Social listening tools in a platform takedown evidence log

What to take away

What the Online Safety Act 2023 requires of regulated platforms

The Online Safety Act 2023 places duties on regulated user-to-user and search services to tackle illegal content and protect children. The Act also includes transparency provisions that allow Ofcom to require information about how platforms moderate content. You can read the full text on the Online Safety Act 2023 page.

For a UK brand monitoring team, the Act matters because it creates a formal framework for platform reporting duties. Platforms must assess risks, take proportionate action, and keep records that show what they did. Those records are the raw material for a takedown and removal record.

Policy responsibility sits with the Department for Science, Innovation and Technology. The Department for Science, Innovation and Technology - GOV.UK page sets out its remit, which includes online safety regulation. Ofcom is the regulator that enforces the platform duties.

In practice, regulated platforms include large social media services, messaging apps, and search engines. Smaller services may have lighter duties, but many brands deal with platforms that fall squarely within the regime. That means your monitoring team can expect more structured reporting than before.

The Act does not give brands a direct right to demand individual takedown records. Instead, it gives Ofcom powers to require transparency, and platforms publish some information voluntarily. Your requests should therefore be framed as part of a commercial or monitoring relationship, not as a statutory entitlement.

Ofcom transparency expectations and the records behind them

Ofcom transparency expectations are the regulator's way of turning platform reporting duties into usable information. Ofcom can require platforms to produce transparency reports, and it has published guidance on what those reports should cover. The aim is to show how platforms handle harmful content and what happens after a report is made.

Ofcom says the duty to publish transparency reports applies only to providers of certain regulated services on its public register of “categorised services” (Ofcom’s transparency reporting statement).

Ofcom has set out its transparency reporting expectations for regulated services, and it can issue formal information notices under the Online Safety Act 2023 when published reports are not enough. In practice, the data you want may sit in two places: the platform's published transparency report, and the fuller returns it makes to Ofcom.

For brands, the key records behind those expectations include: the number of reports received, the number actioned, median time to action, and the outcome. Platforms may also report on automated detection and appeals. These figures help you judge whether a platform is responsive to brand safety concerns.

Ofcom has also signalled that it will name platforms that fail to meet transparency expectations. That reputational pressure gives brand monitoring teams a lever. If a platform is slow to provide removal records, you can point to Ofcom's published expectations and ask for the same data.

The records are not always public. Some are shared with Ofcom under information requests, and only summaries are published. Your team may need to ask a platform directly for its removal records, citing your own monitoring needs and the platform's reporting duties.

When you ask, be specific. Request the number of removals for the harmful content categories you care about, the time period, and the policy basis for each removal. That makes the data comparable across platforms and over time. It also makes your evidence log more credible.

Takedown reporting a UK brand monitoring team can reasonably request

A UK brand monitoring team can reasonably request takedown and removal records that relate to its own content, its ads, or its brand terms. You are not asking for the platform's entire moderation database. You are asking for records connected to your monitoring scope.

Start with a clear request. State the time period, the platforms covered, and the harmful content categories involved. Ask for a summary table plus individual case records where possible. If the platform refuses, ask for its standard transparency report and note the gap.

Each large platform also runs its own reporting route, and some publish removal data you can cite directly. X and YouTube take reports through their Help Centres, Reddit through a report form on each community, and YouTube publishes a Transparency Report on Community Guidelines enforcement while Reddit publishes an annual transparency report. Those public documents are a useful fallback when a case-level request is refused.

Here is a worked example of a request a brand team might send to a platform:

Please provide removal records for content matching our brand terms between 1 January 2026 and 30 June 2026. Include the harmful content category, the date of removal, the policy applied, and whether the content was reinstated. If individual records cannot be shared, provide aggregate counts and median time to removal.

That request is reasonable because it is bounded, tied to your own brand, and asks for data the platform already collects. It also aligns with platform reporting duties under the Act.

You can also ask for reinstatement data. If content is removed and later restored, that is a signal about moderation accuracy. Reinstatement rates help you judge whether a platform's initial decisions are reliable.

Keep a record of every request and response. This becomes part of your brand monitoring evidence log. If a dispute arises later, you can show what you asked for, when, and what you received.

For more on how different platforms produce different evidence, see our article on listening on X, Reddit or YouTube, chosen by the evidence.

Harmful content categories and how they map to listening queries

Harmful content categories are the buckets platforms use to classify removals. They typically include hate speech, harassment, misinformation, violent content, and adult content. The exact labels vary by platform, but the underlying categories overlap.

