You started with one export. Then came the X API to fill in the posts the export left out, a news provider for full articles, a Reddit source, maybe a second listening tool for coverage the first one lacks. Every quarter there's a new gap, and a new integration to close it.
If that sounds familiar, this piece is for you. We'll cover what five listening APIs let you export, why the stitching keeps growing, and how teams get out of it without ripping out the tools their analysts depend on.
Limits checked against each platform's public developer docs on November 11, 2026.
Why the stitching keeps growing
Most of these teams aren't only reporting anymore. They're building their own product: their own models and classifiers, their own metrics, their own view for customers and internal teams who want more than a dashboard can give. The listening tools have quietly become data sources for it.
The catch is that every export API has limits, and they've been tightening. X used to come out with the post text, and now it's an ID. Reddit used to come out, and now mostly doesn't. So each gap means another source to stitch in, and each source brings more work, more upkeep and more specialized skills: data engineering, a different query language for every provider, enrichment that holds up across all of it. None of that is the product you set out to build.
It's a common frustration. In the Social Intelligence Lab's State of Social Intelligence 2026 report, 64% of the 330 practitioners surveyed named tool limitations or integration constraints as the biggest thing holding back their impact. The report also notes teams collecting through a listening tool, then running the analysis in an LLM to get past the platform's own limits.
The short version
- X comes out as post IDs you rehydrate through the X API, on every platform that documents it.
- Facebook, Instagram and LinkedIn content mostly stays in. News comes out as a snippet.
- Volume follows your package: monthly caps, sampling on large exports, and call limits that range from 5 per second to 30 per 10 minutes.
- Some data is licensed for internal analysis, not for building a product on top of.
Many of these limits are set by the networks and publishers behind the data, not by the tools.
Social listening API limits compared
How the five compare at a glance. The details for each platform follow. Sprinklr isn't in the table: its developer portal says rate limits are set by contract, and it doesn't publish what listening data can be exported. For a team planning around those limits, that means finding out after you've signed.
| Platform | X content via API | News | Other sources held back | Published volume or rate limit |
|---|---|---|---|---|
| Talkwalker | IDs + sentiment (paid add-on for full) | Truncated | Facebook, Instagram, LinkedIn, raw Reddit and TikTok, Weibo | 1M results per export task; 60 to 240 search calls/min |
| Meltwater | IDs only | Title + summary | Factiva | 30,000 docs/month included; 2M per export |
| Brandwatch | IDs + a few fields | 256-character snippet | LinkedIn text, Reddit (URL only), some data packs | 30 calls per 10 min |
| Pulsar | Some fields masked | Not stated | Raw display and real-time streaming not supported | 5 req/sec, 100 results per page |
| YouScan | IDs only | Not stated | Reddit, Quora, some small sources | Not published |
Already stitching several of these together? Jump to an alternative approach. For data APIs beyond these five, see our guide to social media monitoring APIs.
Talkwalker API: bulk export tasks, limited X and Meta content
Talkwalker has one of the more complete developer portals in the category, with a Search API, a Streaming API and export tasks built for pulling a project's history.
What you can take out
- Project results through export tasks, delivered to a collector you set up. Each task tops out at 1,000,000 results, and Talkwalker suggests splitting long periods into smaller tasks, for example one month at a time.
- Live results through the Streaming API.
- Aggregated metrics through the Histogram API, including for sources whose raw content can't be exported.
What stays in
- X: exports carry the post ID, the author ID and Talkwalker's sentiment score. You rehydrate the rest through the X API, unless you buy Talkwalker's X data add-on. X exports are also capped at 1.5M documents per month per account.
- Facebook and Instagram: post content can't be exported by API or any automated means. Aggregated metrics can.
- Online news and TVEyes: truncated content only.
- Not exported at all: LinkedIn, raw Reddit and raw TikTok (metrics only), Weibo and other Chinese sources, and some review sites depending on provider agreements.
Limits teams run into
- Search API: 240 calls/min outside a project, 60 inside one. Histogram API: 60 and 30.
