The keyword Kylie Cosmetics matters because posts can reveal product interests, customer experiences, and content that attracts attention. These are search samples, not a representative survey of all customers or a complete count of Bluesky activity.
Comparison pending: add the actual Day 3 pull as data/Scrape2.csv and knit again.
| scrape | posts | collected_at_utc | earliest_post_utc | latest_post_utc | unique_authors | mean_likes_plus_reposts | median_likes_plus_reposts | top_author_share_pct | minimum_100_met |
|---|---|---|---|---|---|---|---|---|---|
| Scrape1 | 100 | 2026-11-19T23:13:22Z | 2026-06-06 09:45:01 | 2026-11-17 15:17:19 | 69 | 3.49 | 0 | 5 | TRUE |
A capped sample of 100 posts on each day will have equal volume by design. This does not show that overall conversation volume was unchanged. Publication time spans may differ between pulls. Search results can overlap, and older posts have had more time to accumulate engagement.
Engagement is likes + reposts; replies are retained separately in the datasets. Ties in the top five are resolved by URI. A frequent account is not automatically an influential account.
On the first knit, a random sample of up to 20 posts per pull is
saved in the manual folder. Open each
Scrape1_labels.csv / Scrape2_labels.csv in
Excel. Fill in sentiment with positive,
negative, or neutral, judging the
attitude toward Kylie Cosmetics. Use neutral for informational posts or
no clear opinion; use notes for ambiguity. Read linked context when
necessary. Fill theme with one main topic per post, using
consistent names across both pulls. Choose 2–3 recurring topics after
reading, rather than forcing posts into invented themes. Save as CSV,
close Excel, and knit again. Blank sentiment is
unlabeled, not neutral. The sample is retained across
knits so your labels are preserved.
| uri | author_handle | text | like_count | repost_count | engagement |
|---|---|---|---|---|---|
| at://did:plc:qobvnkudcv3zlaklxxjduqoi/app.bsky.feed.post/3mpbkwqkchw2p | motherjones.com | Kylie Jenner’s willingness to cozy up to terrible people appears to run in the family. But her ability to sell things is shakier: Kylie Cosmetics is no longer relevant, and Jenner’s clothing line has yet to make an impact since launching in 2023. As for Meta, is there a less cool company? | 169 | 28 | 197 |
| at://did:plc:qobvnkudcv3zlaklxxjduqoi/app.bsky.feed.post/3mpgq2nqkh62y | motherjones.com | Kylie Jenner’s willingness to cozy up to terrible people appears to run in the family. But her ability to sell things is shakier these days: Kylie Cosmetics is no longer relevant, and her clothing line has yet to make an impact since launching in 2023. As for Meta, is there a less cool company? | 87 | 13 | 100 |
| at://did:plc:gsytyhzjivwhl7whwbzmtoxt/app.bsky.feed.post/3n2jdm2ldnq2n | consumerreports.org | Is there lead in your lipstick? We tested 24 popular products and found that most had very low levels, though results varied widely. “Clean” labels and price were not reliable indicators. See the full results and tips for minimizing exposure below. | 11 | 7 | 18 |
| at://did:plc:se7sqhgg6jbshe6myfegowcm/app.bsky.feed.post/3mryufalijc25 | theandblock.bsky.social | 🌎 Große internationale Brands Gymshark (Fitness-Bekleidung) Kylie Cosmetics / KKW Beauty (Kylie Jenner / Kim Kardashian) Allbirds (Nachhaltige Schuhe) Red Bull (Merchandise & Fan-Stores) Bombas (Socken & Bekleidung) Vielleicht haben ja ein paar Leute hier Lust die Emailadressen? 4/4 | 7 | 1 | 8 |
| at://did:plc:5hmkw7azxuuiemi4nh6cvyhk/app.bsky.feed.post/3mzgtzltrgd2y | lexlanham.bsky.social | kylie applies to register her 8 y/o daughter’s name, STORMI JENNER, as a trademark for a range of goods & services incl. skin care, lotions, cleansers, cosmetics, fragrances, hair care & nail polish. neither kylie nor stormi has announced plans to launch a line under that mark | 5 | 1 | 6 |
Your interpretation: read these five posts and describe what they share (topic, format, tone, links, or author). Counts alone cannot establish why they performed well.
