The best MCP for market research is the one that answers the question you are actually asking. For demand sizing, use Rising Trends MCP, which returns absolute US monthly search volume per term. For live SERP and keyword data use DataForSEO, for scraping competitor pages use Firecrawl, for reviews and social at scale use Bright Data, and for public company numbers use SEC EDGAR. Two or three connected servers will cover almost any research brief. One will not.
Key takeaways:
- Pick by question, not by brand. Demand sizing, competitor traffic, customer voice and filings are four different data sources, and no single MCP server holds all four.
- Only one of the eight returns absolute search volume. Rising Trends MCP gives a real number per term, such as 246,000 US searches for "the loaded tea shop" in October 2026. Google Trends wrappers return a 0 to 100 index instead.
- Five of the eight have a usable free tier. Bright Data gives 5,000 requests a month, Tavily 1,000 credits, Firecrawl 1,000 credits, Exa $10 of credits that reset monthly, and SEC EDGAR MCP is open source against a free government API.
- Demand for this is real and new. Searches for "mcp server" went from 260 a month in November 2024 to 60,500 in October 2026 (early estimate), and "mcp connectors" is up 511% over the year.
- Setup is a URL, not a build. Most of these now ship a hosted endpoint you paste into Claude or ChatGPT once. Budget ten minutes, not an afternoon.
Let's get into it.
What an MCP server actually does for a research brief
An MCP server is a standard connector that lets Claude or ChatGPT call someone else's database directly, mid-conversation. Without one, the assistant answers market questions from training data that is months old and cannot cite a number. With one, it makes a real request, gets real rows back, and shows its working.
For market research that changes one specific thing: the assistant stops estimating. Ask how big a category is and you get a figure with a date attached, not a guess shaped like a figure.
The demand curve underneath this is steep and recent.
Two years ago almost nobody searched for this. In the Rising Trends database, "mcp server" sat at 260 monthly US searches in November 2024 and reached 60,500 in October 2026 (early estimate), with "mcp connectors" up 511% over the year and "mcp apps" up 400%. The tooling went from nothing to normal inside eight quarters, which is why most of the roundups you will find are already out of date on prices.
How we picked and checked these
Every server below was checked on 14 November 2026 against its own repo, package registry or pricing page. Nothing here is quoted from another roundup.
- Real and installable. It has a published package or a hosted endpoint you can connect today.
- Maintained. The last release is inside twelve months. Anything older was dropped.
- Answers a research question. Not a generic developer utility that happens to fetch a URL.
- Priced in public. If the price is "contact sales", that is said plainly rather than guessed at.
Freshness, read from npm and PyPI on the day: Apify 0.21.10 (13 Nov 2026), Firecrawl 3.40.0 (12 Nov 2026), Bright Data 4.8.0 (10 Nov 2026), Exa 3.4.3 (3 Nov 2026), DataForSEO 3.1.3 (2 Oct 2026), SEC EDGAR 1.1.0 (20 Aug 2026), Tavily 0.2.22 (5 Aug 2026).
That map is the whole argument of this page. Work down it and the shortlist picks itself.
The 8 best MCP servers for market research, compared
| Server | Answers | Data source | Free tier | Paid from |
|---|---|---|---|---|
| Rising Trends | How big is demand, and is it growing | US search volume, Amazon, web traffic | No | $97 one time |
| Exa | What has been written about this | Own semantic web index | $10 credits monthly | $4 per 1,000 searches |
| Bright Data | What do buyers say, at scale | 1,200+ site scrapers | 5,000 requests monthly | Usage based |
| DataForSEO | Who ranks, and for what keywords | Live SERPs and keyword data | No | $50 minimum top-up |
| Firecrawl | What is on a competitor's page today | Any URL, as clean markdown | 1,000 credits monthly | $19 a month |
| SEC EDGAR | What public filings report | SEC EDGAR, XBRL financials | Open source, free API | Free |
| Apify | Anything needing a custom scraper | Marketplace of hosted actors | $5 of usage monthly | $19 a month |
| Tavily | A cited answer you can footnote | Search with citations | 1,000 credits monthly | $30 a month |
1. Rising Trends MCP
Answers: how big is demand, is it growing, and did anyone catch it early.
