Instagram Tests 90-Minute Videos: What Brands Should Know

Instagram has spent years positioning itself as the home of short, vertical video. Now it is testing the opposite. According to reporting from SocialDay and other outlets, Instagram has begun testing horizontal videos of up to 90 minutes, called Features, with select US creators. For brands, the news matters less as a product launch and more as a signal about where creator partnerships may be heading.

What is actually happening

Here is what has been reported, kept to what the coverage supports.

• The format: Features are horizontal videos up to 90 minutes long.
• The rollout: the US test began October 26, following a trial in Brazil. Reporting suggests it could expand to tens of thousands of US creators.
• Distribution: Features attach to Reels and may be recommended on Explore.
• Clips as Reels: a separate test lets participants link up to eight Reels clips to a longer video, so short content can lead people to the full piece.
• Money: there is no ad revenue share for long-form at launch. Instagram is funding some creators directly, and an Instagram representative said it is "too early to share a long-term monetization model."
• Context: Instagram head Adam Mosseri said, "For a long time, I have been against doing long-form video on Instagram." Instagram is also reportedly testing an option that opens the app directly to Reels.

Everything above is a test. Formats, eligibility and economics can change or disappear. I would plan around the direction, not the details.

Why this matters for brands

Creators will start asking for more

If a creator you work with is in the test, or expects to be, their incentives shift. Some will be funded directly by Instagram and may want different deal structures. Others will want to build long-form work that your brand could either sponsor or be left out of. Expect conversations about integrations, series and ownership to get more complicated.

Long-form changes what a sponsorship is

A 15-second Reel integration and a 40-minute video are different products. A long video can hold a real story, a demonstration or a conversation. It can also bury a brand mention in the middle where few people reach it. The strength of the format is depth, and brands that use it well will behave more like producers than advertisers.

Measurement is unsettled

With no ad revenue share at launch and no confirmed monetization model, there is no established benchmark for what a long-form view on Instagram is worth. I would be cautious about any agency or creator who promises predictable returns right now.

The Reels-to-Features loop is the strategic bit

The "Clips as Reels" test is, to me, the most interesting detail. It suggests Instagram wants short clips to act as the front door for longer videos. That mirrors how many brands already think about content: a long anchor piece, cut into many small ones. If you already work this way, Instagram may become another place to host the anchor. If you don't, this is a reason to start. The idea is covered in more depth in my post on building a brand content series.

What to do now

1. Do not rebuild your Instagram plan. This is a test with select creators. Keep your current Reels and Stories cadence going.
2. Ask your creator partners where they stand. Are they in the test? Are they being funded by Instagram? Do they plan to make Features? Their answers affect availability, exclusivity and pricing conversations later.
3. Review contract language. Check usage rights, exclusivity, disclosure and content-length terms. Agreements written for short posts may not address a 60-minute video. This is a good moment to revisit whether you run a creator program or one-off influencer posts, because long-form rewards ongoing relationships.
4. Inventory your long-form assets. Webinars, interviews, product demos, panels and events you already own are candidates to test if the format opens to you. If the footage exists, you lose little by being ready.
5. Plan the cut-down. For any long video, decide in advance which moments become clips, what each clip's job is, and where it points. Design for the loop, not for the upload.
6. Set a small, honest test budget. If you try a sponsored long-form piece, define success before you start, whether that is watch behavior, saves, follows or traffic, and be clear that the data is exploratory.
7. Keep YouTube in the picture. Long-form already has an established home. Do not move what works there on the strength of a test.

My take

Instagram's own leader has been publicly skeptical of long-form, so this is a notable reversal of direction, even in test form. I read it as a response to where attention and creator income are going, and as an attempt to keep creators from taking their best work elsewhere.

For brands, the opportunity is real but premature. The sensible stance is curiosity with discipline: learn how your creators are thinking, get your contracts ready, and build assets you can use anywhere. Measurement frameworks for the format do not exist yet, so anything you try should be labeled as a test. If you need help structuring that, my guide to a social media strategy framework is a good place to start.

Sources: Reporting from SocialDay and Social Media Today roundups, plus TechRepublic's coverage of Instagram Features and The Next Web's report on the 90-minute test.

---

Want a second opinion on your social?

