-09-15T06:01:30.283039+00:00. Rank 1 was iphone duo. Rank 2 was iphone 18 pro max. Rank 3 was sydney sweeney sports ad. Rank 10 was income. Rank 11 was lady gaga. Rank 17 was costco.
| Token | Trends rank | google search close | 7D | 30D | 12M |
|---|---|---|---|---|---|
| iphone duo | 1 | 100.0 on 2026-09-12 | 999999.0 from 0.0 | 999999.0 from 0.0 | 9900.0% from 1.0 |
| iphone 18 pro max | 2 | 100.0 on 2026-09-12 | +316.67% from 24.0 | +614.29% from 14.0 | 9900.0% from 1.0 |
| sydney sweeney sports ad | 3 | 6.0 on 2026-09-12 | 999999.0 from 0.0 | 999999.0 from 0.0 | 999999.0 from 0.0 |
| ios 27 release date | 4 | 27.0 on 2026-09-12 | +285.71% from 7.0 | +440.0% from 5.0 | +2600.0% from 1.0 |
| income | 10 | 49.0 on 2026-09-12 | -12.5% from 56.0 | -23.44% from 64.0 | -43.02% from 86.0 |
| lady gaga | 11 | 39.0 on 2026-09-12 | +290.0% from 10.0 | +225.0% from 12.0 | -4.88% from 41.0 |
| charlie kirk | 16 | 2.0 on 2026-09-12 | +100.0% from 1.0 | 999999.0 from 0.0 | -98.0% from 100.0 |
| costco | 17 | 78.0 on 2026-09-12 | -3.7% from 81.0 | -3.7% from 81.0 | +6.85% from 73.0 |
Live rank is not a sort of the weekly index. sydney sweeney sports ad sat at rank 3 with close 6.0. costco sat at rank 17 with close 78.0 and estimated volume 36,500,000. income sat at rank 10 while falling on every window in this cut. charlie kirk sat at rank 16 with close 2.0, against a 12M baseline of 100.0 on 2025-09-13. Quote the board stamp and the weekly close. Do not treat rank 1 as the largest audience.
iphone duo and iphone 18 pro max both printed 100.0 on 2026-09-12. Only the second one had a real 7D baseline: 24.0 on 2026-09-05, growth +316.67%. iphone duo's 7D and 30D rows used 0.0 baselines, so growth printed 999999.0, and the handler set data_quality to low with a warning that more than 90 percent of the series is zero. 12M for both phrases compared 100.0 against 1.0 on 2025-09-13 and printed 9900.0%. 7D matched 1W exactly on iphone duo. 30D matched 1M exactly. Quote 100.0. Do not put 999999.0 in a model.
costco is the durable brand in this cut. get_growth printed recent_volume 36,500,000 on 2026-09-12, volume_estimated true, and omitted volume_growth because volume is derived from the trend value for this source, not an independent measurement. get_time_series printed the same close as index 78 with volume 36,452,261. Peak index 100.0 appeared on 2024-11-30 and again on 2025-11-29, both with volume 46,733,668. 7D and 30D both compared 78.0 against 81.0 (-3.7%). A rank-17 breakout can sit on a slightly down week of a large, seasonal series.
sort=rank_change with window=7d and limit 10 used the same as_of_ts, plus baseline_ts 2026-09-08T02:01:35.429763+00:00, then returned count 6. The largest climber was ocarina of time remake at rank 34, prev_rank 170, rank_change 136. None of those six rows sat inside the default top 25. window=1d against baseline_ts 2026-09-14T00:01:29.596449+00:00 returned count 10. fraud climbed 131 to 19 (rank_change 112). ios 27 release date climbed 67 to 4 (rank_change 63). offset 25 with limit 5 returned count 5: dario amodei at 26, rosh hashanah at 27, nintendo at 28, kj biermann at 29, tmobile at 30. The board continued past row 25 on this pull.
Two strings, two tools. REST uses mode get_growth or get_top_trends. MCP uses the matching tool names. The live board page is the Google Trends Trending Now API. Weekly keyword windows live on the Google Search data page. Window defaults live on get-growth. Pagination and limit are documented with get-top-trends.
