Threads Brand Monitoring: AI Sentiment & Mention Alerts on the Apify Store →
A step-by-step tutorial with Threads Brand Monitoring: AI Sentiment & Mention Alerts on Apify.
For the Apify blog guest program, dev.to, Medium and Hashnode — publish with a canonical link to https://apify.com/rich_minds/threads-brand-monitoring-ai. Every output below is from the Actor's free demo run on its 8 sample posts (2026-09-28). The live-search yield ("1 in 3") is an estimate until the first measured live run.
Threads is where people complain about an app update, ask "any recommendations?" before they buy, and compare tools out loud. If you run brand, social or support for a product, those posts are the ones you want to see first.
Today there are three ways to see them, and none works well:
The third route is where most technical teams end up, and it has a specific failure: homonyms and noise. If your brand is Notion, you get "the notion of free will is overrated". You get crypto giveaways that stuff #notion into twenty hashtags. And next week you get the same posts again, billed again. Someone still has to read the whole pile.
It runs a Threads search on your Apify account for your brand, competitors, @handle and #hashtags, then keeps only the posts really about you. Each qualified mention gets:
escalate_support, reply, sales_follow_up, thank_and_amplify, monitor;Off-topic posts, homonyms, spam, reposts, your own posts and repeats from earlier runs are dropped before they are charged. You pay per qualified mention; the first 25 are free.
Open the Actor's page and click Try it. The form is pre-filled with 8 sample Threads posts about Notion, with Obsidian, Coda and Confluence as competitors and notion of as an exclude phrase. Press Start.
The demo takes about a tenth of a second and returns 6 qualified mentions from 8 posts. The philosophy post and the crypto giveaway are gone. The digest the run writes opens like this (verbatim from the demo):
6 new qualified mentions (2 high priority) from 8 loaded
share of voice: Notion 66.7 %, Obsidian 11.1 %, Coda 11.1 %, Confluence 11.1 %
- act first (high): complaint — Author reports Notion crashes on iPad after the recent update,
resulting in lost meeting notes, and asks if it is a known bug.
- act first (high): buying intent — 12-person agency looking to switch away from Notion due to
needing offline mode and faster databases.
The six rows cover six types — a complaint, a buying question, a comparison, praise, a how-to question and a news share — each with its action. That is the whole point: the complaint and the churn risk sit on top.
Replace Notion in Brand terms with your brand (say, Acme), add a rival or two in Competitor terms, and write one line in Brand brief: "Acme — the CRM for small agencies." Typing your brand next to the sample switches the run to the live Threads search on your account; nothing else to toggle.
Max source spend ($0.50 by default) caps what the Threads search may cost on your account. The first run of a search looks back 30 days, so your first result is a month of signal, not a week.
Your first 25 qualified mentions are free on either tier, so leave AI assessment on and see what it adds. Here is the same demo mention with the AI off and on:
{"authorHandle": "maya.builds", "score": 70, "mentionType": "complaint", "priority": "high",
"suggestedAction": "escalate_support", "insight": null, "replyDraft": null, "assessedBy": "rules"}
{"authorHandle": "maya.builds", "score": 100, "isAboutBrand": true, "mentionType": "complaint",
"sentiment": "negative", "priority": "high", "suggestedAction": "escalate_support",
"evidenceQuote": "Notion keeps crashing on my iPad since the last update and I lost a whole page of meeting notes. …",
"replyDraft": "Hi Maya, we're so sorry you lost your notes! … Please reach out at notion.so/help …",
"assessedBy": "ai"}
The AI confirms the post is about your product (the homonym check reads your brief), copies the sentence that proves the verdict and drafts a reply in your voice. Reading, labelling and answering one mention by hand took about 3 minutes when we timed it on the 8 demo posts.
Open the Act now view of the dataset: priority, action, type, author, followers, the quote and the draft, one row per mention. authorProfileUrl is where you reply or DM. Export it as CSV / Excel, or send it to Google Sheets from the Integrations tab. If you work in a CRM, upsert on dedupeKey (threads:<postId>) so a mention seen again never duplicates a record.
A live run with the defaults — 1 brand term × 80 posts:
| Line | Rules only | AI on |
|---|---|---|
| Threads search on your account (automation-lab Threads Scraper) | $0.42 | $0.42 |
| ≈ 27 qualified mentions (1 in 3, estimate) | 27 × $0.0035 = $0.09 | 27 × $0.009 + ≈ $0.01 tokens |
| Total | ≈ $0.51 | ≈ $0.68 |
Your first run is mostly free: 25 of those mentions cost nothing. At $0.0035 per qualified mention the basic tier sits below the $0.004 median of the Threads Actors in the same Store search — and one qualified mention replaces about 3 raw posts you would otherwise read.
Your first real run saves the search as a task and links it in the run status. Open it, click Schedule, choose weekly and put your address in Notify e-mail. From then on:
A busy brand that fills its post cap is told so (scanCapHit) — switch to the daily preset. A PR team can use the Crisis watch preset every 3 hours to catch a negative spike the same afternoon.
from apify_client import ApifyClient
client = ApifyClient("<YOUR_API_TOKEN>")
run = client.actor("rich_minds/threads-brand-monitoring-ai").call(
run_input={"brandTerms": ["Acme"], "competitorTerms": ["Globex"],
"brandBrief": "Acme — the CRM for small agencies.",
"maxQualified": 50, "maxDiscoveryChargeUsd": 0.5},
timeout_secs=3600,
)
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
print(item["priority"], item["mentionType"], item["url"])
It also works as an MCP tool: point Claude or Cursor at https://mcp.apify.com/?actors=rich_minds/threads-brand-monitoring-ai and ask for "the high-priority complaints and buying questions about Acme on Threads this week".
If you need one quick look at one keyword, the free Threads app search is enough. If every post containing your word is about you (a unique brand name), a raw keyword scraper is cheaper per row. This Actor pays off when your name is a common word, when you want a ranked queue instead of a pile, and when you want it every week without reading the same posts twice.
Try it free: https://apify.com/rich_minds/threads-brand-monitoring-ai — the demo needs no setup, and the first 25 real mentions are on us.
Threads Brand Monitoring: AI Sentiment & Mention Alerts on the Apify Store →