Instagram brand tracking: method

Starlecta reads the posts, Reels and Stories of the Turkish Instagram accounts it tracks. The universe is not all of Instagram, so absolute volumes are a lower bound.

• Followers, category and audience (gender, age, city) come from the profile data held for each tracked account.
• Plays, likes and follower counts are read on a stated date. Each weekly report prints that date as the day its counters were read.
• Timestamps are read as Turkey time (UTC+3). A weekly period runs Monday to Sunday.

The weekly reports of 19–25 October 2026 start from about 12 thousand tracked Turkish Instagram accounts. The tracked base widened to about 51 thousand on 24–26 October 2026.

• For the new accounts, posts and Reels were completed 90 days back.
• Their Stories exist only from the day each account was added to tracking.
• Story counts are therefore not directly comparable across weeks that straddle the widening. Weekly trends show direction, and early weeks may look low.
• A weekly change in a report is counted with the same matching in both weeks. A percentage change is given only on a base of 10 or more contents.

A brand is matched to a piece of content from the tags (@) of the brand and its sub-brand accounts, from hashtags, and from brand names in the caption text. The weekly ranking uses tag and link matching.

• A content is counted once for a brand, even when the tag, the mention and the link all appear, or appear several times.
• A content that names more than one brand is counted separately for each brand, so brand totals can add up to more than the number of contents.
• Posts from a brand's own accounts are not counted.
• Accounts that share sector news or discounts are flagged separately and left out of the overlap and example selections.
• The Panel view pages of a report count the way the comparison tabs of the panel count (catalogue matching), not with the report's own brand matching. Their totals differ from the report's tables by design.

• A brand name, tag or link in a post, Reel or Story is a mention. It is not evidence of a paid collaboration.
• Instagram's own Paid partnership label is a separate field, reported on its own.
• Ad disclosure is the share of posts and Reels that carry a #reklam, #işbirliği or #tanıtım tag. It includes the Ücretsiz #reklam tag used on gifted products. Contents noted as not an ad (reklam değil, reklam yok) or as bought by the account itself (kendim aldım) do not enter the rate. Stories are not part of it.

A report keeps the kinds of value in separate fields and never adds them into one number. A dash in a table means there is no such counter in that row.

• Measured: a counter read from the account data on a stated date, such as followers, likes, comments and Reel plays. Plays exist only for Reels. Contents with hidden like counts are left out of medians.
• Modelled: a value calculated by a stated model from measured inputs and labelled as an estimate. Story views and Story interactions are modelled this way (see Story view estimation). When an input is missing the value is left empty with a reason, not shown as zero.
• The reports also carry a reach estimate as a field of its own. Like the Story model, it is not a measured counter and is not added to one.
• Scores such as the Starlecta Quality Score (SQS) are calculated values with their own definitions in the FAQ.

• A weekly report gives the engagement rate (ER) of content as two separate medians: (likes + comments) per follower for posts, and (likes + comments) per play for Reels. The denominators differ, so the two are not comparable with each other.
• Each median states n, the number of contents entering it. Contents with hidden like counts are excluded.
• The account engagement rate used in the panel and in scoring is a different quantity: (likes + comments) over the current follower count, for posts and Reels together, as a 90-day age-weighted median. Plays never enter it, so a per-play Reel rate in a report is not comparable with it.
• Stories are not part of any engagement rate.

Each weekly report prints three separate dates: when it was published, the period it covers, and the day its counters were read. For the reports of 19–25 October 2026 these are 30 October 2026, 19–25 October 2026 and 29 October 2026.

• A change to a published report is dated on the Version history page.