{"mode": "get_top_trends", "type": "Google Trends"}
{"mode": "get_growth", "source": "google search", "keyword": "iphone duo", "percent_growth": ["7D", "30D", "12M"]}
{"mode": "get_time_series", "source": "google search", "keyword": "iphone duo"}
{"mode": "get_growth", "source": "google search, google news", "keyword": "iphone duo", "percent_growth": ["7D", "30D"]}
The MCP equivalents are get_top_trends(type="Google Trends"), get_growth(keyword="iphone duo", source="google search", percent_growth=["7D", "30D", "12M"]), and get_time_series(keyword="iphone duo", source="google search"). REST top_trends with type Google Trends and limit 3 on September 15, 2026 returned the same snapshot timestamp as the omitted-limit MCP pull, with count 3: iphone duo, iphone 18 pro max, sydney sweeney sports ad.
get_time_series for iphone duo returned 261 weekly rows from 2021-09-18 (index 0) to 2026-09-12 (index 100). There is no volume field on those points. Grain is weekly over about five years. The last non-100 close in the strip printed 0 on 2026-09-05. A run of index 1 appeared from 2026-03-28 through 2026-06-27, then the series went back to 0 until the 100.0 close. That is a sparse breakout phrase sitting on the same 261-point grid as costco.
get_time_series for iphone 18 pro max also returned 261 weekly rows from 2021-09-18 to 2026-09-12. Early years are 0. The first 1 printed on 2024-09-14. The series then climbed through 2026: 14 on 2026-08-15, 24 on 2026-09-05, 100 on 2026-09-12. That ramp is why 7D growth is a real percent for rank 2 and a zero-baseline sentinel for rank 1, even though both phrases closed at 100.0.
The comma-separated iphone duo call returned sources_requested ["google search", "google news"], sources_successful 2, and trend_analysis.uptrend_ratio 2/2. Google Search 7D was 999999.0 with 261 points and recent_date 2026-09-12. Google News 7D was +3233.33% (3.0 to 100.0) on the same recent_date, also 261 points. 7D matched 1W on both sources. 30D matched 1M on both sources. Google News 30D was +4900.0% from 2.0. volume_weighted_growth printed null, weight_coverage 0/2 sources. Same preset, same close date, different baseline math. The news index 100.0 is not a search-volume count.
get_time_series on google search, google news returned invalid_source with 16 names and no all. Multi-source strings belong on get_growth only. Omitting percent_growth on costco returned the default 12M window: 78.0 versus 73.0 on 2025-09-13, growth +6.85%, 261 points.
get_growth with source Google Trends and keyword iphone duo returned HTTP 400, error invalid_source, message 'google trends' is not a valid source., plus 16 valid sources: google search, google images, google news, google shopping, youtube, wikipedia, tiktok, reddit, amazon, news sentiment, news volume, app downloads, app rankings, steam, npm, python, all. The handler lowercased the rejected source. It did not map it to google search.
The inverse fold exists on the feed name, and it is wider than case. type google trends on get_top_trends returned HTTP 200 and echoed type Google Trends, with the same top five titles as the canonical label and as_of_ts 2026-09-15T06:01:30.283039+00:00. type google search and type Google Search also returned HTTP 200, echoed type Google Trends, and printed the same ranked queries (iphone duo, iphone 18 pro max, sydney sweeney sports ad at limit 3). REST type google search with limit 3 repeated that alias. llms.txt still describes feed labels as case-sensitive and matching exactly. The live handler accepted these Google-adjacent strings on September 15, 2026 and served the Trends board. Prefer the canonical Google Trends spelling in stored configs. Do not treat the alias as a second feed.
REST omitting type is a different 400. POST https://api.trendsmcp.ai/api with mode get_top_trends and limit 5, no type, returned HTTP 200 with envelope statusCode 400, error invalid_request, message The 'type' parameter is required, then a 24-name list that includes Google Trends. llms.txt still names that miss missing_parameter. Read statusCode inside the JSON, not only the HTTP line.
country GB with limit 5 on Google Trends returned HTTP 200, country GB, as_of_ts 2026-09-15T06:17:23.431574+00:00, count 5: iphone 18 pro max, sydney sweeney, iphone duo, oasis tickets, ios 27. That stamp is later than the default board's 06:01:30 snapshot. Rank 1 flipped from iphone duo to iphone 18 pro max. sydney sweeney lost the "sports ad" suffix. oasis tickets appeared on the GB list and not in the default top 25. Google Trends is on the country-supported list. The GB call fetched a separate board. Quote both stamps if a report mixes them.
get_top_trends MCP pricing is 1 credit per 10 requested rows, max 10 credits per call. REST counts 1 request per call. Free boards still cap at the top 10. The 25-row pulls on this page are the paid default, which is also what the handler returns when limit is omitted on a paid key. The keyword-window follow-up is the Google Search data page. This split stays inside one search product: a weekly phrase series versus live breakout rank.
FAQ