| author_handle | posts | share_pct |
|---|---|---|
| beauty-glowmuse.bsky.social | 5 | 5 |
| headlinesociety.bsky.social | 5 | 5 |
| newsletter-tf.bsky.social | 5 | 5 |
| durumagazine.com | 3 | 3 |
| entertainmentnow.bsky.social | 3 | 3 |
| instyle.com | 3 | 3 |
| stupiddope.bsky.social | 3 | 3 |
| albarjeel-news.bsky.social | 2 | 2 |
| britbrif.bsky.social | 2 | 2 |
| celebritypictures.bsky.social | 2 | 2 |
| sentiment | posts | percent_of_labeled | labeled | sample_size |
|---|---|---|---|---|
| positive | 0 | NA | 0 | 20 |
| negative | 0 | NA | 0 | 20 |
| neutral | 0 | NA | 0 | 20 |
Labeled 0 of 20 sampled posts. Incomplete: label at least 10 posts, preferably all 20.
Percentages describe labeled sample posts only. Each post has equal weight; they are not percentages of customers.
Pending manual theme labels.
Sentiment among manually labeled sample posts; incomplete samples are marked.
A second collection is required. One dataset split into groups cannot demonstrate a change between collection days.
Open results/Bluesky_Analysis.xlsx in Excel. It contains the datasets, volume, top posts, accounts, manual labels, sentiment, themes, and available comparisons. Screenshot a relevant Excel table or create and screenshot a chart.
For an Excel PivotTable, open a labels sheet, select the table, and
choose Insert → PivotTable. Put sentiment
in Rows and uri in Values as Count; exclude blanks, then
choose Show Values As → % of Grand Total. Screenshot
the result. Your assignment asks for Excel evidence even though this
report automates calculations.
Initial post (write 200–250 words using your actual results): Explain your keyword and project connection; give the post count, dates and time span; describe 2–3 supported findings using the top posts, sentiment and account tables; identify 2–3 recurring themes; state a limitation and one marketing question. Include an external reference and your Excel screenshot. Do not submit this scaffold as findings.
Day 3 update (complete these four sentences): “On [actual date], I collected [actual n] posts using the same keyword. Compared with the first pull, [numeric change], while [supported similarity] remained similar. [Theme/top-post evidence] suggests [cautious interpretation], although [sample limitation]. I recommend our team [specific action connected to this evidence].”
Peer replies: Read two classmates’ posts and write 75–100 words for each. Address a specific finding, then suggest an alternative interpretation, a sample or keyword limitation, or a follow-up analysis. Their posts are needed to write accurate personalized replies.
Reflection: Are the people posting here likely to match your brand’s target customers? Do a few days show a meaningful shift or ordinary variation? What did human judgment contribute when interpreting sentiment, sarcasm, and themes? If instructed, post your reflection on your own account with #MKTG4000fiasco.
Clark-Keane, C. (n.d.). A marketer’s guide to Bluesky (& is it worth it?) WordStream. https://www.wordstream.com/blog/bluesky-for-marketing
The guide recommends experimentation to find a brand’s audience and voice. Use your observed themes and high-engagement posts to propose a small content experiment. This is a marketing interpretation, not proof that the brand’s customers are represented in the sample. The reference supplies context; it does not verify the results in this report.
Technical reference: Gruber, J., & Guinaudeau, B. (n.d.). atrrr: Basic usage. https://jbgruber.github.io/atrrr/articles/Basic_Usage.html
Tool disclosure: ChatGPT helped prepare the R collection and reporting code. R computed descriptive summaries; the student manually labeled sentiment and interpreted themes. Change this disclosure if your actual workflow differs.