Table of Contents
- What you'll end up with
- Before you start
- The four shapes a mention takes on X
- The free setup, step by step
- Step 1 — Read the tags
- Step 2 — Build the name search
- Step 3 — Build the link search
- Step 4 — Save both searches
- Step 5 — Put a slot in the calendar
- Where the free setup breaks
- The X API route (only if you're building it yourself)
- Triage: four labels, applied the day a post lands
- Ask before it goes on your site
- Doing this with Testimonial
- Where people get stuck
- Wrapping up
Image URL
Do not index
Do not index
AI summary
A guide to monitoring brand mentions on X (Twitter) that captures untagged posts by using saved name and link searches, setting up a weekly review routine, applying a four‑label triage system, requesting permission before publishing, and optionally leveraging Testimonial’s Brand Monitor for automated alerts and import, while noting limitations of the free setup and API options.
Title
How to Monitor Brand Mentions on X (Twitter)
Date
Oct 19, 2026
Description
Twitter monitoring that catches the posts without your @: the free X search setup, what the API costs, and how to turn the best posts into proof on your site.
Status
In progress
Current Column
In the queue
Person
Writer
Your Notifications tab shows every post that tags @yourhandle.
It doesn't show "switched the whole team to [your product], no regrets" typed without the @.
Or the post that misspells you. Or the one that just pastes your URL.
Those untagged posts are what this guide is for. You'll get the free setup on X itself, an honest look at when the X API is worth paying for, and the loop that turns a good post into proof on your site.
For the cross-platform picture, start with our guide to brand monitoring. This page is X only. And monitoring only finds praise nobody asked for; collecting the rest is still the whole testimonial pipeline.
What you'll end up with
Two or three saved searches on X that catch the posts your notifications miss.
A weekly slot where you actually open them.
A four-label triage rule for every match.
A permission ask you can send the same day.
And the best posts on your site with the author's name, handle and link intact. Not in a screenshots folder.
Before you start
You need:
- Your handles. The company account, plus the founder account if people tag you personally.
- A short name list. Product name, company name if it differs, the one or two misspellings people actually type, a branded hashtag if you have one. Five entries, not twenty.
- Your domain. People paste
yourproduct.comwithout naming you.
- A signed-in X account to run the searches from.
- A place on your site where approved posts will go: homepage strip, pricing section, or a Wall of Love.
The four shapes a mention takes on X
Most people only watch the first row. The other three are where the unprompted praise sits.
Shape | What it looks like (illustrative) | In your Notifications? | How you catch it |
Tag | "@yourhandle saved me an afternoon" | Yes, under "Mentions" | Notifications |
Name, no tag | "[Product] is the only tool from that stack I kept" | No | A name search |
Link | A post sharing yourproduct.com/pricing with no name in the text | No | A domain search |
Reply in someone else's thread | "We use [Product] for exactly this" under a stranger's question | Only if tagged | The name search, read with the parent post |
Watch all four or you are monitoring your notifications, not X.
The free setup, step by step
Step 1 — Read the tags
Open Notifications and select "Mentions". That's the tagged layer.
Reply where a reply adds something. Then move on, because this tab covers one shape out of four.
Step 2 — Build the name search
Type this into X's search box, swapping in your own names:
"yourproduct" OR "your product" OR yourprodcut -from:yourhandle- Quotes match the exact phrase.
- OR (capitals) catches any of the variants.
-from:yourhandledrops your own posts, which otherwise fill the first page.
Then switch the results tab from "Top" to "Latest". "Top" is X's selection. "Latest" is everything the search returns, newest first.
If your product name is also an ordinary word, add context or subtract noise:
"yourproduct" (app OR tool OR pricing) or "yourproduct" -hiring -job. Expect to tune it a few times before it's clean.Prefer forms to operators? Run any search, then open "Advanced search" from the results page. Its fields for exact phrase, any of these words, none of these words, accounts and dates build the same query.
Step 3 — Build the link search
yourproduct.com -from:yourhandleThis catches the shares that never type your name. It's the quietest search, and I'd still read it first: a stranger linking straight to your pricing page is pointing someone at you.
Step 4 — Save both searches
With the results open, use the three-dot menu by the search bar and choose "Save search". Saved searches appear when you tap into the search box, so next week is one tap, not retyping.
Step 5 — Put a slot in the calendar
Saved searches don't notify you. Nothing does on the free setup.
Pick a day. Open each search on "Latest" and read back to where you stopped last time. Adding
since:2026-10-12 (your last check date) to the query trims what you've already seen.Where the free setup breaks
Four problems with it, and I'd still start here:
- Nothing pings you. A great post on Monday waits until you open the search on Friday.
- X decides what search returns. You will not see everything, and posts from protected accounts are only visible to their approved followers, so they never appear for you at all.
- No memory. Nothing marks a post as handled. You re-read the same posts every week.
- Name collisions. A common-word name turns the search into a job board and a sports feed.
That's enough if you're named a few times a month and you'll open the searches on schedule. If you won't, the setup is decoration.
The X API route (only if you're building it yourself)
As of October 2026, the X API is pay-per-use: no subscription, you buy credits, and reading a post costs $0.005 per post returned. Pay-per-use accounts are capped at 3 million post reads per billing month; above that, X points you to an Enterprise plan.
Two endpoints matter for monitoring:
- Mentions (
GET /2/users/{id}/mentions) returns posts that tag an account. Called by your own developer app for your own account, X prices it as an "Owned Read" at $0.001 per resource. It only sees tags, the first row of the table above.
- Search catches untagged names and links. Recent Search covers the last 7 days; Full-Archive Search goes back to 2006 and is available to pay-per-use and Enterprise customers.