- Project data access stream: 5 calls/min. Export task creation: 40 calls/min.
- Usage is metered in credits, so a large backfill has a cost to plan for.
Meltwater API: monthly caps, sampling and internal-use terms
Most Meltwater packages now include some API access, with larger volumes sold as an add-on package. Exports run off saved searches you build in the app or through the API.
What you can take out
- Mentions from saved searches through one-time and recurring exports, in JSON or CSV templates.
- Real-time mentions through Data Streams, if your package includes them.
- Analytics on earned media, plus a small search endpoint meant for dashboards rather than bulk pulls.
What stays in
- X: exports return the post ID and author ID only. Meltwater's docs walk you through rehydrating the content yourself with the X API.
- News: title and summary, subject to publisher restrictions. Factiva content isn't provided.
- Use: Meltwater's own help content describes API data as meant for internal use, not for building or reselling commercial software.
Limits teams run into
- Included API access: 30,000 documents per month, 50 exports per month and 12 months of history. Paid packages set their own numbers by contract.
- Exports aren't de-duplicated, so overlapping exports count twice against the monthly total.
- Exports are processed one at a time. Each tops out at 2,000,000 documents, and a larger one is randomly sampled down to fit.
- Time window: 12 months per export or analytics request unless the contract says otherwise. Export creation is limited to 20 calls/min.
Brandwatch API: 30 calls per 10 minutes, X and Reddit limited
Brandwatch's Consumer Research API covers queries, mentions, categories, tags and charts. Its publishing and engagement APIs are separate, with their own keys.
What you can take out
- Mentions from your queries, with full text through a dedicated endpoint where the source allows it.
- Aggregated chart data and counts.
- Metadata edits going the other way: you can tag, categorize and annotate mentions through the API.
What stays in
- X: full text and most metadata are stripped. A short set of fields remains, including the post ID, language, sentiment, country and tags. You fetch the post body from the X API.
- Online news: a 256-character snippet.
- LinkedIn: no snippet and no full text.
- Reddit: URL only, due to licensing.
- Data packs: some add-on data packs carry their own restrictions, so the API can show less than the app does.
Limits teams run into
- 30 calls every 10 minutes per API client. Use cases that need more go through your account manager.
- Paging with page and pageSize alone stops at 10,000 mentions (up to 5,000 per page). Beyond that you page through history with a cursor.
- No webhooks on Consumer Research, so new mentions are polled. Brandwatch recommends running requests one at a time rather than in parallel.
Pulsar API: approved use cases only, no raw display
Pulsar offers a GraphQL API, and access is approved case by case. Its docs say everything in the API is also available in the platform, either as charts or as exports.
What you can take out
- Results and metrics from your searches, in JSON, including aggregated performance such as sentiment, impressions and visibility.
- Searches can be created and launched through the API.
What stays in
- X: the only source Pulsar flags with platform limits. Some fields are masked to comply with data providers' policies.
- Use cases: Pulsar doesn't support streaming data in real time into your own app, or collecting and displaying its data in raw form.
Limits teams run into
- Each use case goes through an approval form and Pulsar's product team before access is turned on.
- 5 requests per second, with up to 100 results per page. No overall cap on the number of calls.
YouScan API: paid add-on, no Reddit or Quora
YouScan sells API access as a paid add-on.
What you can take out
- Topic metrics such as mention counts, sentiment breakdowns and word clouds.
- Full mention text, where the source allows it.
What stays in
- X: mention text, full text, URLs and author details are removed. The post ID and profile ID remain.
- Reddit and Quora: not available through the API.
- Some smaller sources: restricted, with no full list published.
Limits teams run into
- Each API key belongs to one user and only sees the topics that user can see in the app.
- No rate or volume limits are published, so ask before planning a large backfill.
Where the limits bite
One limit is easy to miss in the tables: the query lives in the tool. Exports run off saved searches and projects inside each platform, so the data's shape follows the tool's model rather than yours.
