Consulting and recruiting
Social listening data: why agencies usually can't relicense it
By SourceX Editorial · Reviewed by Noah Loul ·
Short answer
Agencies usually can't relicense social listening data because they never held the rights: posts come from platforms under developer terms, pass through a listening vendor under a subscription license and remain the authors' content. Those terms govern. What an agency can usually keep is its own work, such as codeframes, analyst reports and aggregate findings.
Key takeaways
- Social listening data reaches an agency through a chain of licenses, and each layer narrows what can be done with it.
- Platform developer terms increasingly bar AI training on their content; X changed its terms in June 2025.
- Platforms license their own data for AI directly, which is one reason they restrict others from doing so.
- Analyst-authored reports, taxonomies and aggregate findings are usually the agency's; raw posts are not.
- A public post is not free to copy, store or license just because anyone can read it.
Why can't agencies relicense social listening data?#
Agencies usually can't relicense social listening data because the rights to the raw posts never passed to them. The agency subscribes to a listening tool; the tool's provider receives posts under platform developer terms or through licensed data providers; the people who wrote the posts keep their own rights. Each layer grants access for analysis, not a right to redistribute.
That is why a large archive of exported mentions, however carefully coded, is rarely an asset the agency can license to anyone. The posts belong to a chain of other parties, and the subscription almost always forbids passing them on.
The authors add a separate layer. A post can be expressive work protected by copyright, and it often carries personal data: a handle, a photo, a location or a story about the writer's health or family. Even where a contract allowed onward use, the agency would still need to consider those rights, which is a burden that survey data collected under the agency's own consent does not carry in the same way.
The chain of terms behind every mention#
Each link in the chain sets its own limits, and the narrowest link controls what the agency can do. Reading only the listening tool's subscription misses the platform terms that sit behind it.
| Layer | Who sets the terms | What it usually restricts |
|---|---|---|
| Platform | The social network's developer and API terms | Storage, redistribution, AI training and display of deleted content |
| Data provider | A licensed reseller or the platform's own data program | Onward licensing, resale and use beyond the subscriber's analysis |
| Listening tool | The vendor's subscription agreement | Export volume, retention after cancellation and sharing outside the account |
| Client | The agency's MSA and statement of work | Use of client briefs, brand data and results |
| Authors | Copyright and privacy law | Copying expressive content and processing personal data |
Platforms now price and restrict AI use of their content#
Platforms have moved to control AI use of their content directly. On April 18, 2023, Reddit announced premium paid access for third parties that need large-scale or commercial use of its Data API, a change its CEO linked to companies using Reddit data to train AI models. In its Form S-1, filed February 22, 2024, Reddit disclosed that it had entered into data licensing arrangements in January 2024.
X took the restrictive route. In June 2025 it updated its Developer Agreement to bar developers from using the X API or X Content to fine-tune or train a foundation or frontier model, as TechCrunch reported on June 5, 2025. When a platform licenses its content for AI itself and bars others from doing so, a downstream agency has no room to offer the same content.
What an agency can usually keep and reuse#
The line runs between the posts and the agency's own work. The table sorts common assets from a listening program; check each one against your actual subscription and client terms.
Labels deserve a note. An analyst's sentiment or topic labels are the agency's work, but while they sit attached to post text they travel with the platform's restrictions. A taxonomy or codebook described on its own, without the posts, is a different and much cleaner record.
| Asset | Keep and reuse internally | License to a third party |
|---|---|---|
| Raw post exports | Only as the tool subscription allows | Generally no |
| Post IDs, URLs and author handles | Only within the subscription | Generally no |
| Vendor-computed sentiment and topic scores | Within the subscription | Generally no |
| Agency codeframes and topic taxonomies | Yes, if built without client confidential material | Possibly, after a rights check |
| Analyst-written reports and commentary | Yes, subject to client confidentiality | Only with client permission |
| Aggregate counts and trend findings | Usually yes, subject to vendor terms | Check vendor terms; often restricted |
| Agency labels applied to posts | Within the subscription | Generally no while tied to post text |
Mistakes that create exposure#
Most problems come from treating social data like the agency's own survey data, which it collected under consent and contracts it could read and negotiate. These patterns are worth stopping before they become a dispute with a vendor, a platform or a client.