This is the one that gives you a number instead of an index. Every tracked term carries absolute US monthly search volume with five years of monthly history, growth over three months to five years, and the date the term entered the database. That last field is the one analysts underrate: it tells you how early the signal was available, which is the difference between research and hindsight.
Connect it at the Rising Trends MCP page, which takes one URL in Claude or ChatGPT.
The tools are find_trends, search_trends, get_trend, get_trends, list_mega_trends, get_mega_trend, find_products for Amazon categories with price and monthly sales, find_websites for traffic growth, find_apps, and list_categories. Searching is free and does not count against the quota.
Price: Pro, $97 as a one-time payment for lifetime access, which includes the connector, the REST API and a 100-trend watchlist with monthly alerts. There is no free tier, and that is the honest trade: every other server on this list lets you try before you pay.
What it does not do: no reviews, no social posts, no filings, and no live Google Trends index or regional breakdown. For those, use Bright Data, SEC EDGAR or DataForSEO.
Example conversation (Rising Trends MCP in Claude, data through October 2026, October is an early estimate)
You
I am sizing the skincare category. What is growing fastest with real volume behind it?
find_trends({ "category": "Skincare", "timeframe": "1y", "min_volume": 5000, "sort": "growth", "limit": 5 })
Claude
Five fastest-growing Skincare terms with at least 5,000 US monthly searches, October 2026 (early estimate):
- "reale actives": 135,000 a month, up 233% over three months. Added September 2026.
- "ascorbic acid serum": 74,000 a month, flat over three months but up from 140 a month a year ago.
- "clerivy": 33,100 a month, from 170 in August 2026. Added October 2026.
- "target beauty studio": 14,800 a month, up 410% over three months.
- "more naturals": 12,100 a month, up 124% over three months.
Three of these did not exist as searches twelve months ago. "ascorbic acid serum" is the steadiest: it stepped up to 74,000 in July 2026 and has held there for four months.
You
Show me one you flagged early, so I can see how much warning the data gives.
get_trend({ "name_or_slug": "the loaded tea shop" })
Claude
"the loaded tea shop" entered the database on 22 March 2026, when it was running at 27,100 US searches a month. The monthly series since: 3,600 in November 2025, 8,100 in January 2026, 27,100 in March, 60,500 in May, 165,000 in July, 201,000 in September and 246,000 in October 2026 (early estimate). That is up 5,491% over the year and still up 49% over the last three months. Categories: Beverages, Vitamins and Supplements.
Every number in both answers came back from the two tool calls. That is the point of connecting a demand server: the assistant quotes the database instead of its memory.
Seven months of warning, in a category most people only noticed at the top of that curve. If you would rather do this step by hand first, our guide to using Google Trends for market research covers the free version of the same job, and trend analysis covers how to read a curve like the one above.
2. Exa MCP
Answers: what has been written about this, including the pages Google buries.
Exa runs its own semantic index, so it searches by meaning rather than keyword match. For a literature sweep at the start of a brief, that is the difference between twenty relevant pages and two hundred near-misses. The tools are web_search_exa and web_fetch_exa by default, with web_search_advanced_exa and an agent_run multi-step research tool available, all listed in the Exa MCP server repo.
Setup: add https://mcp.exa.ai/mcp as a remote server, or install the Claude plugin. MIT licensed, version 3.4.3 published 3 November 2026.
Price: per Exa's pricing page, $10 of credits on signup, resetting on the first of each month with no card required. Then $4 per 1,000 lite or instant searches, $7 for auto or fast, and $12 to $15 for deep search.
Weakness: it is a search engine, not a database. It will find you an article claiming a market is worth $4 billion. It will not tell you whether that number is sound.
3. Bright Data MCP
Answers: what customers actually say, in volume.
This is the customer-voice entry, and it is the one most research roundups miss. Thirty tools, including six that find and run any of 1,200+ pre-built scrapers for Amazon, Reddit, LinkedIn, TikTok, Instagram, YouTube, X and Crunchbase, returning clean JSON rather than HTML you have to parse. Every request routes through unblocking infrastructure, so pages that reject ordinary scripts come back normally. The full tool list is in the Bright Data MCP repo.
For review mining or share-of-voice work on a brand, this does in one prompt what used to be a week of scraper maintenance.
