Stop copy-pasting from Google Trends into your AI. Connect Trends MCP and your AI assistant can pull consumer demand signals, cross-platform trend momentum, and competitive intelligence directly - across Google, TikTok, YouTube, Reddit, Amazon, Wikipedia, and news in one session.
Updated 2026-08-19
Market research with live trend data starts by asking whether demand for a category, brand, or concept is rising, flat, or fading across public platforms, then reading those curves before anyone commissions a survey. As of August 2026, one MCP or REST API covers 25+ platforms, stores weekly history of about five years, and scores each series on a 0-100 index so search intent, shopping demand, community talk, and news volume sit on the same scale. The free plan is 100 requests per month. Starter is $19 per month. A typical session pairs a multi-source growth call with a weekly history pull on the one or two series that still look unresolved.
A live session starts with a hypothesis keyword, pulls weekly history of about five years on a 0-100 index, then compares 3 month, 6 month, and 1 year growth across search, commerce, and community sources before anyone writes a TAM slide. The assistant stays in one conversation. There is no export hop into a spreadsheet unless a stakeholder asks for a table.
The first call is usually a growth query with several sources rather than a single Google Trends screenshot. Search shows information-seeking. Amazon shows shopping. Reddit shows enthusiast talk. News volume shows whether journalists have already arrived. Wikipedia page views catch information spikes that often precede coverage. When those series disagree, the research question changes. A short-form climb with flat Amazon demand is awareness without purchase. A Google climb with quiet Reddit is mainstream search without a community yet.
Weekly history comes second. It answers whether the latest growth window sits on a multi-year rise, a seasonal bump, or a cliff after a peak. Ranked and live top-trend feeds belong at the start of a greenfield scan, when the category name itself is still unknown. Analysts who skip history and only read a 7-day print often treat a news spike as a market.
Category sizing from trend series estimates whether interest is expanding, stable, or shrinking, then decides whether a survey or expert interviews are still worth the spend, because public curves do not replace a TAM model and only show which boxes in that model deserve money this quarter.
A practical sizing pass uses the same keyword on Google Search and Amazon, then a brand or SKU list for competitor tracking. Search growth without Amazon growth often means people are learning, not buying. Amazon growth without search growth often means shoppers already know the category and are comparing packs. Both can be real markets. They imply different briefs.
Five years of weekly points matter here because many consumer categories are seasonal. A 3 month rise that repeats every winter is not a new market. A 1 year rise that keeps its shape after the seasonal peak is closer to a structural shift. The 0-100 index makes those shapes comparable across sources. Absolute volume estimates, where they exist, sit beside the index so a small niche is not mistaken for a mass category just because the percentage looks large.
Each source answers a different research question, and mixing them without labeling the job produces tidy slides that contradict the evidence, because search, shopping, community talk, short-form video, news, Wikipedia, and site visits measure different stages of demand rather than one generic buzz score. Google Search is the default intent meter. Amazon search trends are the purchase-intent meter. Reddit trends are the enthusiast and early-adopter meter. TikTok is often the earliest consumer-facing awareness meter. News volume and sentiment are the media-attention meter. Wikipedia is the definition-seeking meter. Web traffic is the site-visit meter for named companies.
A product idea check follows a fixed order so the assistant does not wander. Search and Amazon first, to see if anyone is looking or shopping. TikTok and Reddit next, to see if a community already has language for the problem. Wikipedia after that, to see if people are trying to understand a term. Competitor brand search last, to see who already owns the name.
Channel breakdown is the same data read sideways. If TikTok and YouTube are rising while Google Search is flat, the category is still in discovery formats. If Google Shopping or Amazon is rising while social is flat, the job is conversion content and retail placement, not more awareness creative. Researchers who treat every source as "buzz" lose that split.
Trend APIs that sit inside assistants usually come first in a research stack because one MCP or REST connection can query 25+ platforms, return weekly history of about five years, and score each series on a 0-100 index before paid surveys or panel fieldwork begin. Survey platforms still own stated preference, concept tests, and price ladders. Panel and social intelligence suites still own respondent-level quotes and mention operations. The order of use is public signal, then paid sample.
Syndicated reports and dashboard suites remain useful when a team needs a formatted deliverable for a steering committee. They are slower when the question is "is this category still climbing this month, and on which channel." Copying charts into a chat window is the failure mode the MCP path is built to avoid. The assistant reasons over the same JSON a human analyst would otherwise paste by hand.
Cost is part of the workflow, not a footnote. As of August 2026, 100 requests per month on the free plan cover a weekly category scan for a small roster of keywords. Starter at $19 per month covers a heavier brief cycle. Survey fieldwork and enterprise listening contracts are a different budget line. Teams that quote competitor dollar amounts without a current vendor quote should check current pricing rather than reuse an old slide.
Trend-based research cannot replace a representative sample, a quality interview, or a unit-economics model, and treating a 0-100 index as a revenue forecast is the most common over-read, because public attention can rise while margins fall or while a scandal inflates the same keyword. A keyword can climb because of a scandal, a stock meme, or a one-week shipping delay.
History depth also has a bound. About five years of weekly points is enough to see several seasonal cycles and many product generations. It is not enough to reconstruct a decade-old category origin story. Sparse series on small terms swing hard in percentage terms. Growth calls should be read with the index level and, when present, an absolute volume estimate.
Repeatability is the reason to keep the workflow in tools rather than in screenshots. The same keyword, sources, and windows can be rerun next month. If the story changes, the market changed, or the last brief was a spike. That loop is the research product. The tools below are the calls that loop uses.
Tools for this workflow
get_trendsChart the full demand history for any product category, brand, or market concept - see whether consumer interest is in an early growth phase, peak, or structural decline.
get_trends(keyword='plant based meat', source='google search', data_mode='weekly')get_growthRun a multi-signal market research query in one call: compare Google Search consumer intent, Amazon purchase demand, Reddit community interest, and news coverage growth simultaneously.
get_growth(keyword='plant based meat', source='google search, amazon, reddit, news sentiment', percent_growth=['3M', '6M', '1Y'])get_ranked_trendsRank the fastest-growing topics in a category or sector to identify where consumer attention is building - the equivalent of a market sizing scan across live behavioral data.
get_ranked_trends(source='google search', sort='yoy_pct_change', limit=40)get_top_trendsDiscover what consumers are trending toward right now without a hypothesis - use this at the start of a market research session to surface the categories worth investigating.
get_top_trends(type='Google Trends', limit=25)FAQ