Meta's Deprecation Dates for Instagram Impressions vs Views

What changed between Instagram impressions and views?

Meta settled Instagram impressions vs views on January 21, 2025, when one changelog entry introduced the views field and deprecated impressions on both media and user insights. The deprecation took effect immediately for v22.0 and above, and Meta set April 21, 2025 as the date it would reach every remaining version. Three sibling fields went out in the same entry: plays, clips_replays_count and ig_reels_aggregated_all_plays_count, all on media insights.

The date matters more than the definition here. Instagram's Help Center describes views without saying when it arrived or what it displaced, so anyone reconciling a reporting break has to work backwards from the developer changelog. That is the only Meta surface that carries the dates, and it states them in two lines:

Metric Deprecations. Applies to v22.0+. Will apply to all versions April 21, 2025.

A fourth field had already gone a few months earlier. On October 2, 2024, Meta wrote that "the video media metric video_views will no longer be supported," applying to v21.0 and above, with the same note that it would apply to all versions on January 8, 2025. So a team that upgraded once in late 2024 and again in early 2025 lost four different counting fields in four months.

When did each Instagram metric actually stop working?

Here is the full sequence from Meta's Instagram Platform changelog, with the version each date applied to.

Date	What Meta changed	Applies to
October 2, 2024	video_views no longer supported	v21.0 and above
January 8, 2025	Same video_views removal	All versions
January 21, 2025	views introduced; impressions, plays, clips_replays_count and ig_reels_aggregated_all_plays_count deprecated	v22.0 and above
April 21, 2025	Same four deprecations	All versions

Two of those rows describe the same change twice. That is the pattern Meta uses for metric removals: the field breaks first on the newest version, then everywhere about three months later. If your integration was pinned to an older version, the January date passed without anything visible happening and the April date is the one that broke your dashboard.

Which field replaced impressions and plays?

views replaced all four of the deprecated counting fields at once, on both the media and user insight edges.

Deprecated field	Where it lived	Replacement
impressions	Media and user insights	views
plays	Media insights	views
clips_replays_count	Media insights	views
ig_reels_aggregated_all_plays_count	Media insights	views
video_views	Video media	views

The consolidation is the point of the change. Before it, a reel produced a plays number, a replays number, an aggregated all-plays number and an impressions number, and none of them answered the same question. After it, one field covers feed posts, reels and stories.

Before and After the Cutover

Before January 21, 2025

• impressions counted views on both media and user insights
• plays, clips_replays_count and ig_reels_aggregated_all_plays_count each counted reels separately
• video_views was already gone on v21.0 and above, since October 2, 2024

After April 21, 2025

• views covers feed posts, reels and stories on every version
• impressions, plays, clips_replays_count and ig_reels_aggregated_all_plays_count are deprecated everywhere
• impressions still returns data only for media created on or before July 1, 2024, and only on v21.0 and older

Why does a request for impressions still return numbers?

Because Meta scoped the shutdown by media creation date, not by request date. The changelog note reads:

API requests with the impressions metric will continue to return data for media created on or before July 1, 2024 for v21.0 and older. API requests made after April 21, 2025 for media created on or after July 2, 2024 will return an error.

Read that carefully, because it produces a result most people do not expect. On v21.0 and older, impressions is not dead. It is frozen. Posts published on or before July 1, 2024 keep returning impression counts indefinitely, while anything published from July 2, 2024 onward errors out. One archive query across a two-year date range will therefore succeed for the first half of the range and fail for the second half, from a single field name in a single request.

That is also why "impressions still works for us" and "impressions is gone" are both true statements from different teams. The team reporting on a 2023 evergreen library sees data. The team reporting on last week sees an error. Meta does not publish the error code or message for that failure, so the only reliable test is to request impressions for a known post created after July 2, 2024 and read what comes back.

What does Meta say a view counts?

Meta gives two different definitions of views in two of its own reference pages, and neither one states a playback threshold.

The media insights reference defines it as "Total number of times IG Media has been played on Instagram," and lists it for FEED, REELS and STORY media types. The user insights reference defines the same field as "The number of times your content was played or displayed. Content includes reels, posts, stories."

The gap between "played" and "played or displayed" is not cosmetic. A static photo in the feed is displayed and never played. Under the media-level wording, it is unclear what registers; under the user-level wording, a display is enough. Meta ships both sentences and reconciles neither.

The second gap is the one that trips up anyone benchmarking Instagram against another network: Meta publishes no minimum playback duration for a view. There is no documented number of milliseconds, no documented number of seconds, and no documented percentage of the player that has to be on screen. Numbers circulate claiming a one millisecond threshold, and none of them trace back to a Meta page. Treat the threshold as undocumented rather than as one millisecond.