Setup: npx @brightdata/mcp, or the hosted endpoint. MIT licensed, version 4.8.0 published 10 November 2026.
Price: 5,000 requests a month free, renewing monthly, no card required. Beyond that it is usage based, and Bright Data does not publish a flat rate on its main pricing page.
Weakness: scraped social text is evidence of what people say, never a fact about the world. Treat it as qualitative input, and do not let an assistant turn a hundred Reddit comments into a market share estimate.
4. DataForSEO MCP
Answers: who ranks for the money terms, and what the keyword landscape looks like.
The competitive-search entry. Live SERPs, keyword data and backlinks, which together tell you who is winning the organic traffic in a category and what they are winning it with. Pair it with a demand server: DataForSEO tells you who ranks, Rising Trends tells you whether the thing they rank for is growing.
Setup: npx dataforseo-mcp-server@latest, with OAuth on the HTTP transport or login and password on stdio. Apache-2.0, version 3.1.3 published 2 October 2026.
Price: pay as you go with a $50 minimum top-up and balances that do not expire. On DataForSEO's pricing page, SERP requests start at $0.0006, the Keywords Data API at $0.0012 a task, and DataForSEO Labs at $0.012 a task plus $0.00012 an item.
Weakness: no free tier, and the pricing is endpoint-by-endpoint, so your first month is hard to forecast. If you only need keyword volumes, our rundown of the Google Trends API and its alternatives covers cheaper routes.
5. Firecrawl MCP
Answers: what is on a specific page right now.
Point it at a competitor's pricing page, changelog or careers page and it returns clean markdown an assistant can actually reason over. Research briefs are full of this: thirty tabs of competitor pages that someone has to read and summarise. This is the server that does that part.
Setup: npx firecrawl-mcp, or the hosted endpoint. Version 3.40.0 published 12 November 2026, the most recently updated server on this list.
Price: per Firecrawl's pricing page, a free tier of 1,000 credits a month, roughly one credit per page. Hobby $19 a month, Standard $99, Growth $399, Scale $749.
Weakness: it reads pages, it does not find them. Pair it with Exa or Tavily to build the URL list first.
6. SEC EDGAR MCP
Answers: what public companies report, in their own filings.
The entry that separates a market research stack from a marketing stack. It connects filings, financial statements and insider trading data from SEC EDGAR, with XBRL figures returned at exact numeric precision rather than parsed out of a PDF. If the category you are sizing has a public pure-play in it, segment revenue from a 10-K beats every analyst estimate you can buy.
Setup: pip install sec-edgar-mcp, or run the published Docker image from the SEC EDGAR MCP repo. AGPL-3.0, version 1.1.0 uploaded 20 August 2026. It is not affiliated with the SEC, and the underlying EDGAR full-text search is public.
Price: free. The server is open source and the EDGAR API carries no charge.
Weakness: US public companies only. Private competitors, which is most of them in a young category, are invisible here.
7. Apify MCP
Answers: the scraping job no off-the-shelf tool covers.
A marketplace of hosted scrapers that the assistant can search and run. When your brief needs something specific, such as every listing in one marketplace category or a particular forum, there is usually an actor for it already. Version 0.21.10 published 13 November 2026, MIT licensed.
Price: per Apify's pricing page, a free plan with $5 of platform usage a month, then Starter $19, Scale $199, Business $999.
Weakness: actor quality varies, because anyone can publish one. Check the actor's own run history before you build a brief on it. We cover this stack in more depth in our guide to the best MCPs for finding business ideas.
8. Tavily MCP
Answers: a search result you can footnote.
Tavily returns search output built for citation, with search, extract, map and crawl tools. For a deck that has to survive a legal read, having the assistant hand back sources with every claim is worth more than raw index quality.
Setup: the remote endpoint at https://mcp.tavily.com/mcp/ with your API key, or npx tavily-mcp. MIT licensed, version 0.2.22 published 5 August 2026, the oldest release here but still inside the twelve-month cut.
Price: on Tavily's pricing page, Researcher is free with 1,000 API credits a month, Pay As You Go is $0.008 a credit, and Project is $30 a month for 4,000 credits.
Weakness: it overlaps heavily with Exa. Connect one of the two, not both.
Which servers should you actually connect?