None of this is a criticism. These tools are built to give analysts answers, and their APIs reflect that. They weren't designed to be the data layer for another product.
What stitching really costs
Each fix makes sense on its own. Together they add up. We talk to teams running one or two extra sources, and teams running seven or eight, with more than $1M a year in licenses across them. Every source has its own access rules, pricing, rate limits, schema and changes, and someone has to keep all of it working. Plenty of teams started building six months earlier and found out it's not just connecting to an API.
You end up spending your team on working around other people's limits instead of on the product you're building.
An alternative approach
Datastreamer is the orchestration layer these teams have been trying to build themselves. We help insight and intelligence platforms build on web and social data with a managed sourcing, enrichment, and integration platform.
That makes it the piece underneath the stitching, not a fifth tool on the list. It fits into what you already have. You add, swap or drop sources and enrichments to get to what you need, without the extra noise. There's no rip and replace: the tools your analysts depend on can stay, and their exports can come in as one source among the others.
- Sources: connect to providers that already hold the platform access, through one platform. Datastreamer picks the best provider for each job (we call this Auto Sources), and you can bring your own provider keys. The full list of sources is on our integrations page.
- Enrichment: choose what you need, such as sentiment or language, and apply it across every source.
- Integration: data lands where your team works in one consistent schema, whichever provider it came from. You own the jobs, the pipelines and the results.
- Cost control: volume and cost caps, with spend broken down by job, client or project. Pricing is usage-based, and it's on our pricing page.
Short on the people for it? The specialized work, from provider queries to schemas to enrichment, is the part we take on. Teams that want more hands-on help can get it on the same platform.
Where to start
Teams rarely move everything at once. They usually start with one specific gap: a data source they can't get, or a schema problem that keeps breaking an integration. Once that runs through Datastreamer, the next gap goes the same way.
A few of the starting points we've seen:
- A missing source. A European media monitoring company moving off its listening tool already had a text news provider. Its first gap was broadcast: radio, TV and podcasts, with about two years of history behind it. Meanwhile, its listening tool's export runs into the same pipeline as everything else.
- A source that disappeared. When its LinkedIn provider dropped the source, a team running real-time client alerts checked a week of LinkedIn Auto data against its own records, then moved straight to production.
- Query and schema friction. An analytics team replacing a listening tool with multiple raw providers found the hardest part was provider query language that worked nothing like the one they knew. Easily managed by the Datastreamer Jobs system.
If you have a gap in mind, book a call with our team and we'll look at your setup together. If you're still weighing your options, our Build vs Buy page lays out what each path involves.
FAQ
Can you export X (Twitter) posts through Brandwatch, Talkwalker or Meltwater?
You get post IDs, plus a few fields such as sentiment on some platforms. The post text comes from rehydrating the IDs through the X API, or from Talkwalker's paid X data add-on.
Can you export Facebook and Instagram content through a listening tool's API?
Talkwalker's docs say post content can't be exported by API or any automated means. Aggregated metrics can.
How much data can you export from the Meltwater API?
Included API access covers 30,000 documents and 50 exports a month. Paid packages set their own limits by contract. A single export tops out at 2,000,000 documents and is sampled above that.
What is the Brandwatch API rate limit?
30 calls every 10 minutes per API client. Use cases that need more go through your account manager.
Can you combine several listening tools and data providers in one Datastreamer-hosted pipeline?
Yes. Teams often bring a listening tool's export in as one source alongside data providers, then apply the same enrichment and schema across all of it. That's the layer Datastreamer runs.
Sources
All checked November 11, 2026.
- Talkwalker: API restrictions, export task principles, past data export, project data access
- Meltwater: usage limits, rehydrating X posts, exporting your data, data use note, FAQs
- Brandwatch: best practices and rate limits, retrieving mentions, data restrictions
- Pulsar: GraphQL API FAQs, Pulsar API
- YouScan: how to use the YouScan API
- Sprinklr: developer portal
- Social Intelligence Lab: State of Social Intelligence 2026