The bill is rarely the problem. Everything after the request is: storing what you've seen, deduplicating, a triage view, someone reading it.
Skip this unless you already have an engineer who wants that pipeline. A cron job that posts raw JSON into Slack is an alert, not a programme.
Triage: four labels, applied the day a post lands
Label | On X it looks like | What you do |
Publish candidate | First-hand, specific, from a public account, and it makes sense to a stranger without the thread above it | Send the permission ask |
Product note | "Love it, but I wish it…", a feature request, a workaround someone shared | Route to product. Not homepage material |
Support | A complaint, a billing problem, "is this down?" | Reply once in public, move to DM or email |
Ignore | Giveaway bots, job posts, name collisions, a round-up thread that listed you among twenty | Move on |
Three X-specific traps before you label anything a publish candidate:
Replies that need their parent. "Same, it's the only one that worked for us" is warm and useless on its own. If the card needs the thread above it to make sense, skip it.
Quote posts. Open the post being quoted. Praise quoting a complaint about a competitor is a different card than it looks. Praise quoting your own outage announcement might be sarcasm.
Dated context. "We just launched with [Product]" from three years ago can mislead next to this year's pricing. Keep the date visible or pass.
Two strong posts on your pricing page beat forty in a screenshots folder.
Ask before it goes on your site
Public on X is not permission for your homepage.
Reply or DM. A message might read:
Hey [Name] — your post about [the specific thing they said] made our week.
Mind if we feature it on our [homepage / pricing page / Wall of Love]? We'd keep your name, handle and a link back to the original.
Totally fine if you'd rather not.
If their DMs are closed, reply publicly and ask whether you can DM or email them. Save the yes somewhere you'll find it in six months.
If anything changed hands, such as a free month, an affiliate link or an employee relationship, disclosure rules may apply. I'm not your lawyer: start with the legal side of testimonials and the FTC Endorsement Guides, then check your own jurisdiction.
The full permission and attribution checklist, and the trade-off between X's native embed and importing the post, lives in turning a tweet into a testimonial. Once you have the yes, that page is your next step.
Whichever way you publish: exact wording, display name, handle, link to the original. If the author deletes the post later, take your card down.
Doing this with Testimonial
The saved-search routine above works. You don't need us for it.
What Brand Monitor changes is that the watch runs without you opening a search, and X lands in the same inbox as LinkedIn, Reddit, Hacker News, YouTube, TikTok and web search. Brand Monitor is on the Ultimate plan and above.
Setting it up for X, per the help centre:
- Open the Space and click "Brand Monitor" in the left sidebar.
- Click "Settings" in the upper-right corner.
- Fill in "Business Name", "Business URL" and "Twitter Handle". Add your LinkedIn and YouTube URLs too if you want those watched.
- Add keywords: the same short name list from "Before you start". The feature page's own example list includes a domain and an @handle, so put yours in. 10 keywords on Ultimate, 15 on Ultimate +. Competitor names can go here as well, but each one spends a slot.
- Save. The first scan starts straight away; after that, Brand Monitor scans every 24 hours.
- In the inbox, filter by source to see X on its own, then by sentiment (positive / neutral / negative / mixed) when you're hunting publish candidates.
- Send the permission ask. Then approve the post: it imports onto your social media inbox and your Wall of Love with the original author, link and screenshot kept.
Already have the post's URL? You don't need Brand Monitor. Every plan, Free included, has unlimited X imports. Click "Import" in the Space sidebar, then "+ Import", choose "Twitter / X", paste the URL, tick "Add to my Wall of Love" if you want it there now, and click "Add". With the browser extension, it's the three dots on the post, then "Add testimonial". Importing from X covers the details.
Where it breaks, including for us:
- Daily, not live. A scan every 24 hours suits proof. It is not a crisis pager.
- The back catalogue. The first scan's all-time sweep is described for the web and TikTok. Treat X as a watch that runs forward from setup, and import older posts you already know about by URL.
- Caps. 2,000 mentions a month on Ultimate, 5,000 on Ultimate +. You can buy extra mention credits, and they don't expire, but a noisy common-word keyword will still burn through them.
- Sentiment is a sort order. It will mis-tag sarcasm. Read every post before you publish it.
- Approving is not asking. The approve button imports the post. It doesn't speak to the author. I'd send the ask first, every time.
Ultimate is $50/month per space and Ultimate + is $95/month per space; check plans and pricing for the current numbers before you decide.
If you're named a few times a month and you really will open those saved searches on Friday, stay with the free setup. Brand Monitor earns its place when posts arrive faster than you search for them, and some of them belong on your site.
Where people get stuck
They only watch Notifications. Fix: add the name search and the domain search. Tags are one shape out of four.
They read "Top" and think they've seen everything. Fix: switch to "Latest".
Their name is an ordinary word. Fix: add context words or exclusions to the search, and use the domain and handle as Brand Monitor keywords rather than the bare word.
They reply to every mention as the brand. Fix: reply when you add something. A like is not triage, and a reply under every post reads like a bot.
They publish a reply that needs its thread. Fix: if a stranger can't follow it alone, it's not a card.
They tidy the wording. Fix: the exact words or nothing. A cleaned-up post is your copy with their name on it.
For the same loop across every platform, from alert to published proof, see brand mentions monitoring.
Wrapping up
That's Twitter monitoring that catches more than your tags: a name search, a domain search, a weekly read, four labels and an ask.
If you do one thing today: run
"yourproduct" -from:yourhandle on "Latest", scroll back a month, and send the ask for the best untagged post you find.Getting a page of noise back? Reply with your product name and I'll write the search for you.
Written by