Connecting all eight is the common mistake. Every server adds tool definitions to the model's context and more chances for it to pick the wrong one. Three is the practical ceiling for a single brief.
- Sizing a consumer category: Rising Trends for demand, Bright Data for reviews, Firecrawl for competitor pages.
- Competitive intelligence on a known rival: DataForSEO for their rankings, Firecrawl for their site, SEC EDGAR if they are public.
- Scoping something brand new: Exa to find the literature, Rising Trends to check whether anyone is actually searching for it.
- Writing something that gets published: Tavily for cited sources, Rising Trends for the numbers.
Rising Trends appears in three of those four because demand sizing is the step the others cannot do: the Trends API and MCP return absolute monthly volume per term, where every search-based alternative returns a relative index. If your work is more about watching categories move than sizing one, the market research use-case walkthroughs show the same data without a connector, and the trending products list and trending websites list are browsable without one too.
How to add an MCP server to Claude or ChatGPT
The hosted servers all follow the same three steps.
- In Claude: open Settings, go to Connectors, add a custom connector, and paste the server URL.
- In ChatGPT: open the connectors page, add a custom MCP server, paste the URL and choose OAuth.
- Sign in and allow access, then ask your first question in plain words. The assistant picks the tool.
For Claude Code or Cursor, create an API key in the service and send it as an Authorization: Bearer header alongside the URL. Setup for Rising Trends specifically is on the Rising Trends MCP page.
Copy-paste prompts
- "Using Rising Trends, find the fastest-growing trends in my category over the last year with at least 5,000 monthly searches, and tell me when each one entered the database."
- "Size this category for me: give me absolute monthly search volume for the five main terms, then check which competitors rank for them."
- "Pull the last three years of segment revenue for this public company from its filings, then compare it to search demand for the category."
What an MCP server will not do for you
Worth saying plainly, because most roundups skip it.
It will not design your research. The assistant picks a tool and reports what comes back. Deciding which question matters, which comparison is fair and which number is load-bearing is still the job.
It will not replace a survey. None of these servers asks anyone anything. Search volume tells you how many people looked for a term, not why, not who, and not whether they would pay. Behavioural data and stated preference are different evidence, and a panel is still the only way to get the second.
And it will not fix a bad source. An assistant that scrapes a wrong number repeats it confidently. Check anything that is going to drive a decision, the same way you would check a figure from a junior analyst. The method behind that check is in our guide to identifying market trends.
FAQ
Which MCP is best for market research?
There is no single best one, and any page that names one is usually selling it. Rising Trends MCP is the strongest demand-sizing server because it returns absolute US monthly search volume per term, but it holds no reviews, no filings and no SERP data. Connect it alongside one scraping server and one search server.
Can an MCP server give me real search volume?
Only some. Most search-trend servers wrap Google Trends, which publishes a 0 to 100 relative index with no absolute numbers behind it. Rising Trends MCP returns the actual monthly figure, such as 60,500 US searches for "mcp server" in October 2026. We compare the options in our roundup of the best trends MCPs and skills.
Which market research MCPs are free?
SEC EDGAR MCP is fully free and open source. Bright Data gives 5,000 requests a month, Tavily 1,000 credits, Firecrawl 1,000 credits, Apify $5 of usage, and Exa $10 of credits that reset monthly. DataForSEO and Rising Trends are paid only.
Is there an official Similarweb or Crunchbase MCP?
Not a first-party one worth connecting as of November 2026. The entries in the public MCP registry under both names are third-party wrappers of varying quality. Bright Data's scraper set covers Crunchbase company pages, and for traffic estimates the options are in our list of Similarweb alternatives.
How many servers should I connect at once?
Two or three. Each one adds tool definitions the model has to choose between, and accuracy drops as that list grows. Connect what the current brief needs, then swap rather than stack.
Can an MCP replace a market research agency?
For desk research, it covers a large share of what a junior analyst does in a week: demand figures, competitor pages, filings, review sentiment. For primary research, panels and qualitative work, no. The two are different evidence, and the fastest briefs use both.
Want the data without the chat? Read our guide on how to identify market trends, or browse what is breaking out right now in our trending topics list.
Never miss a trend again
- One trend broken down with the numbers
- Who's making money, and what to build
- A website on the rise
- A breakout app
Join 5,000+ founders