I offer a free social audit for brands: send me your handles, I review what you're doing, and we talk through what to keep, cut and try next. Request your free social audit →

LinkedIn's AI Slop Crackdown: What Brands Should Do Now

LinkedIn has been telling us for a while that it wants fewer generic posts in the feed. Now we have specifics. According to reporting from SocialDay, LinkedIn VP of Engineering Tim Jurka said the platform's "AI slop" classifiers now check every post that is distributed beyond the author's immediate network. If you run a company page or ghostwrite for executives, this changes how you should plan content.

What LinkedIn announced

Here is what has been reported so far, and I'm flagging where details come from the platform versus secondary coverage.

• Scope: the classifiers review posts before they travel past first-degree connections.
• Accuracy: the system reportedly has 94% precision at detecting generic content.
• Definition: LinkedIn describes slop as "generic and/or repetitive posts that added no real insight."
• Consequence: flagged posts lose reach in recommendations beyond first-degree connections. They are not removed.
• Impact: reporting cites about 40% fewer views for content classified as slop.
• Member signals: people can now flag a post with a "Seems like AI slop" option, and LinkedIn plans to notify authors.

The detail I would underline is the definition. LinkedIn's Chief Product Officer said "AI and slop are not the same thing." The target is not AI-assisted writing. The target is content that adds nothing. That is a quality standard, and it is one good brands should already be holding themselves to.

What it means for brands and executives

Distribution is now a content-quality test

Your network will still see your post. The risk is that it never reaches anyone beyond it. For company pages that rely on employees and followers to extend reach, and for executives building authority with people outside their immediate circle, that outer ring is the entire point of posting.

Templated thought leadership is the biggest exposure

The posts most likely to be caught are the ones I see constantly: a bold opening line, a numbered list of obvious tips, a tidy lesson, and a question at the end. Whether a person or a model wrote it, it reads the same as a thousand others. If your content calendar is built from prompts like "write a LinkedIn post about leadership," you are producing exactly what a classifier is trained to recognise.

Repetition counts too

The definition includes "repetitive." Brands that cross-post identical copy from executives, employees and the company page, or recycle the same talking points every week, may be easier to flag than they realise. I would treat that as an inference from LinkedIn's wording, not a confirmed rule about how the system treats duplicate content.

Reader flags add a second layer

With a "Seems like AI slop" option in the hands of members, your audience now has a direct way to signal that a post wasted their time. Expect your most visible, most polished-but-empty posts to be the first to get that reaction.

What to do now

1. Audit your last 20 to 30 posts. Mark each one as either "adds something only we could say" or "could have been written by anyone." Compare reach between the two groups. This is the fastest way to see whether you have a problem. The same method works as part of a broader social media audit.
2. Require a point of view in every post. Before anything is published, someone should be able to finish the sentence: "The insight here is..." If they cannot, the post is not ready.
3. Anchor posts in proprietary material. Customer questions, internal lessons, real decisions, product trade-offs, event observations and original data you own. Slop is generic by definition, so specificity is your protection.
4. Keep AI as a tool, not an author. Use it for outlines, edits, summaries and research. Do not use it to generate the opinion. Executives' posts in particular should start from something they actually said or did.
5. Vary structure and voice. Retire the repeated hook formulas. Mix short observations, longer essays, documents and video, and let each person on your team sound like themselves.
6. Brief executives and employee advocates. Make sure everyone who posts on the brand's behalf understands what is changing, and that copy-paste amplification of company posts is not a strategy.
7. Watch the numbers that matter. Track reach beyond your followers and engagement quality, such as substantive comments, rather than raw impressions alone. If you report upward, frame these as executive-level KPIs so a dip caused by lower-quality posts is visible early.
8. Prepare for author notifications. LinkedIn plans to notify authors. When that arrives, treat it as feedback and log which posts get flagged so you can learn from patterns.

My take

I don't see this as a punishment for using AI. I see it as LinkedIn raising the cost of being forgettable. For brands that already publish with a clear point of view, the effect could be positive, because there should be less noise competing with them. For brands that have been scaling output without scaling thinking, it is a correction that was coming.

Some details are still unclear, including exactly how the classifier weighs different signals and how appeals or notifications will work. I would not rewrite your whole approach on the strength of one report. I would, however, start measuring now so you have a baseline before the changes fully land.

Sources: Reporting from SocialDay on Jurka's announcement; see also Social Media Today's coverage of LinkedIn's approach to AI slop and Campaign's report on the "Seems like AI slop" button.

---

Want a second opinion on your social?

I offer a free social audit for brands: send me your handles, I review what you're doing, and we talk through what to keep, cut and try next. Request your free social audit →