Labels differ by platform. X's rules use categories such as hateful conduct and harassment; YouTube's Community Guidelines cover hate speech, harassment and cyberbullying, and medical misinformation; Reddit's content policy covers hateful content and harassment. Map the platform's own label to your query before you ask for a record.

Your listening queries should map to those categories so you can compare platform removals with your own monitoring. If your queries use different language, you will struggle to match records. Build a mapping table before you request data.

The table below shows a simple mapping you can adapt. It links common harmful content categories to example listening queries and the kind of removal record you might request.

Use the same categories in your evidence log. That way, when you report to your board or a client, you can show a consistent picture across platforms.

Some categories are harder to monitor than others. Misinformation, for example, may not be removed at all, but labelled or downranked. Your listening queries should capture labels as well as removals.

For a fuller treatment of how to define metrics that withstand challenge, see our piece on metrics defined completely.

Where platform duties meet UK GDPR when you log removal evidence

Removal evidence often contains personal data. A username, a profile link, or a quoted post can identify a person. That means UK GDPR and the Data Protection Act 2018 apply when you collect, store, and share that evidence.

The ICO has published guidance on how online safety and data protection interact. The Online safety and data protection | ICO page explains the balance between safety duties and data rights. You should read it before building a logging process.

Key principles apply. You need a lawful basis for processing, such as legitimate interests. You should minimise the personal data you keep, store it securely, and set retention periods. You should also be able to respond to subject access requests.

When you request removal records from a platform, you are receiving personal data. The ICO's Data sharing | ICO guidance helps you handle that flow lawfully. It covers data sharing agreements and the need for a clear purpose.

Your evidence log should separate personal data from aggregate counts. Keep aggregate figures for reporting, and restrict access to case-level records. That reduces risk and makes your log easier to defend.

If you share removal evidence with a client or agency, put a data sharing agreement in place. Document the purpose, the categories of data, and the retention period. This is good practice and aligns with ICO expectations.

For a wider view of how UK law reaches social listening, including data protection and platform rules, see our guide to the layers of UK law.

Building a takedown evidence log that survives scrutiny

A takedown evidence log is a structured record of what you monitored, what you reported, what was removed, and what happened next. It should be consistent, time-stamped, and traceable. If challenged, you should be able to show the source of every entry.

Use a checklist to make sure each entry is complete. At minimum, a defensible log records:

Keep the log in a system that records who changed what and when. A spreadsheet can work for small teams, but an audit trail is better. If you use social listening tools, check whether they can export removal signals with timestamps.

Review the log regularly. Look for gaps, such as missing removal dates or unclear categories. A log that shows its gaps as plainly as its numbers is more credible than one that hides uncertainty.

Train your team on the log. Everyone should use the same categories and the same fields. Inconsistency is the quickest way to undermine an evidence log.

Finally, align the log with your retention policy. Do not keep personal data longer than necessary. Aggregate the data for long-term reporting, and delete case-level records when the purpose has passed.

Using social listening tools in a takedown evidence log

Social listening tools are not moderation systems, but they can capture removal and reinstatement signals. You need to ask for the right features. Not every tool records when a post disappears or reappears.

Vendors UK brand teams commonly shortlist for this work include Brandwatch, Talkwalker, Meltwater, Sprout Social, Hootsuite and Brand24. Put the same questions to each of them and compare the answers side by side, because removal and reinstatement coverage varies more between products than sales material usually admits.

Start by asking your vendor how they detect removals. Some tools re-check URLs and flag when content is no longer available. Others rely on platform APIs that do not report removals. The difference matters for your evidence log.

Ask for reinstatement tracking too. If a post is removed and later restored, you want to know. That signal helps you judge platform accuracy and your own reporting decisions.

Check how the tool handles harmful content categories. Can you tag a post with the same category labels you use in your log? If not, you may need to map them manually.

Ask about data export. You need to get removal signals out of the tool and into your log. Look for timestamps, platform identifiers, and category fields in the export.

A workable export schema is date_reported | platform | post_id | category | removal_date | reinstated | source_url. If your tool cannot produce those columns, plan to add them by hand and note the gap in the log.

Ask about UK GDPR compliance. Where is the data stored? Who can access it? What is the retention period? Your vendor should be able to answer these questions clearly.

Finally, test the tool on a small set of known removals. See whether it catches them and how quickly. That test tells you more than any sales claim.

Common questions