- Assuming a public post is free to copy, store and reuse.
- Keeping bulk exports on shared drives after the tool subscription ends.
- Sending raw exports to clients in a form the subscription does not allow.
- Using exported posts to train or fine-tune an internal model.
- Supplementing a listening tool with scraping that breaches platform terms.
- Describing client work as built on licensed data when the license covers analysis only.
Questions to put to your listening vendor#
The subscription agreement is the document an agency can actually read and negotiate, so ask the vendor to answer these points in writing before renewal. The answers tell you what can stay in your project folders and what must be deleted when the relationship ends.
- Which platforms' terms flow down to us, and where can we read them?
- What may we export, in what volume, and for which purposes?
- Must exported content be deleted when the subscription ends, and on what timeline?
- May we share exports or excerpts with clients, and in what form?
- Are we allowed to use exported content to train or test any model, including internal tools?
- Which outputs, such as aggregate counts or charts, may we keep and publish after cancellation?
Illustrative: an insights agency reviews its listening archive#
Illustrative: a fictional insights agency has run social listening for consumer brands for years. Analysts exported mentions to spreadsheets for coding, and those files now sit in project folders beside survey data, reports and the agency's own topic taxonomy.
When the CEO asks whether the archive could be licensed, counsel reads the listening tool subscription and finds that it limits exports to the subscriber's internal analysis and requires deletion of stored content on cancellation. The raw exports are deleted under the agency's retention policy.
The topic taxonomy, the coding guidelines and analyst reports with client details removed are kept as firm records, documented separately from any post content. Those records, together with the agency's survey studies, become the scope worth discussing.
How SourceX handles social data#
SourceX generally treats third-party social content as out of scope, because the agency is not the rights holder. In the Rights step of the SourceX five-step transaction, platform-sourced posts are separated from the agency's own records, such as taxonomies, codebooks, survey studies and project records.
Only records the agency can show it controls move forward, documented in a SourceX Evidence Packet with their provenance, licensing rights and permitted use. The agency approves every step, and nothing is shared during the initial assessment.
Frequently asked questions
What if we collected posts ourselves rather than through a tool?
Collecting directly does not avoid platform terms. Scraping or API collection is governed by the platform's terms of service and developer agreement, which may forbid storage, redistribution or AI training. The authors' copyright and privacy rights also still apply.
Can we quote individual posts in client reports?
Often in limited form, but check the vendor subscription and any platform display rules that flow down to you. Quoting a handful of posts to illustrate a theme is different from handing over a file of mentions. Remove handles and personal details unless there is a clear reason to show them, and avoid quoting posts that were later deleted.
Is a client's own community or review data different?
It can be. A client may control content from its own forums, reviews or support channels under its own terms of use. That data sits in the client's rights chain, so any reuse depends on the client's terms with its users and its contract with the agency.
Are survey open-ends different from social posts?
Yes. Survey verbatims are collected by the agency or its sample providers under consent and contracts it can read and negotiate. Social posts arrive under platform terms the agency did not write. That difference is why survey archives are sometimes licensable and social exports rarely are.
Can we license a model trained on social data?
That carries the same restrictions in another form. If platform or vendor terms bar using content to train models, a model built on that content is likely affected as well. Ask counsel before building or offering any model trained on listening exports.
Sources
- On April 18, 2023, Reddit announced premium paid access for third parties needing large-scale or commercial use of its Data API, a change CEO Steve Huffman linked to companies using Reddit data to train AI models. Source
- In its Form S-1 filed with the SEC on February 22, 2024, Reddit disclosed that in January 2024 it entered into certain data licensing arrangements. Source
- In June 2025, X updated its Developer Agreement to bar developers from using the X API or X Content to fine-tune or train a foundation or frontier model, as reported by TechCrunch on June 5, 2025. Source
Related resources
See if your company qualifies
A short company assessment. No data uploads are needed.