For contrast, the IAB and MRC viewable ad impression standard does set a bar, and it is far higher than anything Instagram documents: at least 50 percent of pixels in view for one continuous second for display, and two continuous seconds for video. Instagram's views field is not measured against that standard and Meta never claims it is.

Which Instagram field should you use now?

Use views for volume and reach for people. Those are the two that survived the cleanup and they answer different questions.

Field	Edge	Meta's definition
views	Media and user insights	"The number of times your content was played or displayed"
reach	User insights	"The number of unique accounts that have seen your content, at least once, including in ads"
impressions	Media and user insights	Deprecated; documented but returns data only for media created on or before July 1, 2024 on v21.0 and older

Note the third row. impressions is still described in Meta's media insights reference, still listed among the available metric names, and still used inside the definition of another live field: the reels reach description says reach "is different from impressions, which can include multiple views of a reel by the same account." Meta deprecated the field in the changelog and left it standing in the reference documentation. Do not read its presence in the metric list as a sign that it works.

Meta's media insights reference also lists total_views, crossposted_views and facebook_views alongside views. Those are separate fields with separate scopes, so read their definitions before you sum anything.

If you want the general definition of an impression across networks, and how the platforms that still report impressions count them, our guide to social media impressions covers that ground. This page is only about which Instagram field exists and when it stopped.

What this means for your historical reporting

Meta never published a conversion factor between impressions and views, and the two fields were never defined against each other. An impressions number from March 2024 and a views number from March 2025 are two different measurements of two different events, so charting them as one continuous line invents a trend that the data does not support. Break the series at the cutover and label it.

For anyone pulling numbers programmatically, the practical work is small: request views instead of impressions, keep reach where you were already using it, and stop treating pre-cutover and post-cutover totals as comparable. AdaptlyPost reads current Instagram metrics through the social media analytics feature, and the Instagram analytics tool page covers what is available per account. If you are building against the API yourself, the Instagram posting API page describes the publishing side of the same platform.

Frequently Asked Questions

When exactly did Instagram deprecate impressions?

January 21, 2025 for v22.0 and above, and April 21, 2025 for all remaining versions. Meta published both dates in the Instagram Platform changelog under the heading Metric Deprecations. The same entry introduced views and deprecated plays, clips_replays_count and ig_reels_aggregated_all_plays_count.

Why am I still getting impressions data from the Instagram API?

Your request is hitting v21.0 or older and asking about media created on or before July 1, 2024. Meta kept impressions returning data for that specific combination. Ask the same version for media created on or after July 2, 2024 and the request returns an error.

How many milliseconds of playback does an Instagram view require?

Meta does not publish a threshold. Neither the changelog nor the media and user insights references state a minimum playback duration, a minimum on-screen time or a minimum visible percentage for views. Any specific millisecond figure attributed to Instagram is unsourced.

Is an Instagram view the same as an impression?

No, and Meta does not define one in terms of the other. impressions counted times the media was seen, views counts times it was played or displayed, and Meta published no conversion between the two. Keep the two series separate in reporting rather than merging them.

Does the IAB and MRC viewable impression standard apply to Instagram views?

No. That standard requires at least 50 percent of pixels in view for one continuous second for display and two continuous seconds for video. Instagram's views field is not measured against it, and Meta makes no claim that it is.

What replaced the plays metric on Instagram Reels?

views. Meta deprecated plays, clips_replays_count and ig_reels_aggregated_all_plays_count on the same date as impressions and folded all of them into the single views field, which covers feed posts, reels and stories.

How many metrics did Meta deprecate in the January 21, 2025 changelog entry?

Four: impressions, plays, clips_replays_count and ig_reels_aggregated_all_plays_count. impressions applied to both media and user insights, while the other three applied to media insights only. Meta introduced views in the same entry to absorb all four counts into a single field.

Is video_views the same field as the new views metric?

video_views was a separate field on video media, deprecated on October 2, 2024, three months before views existed. Meta's own replacement table lists video_views alongside impressions, plays, clips_replays_count and ig_reels_aggregated_all_plays_count, with views as the field that now covers all five.

What is the difference between views and reach on Instagram?

views counts the number of times content was played or displayed, while reach counts unique accounts that saw the content at least once, including in ads. The post recommends views for volume and reach for people, since those are the two fields that survived Meta's cleanup. reach lives only on user insights, while views appears on both media and user insights.

Did Meta publish a formula for converting impressions into views?

Meta never published a conversion factor between the two fields, and neither one was defined in terms of the other. Charting a March 2024 impressions number against a March 2025 views number as one continuous line invents a trend the data does not support. The post recommends breaking the series at the cutover date and labeling each half separately.