Data reference¶

Observed on 18 October 2026, on one consenting account. A field's presence in that response does not guarantee availability for every account, Reel or period.

QueryResult.get_data(*fields) starts at payload["data"]. It resolves canonical names such as 1$xig_fetch_media(id:$media_id) and numeric aliases such as 1fvalue. Compatible fragments of the same node are combined; conflicts are rejected. Explicit null is returned as None; a missing or ambiguous path is an error.

Detailed Reel statistics¶

BaselReelDetailsQuery → client.reel_insights(reel_url)

reel = client.reel_insights(reel_url)
plays = reel.metrics.get("play_count")
skip = reel.metrics.get("skip_rate")
skip_value = skip.get("value") if isinstance(skip, dict) else None

Fields	Observed data
`play_count`, `like_count`, `comment_count`, `repost_count`, `save_count`, `reshare_count`	Playback and interaction counts
`skip_rate`, `reshares_rate`, `likes_rate`, `comments_rate`, `reposts_rate`, `saves_rate`	Metric objects with values and, depending on the field, deltas, labels and trends
`average_watch_time`	Average watch time; the SDK does not assume its unit is seconds
`follow_count_from_media`, `reach_count`, `viewer_count`	Follows, reach and viewer metric objects
`views_over_time`	Raw nested playback time series
`content_view_source_breakdown`, `viewer_distribution`, `top_ten_badge`	Sources, viewer distribution and a badge
`reels_retentions`	Retention array, exposed separately as `reel.retention`

The SDK normalizes top-level metric aliases 1fvalue/1fdelta to value/delta. Nested structures remain unchanged. Retention guide.

Public Reel metrics¶

BaselPublicInsightsQuery → client.public_insights(reel_url)

response = client.public_insights(reel_url)
metrics = response.get_data("xig_fetch_media", "basel_media_public_insights")

Returned interaction counts and rates. This block has no retention curve. The name “Public” does not prove unauthenticated access: only an owned Reel was checked.

Paginated Reels¶

BaselAccountLevelReelsQuery → client.reels_page()

page = client.reels_page(first=3, timeframe="LIFETIME", order_by="BY_DATE")
for item in page.items:
    print(item["id"], item.get("permalink"))
if page.has_next_page:
    next_page = client.reels_page(first=3, after=page.next_cursor)

Raw path: xig_user_by_igid_v2 → basel_insights → user_reel_dicts → edges[].node. Observed node fields include id, permalink, taken_at, caption, image_versions2, clips_metadata, is_promoted, is_active_trial, coauthor_producers and basel_media_level_insights.

Pagination uses page_info.has_next_page and end_cursor. Two pages of three Reels were checked without overlapping IDs. account_reels() returns the raw response.

Parameter	APK values
`timeframe`	`LIFETIME`, `ONE_WEEK`, `TWO_WEEKS`, `ONE_MONTH`, `ONE_YEAR`
`order_by`	`BY_DATE`, `BY_VIEWS`, `BY_LIKES`, `BY_COMMENTS`, `BY_REPOSTS`, `BY_SHARES`, `BY_SAVES`, `BY_FOLLOWERS`, `BY_SKIP_RATE`
`sort_direction`	`ASCENDING`, `DESCENDING`, `None`

The live test used LIFETIME, BY_DATE, DESCENDING, and skip_follow_count=False. Other combinations were recovered statically. Server page-size limits are unknown.

Account statistics¶

BaselAccountLevelInsightsQuery → client.account_insights()

response = client.account_insights(period="LAST_30_DAYS", include_time_series=True)
metrics = response.get_data("xig_user_by_igid_v2", "basel_insights")
views = response.get_data("xig_user_by_igid_v2", "basel_insights", "reels_views", "value")

Observed metrics: reels_views, total_reach, followers_net_growth, like_count, comment_count, repost_count, reshare_count, save_count, viewer_count. Metric objects may include a value, delta and time_series.

Access flags: has_access_to_audience_insights, has_access_to_demographic_insights. can_see_organic_insights is in the user node.

Periods: LAST_7_DAYS, LAST_14_DAYS, LAST_30_DAYS, LAST_60_DAYS, LAST_90_DAYS, LAST_180_DAYS. Only LAST_30_DAYS was exercised live.

Audience statistics¶

BaselAudienceInsightsQuery → client.audience_insights()

response = client.audience_insights(period="LAST_30_DAYS", client_timezone="Europe/Moscow")
ages = response.get_data("xig_user_by_igid_v2", "basel_insights", "audience_age_distribution")
countries = response.get_data("xig_user_by_igid_v2", "basel_insights",
                             'audience_top_locations(breakdown_type:"COUNTRY")')

The response had age groups, countries, cities, gender groups, total followers, follower growth/follows/unfollows with time series, and most-active follower times. popular_topics_preview was empty. The account-statistics periods above also apply.

Delta updates¶

XIGBaselAccountLevelReelsDataDeltaQuery → client.reels_delta()

response = client.reels_delta(["NUMERIC_MEDIA_ID"])
items = response.get_data("xig_basel_multi_media_insights_query")

Returns a list of id/basel_media_level_insights nodes with media ID, interaction counts, follows and a badge. Only a single-ID request was tested; multi-ID limits and responses remain unverified.

Comment analysis¶

XIGBaselInsightsCommentsQuery → client.comment_insights()

response = client.comment_insights(reel_url, sort_order="recent")
analysis = response.get_data("xig_fetch_media", "ai_comment_insights")

APK values: recent, popular. The live recent request returned HTTP 200 and ai_comment_insights=null. A populated analysis was not verified. Null does not mean the Reel has zero comments.

Kits¶

BaselKitsQuery → client.kits()

response = client.kits(after=None)
connection = response.get_data("xig_basel_kits", "kits")
edges = connection["edges"]

The persisted query uses first:50; the SDK supplies after. The live response had edges=[] and has_next_page=false. Populated items and further pages were not verified. No Kits were created for testing.

Alternative Reel insights¶

ReelInsightsQuery → client.extended_reel_insights()

response = client.extended_reel_insights(
    reel_url,
    bloks_versioning_id="8dc50e9193c9b70a3f9dbcc25fc5f444bbac27bb56a8c1359ede836b8069c4b8",
)
insights = response.get_data("ig_reel_insights")
curve = insights["retention"]["retention_curve"]

The SDK assembles the observed NDJSON response and combines compatible canonical fragments. Curve fields are timestamp_seconds and 1fpercentage; here 1 means 100%. Some other percentage fields, including country percentages, already use percent units. Do not multiply every percentage in this response by 100.

The explicit Bloks version comes from the researched APK and may need to change. This curve can differ from BaselReelDetailsQuery. Its use by the Edits UI was not established. Method contract.