How to Find a Competitor's Best-Performing Tweets
Search Operators That Surface Them in Minutes, and the Ratios That Tell a Hit From a Pile-On
Short answer
Search from:handle min_faves:N on X to list a competitor's most-liked posts, raising N until a manageable set remains. Then rank each post by views relative to the account median, and check its reply-to-like and bookmark-to-like ratios. High replies with few likes usually means backlash, not a hit.
1. How do I find a competitor's top tweets quickly?
Use X search operators. from:handle limits results to their posts, min_faves:N keeps only posts with at least N likes, and since: and until: restrict the date range. The Top tab then sorts what remains.
| Query | What it returns |
|---|---|
| from:veltrix min_faves:100 | Posts by @veltrix with at least 100 likes |
| from:veltrix min_retweets:25 | Posts reposted at least 25 times, a better proxy for spread |
| from:veltrix min_replies:20 | Posts that started a conversation, good or bad |
| from:veltrix min_faves:50 since:2026-06-01 until:2026-09-01 | Their strongest posts in one quarter |
| from:veltrix min_faves:50 -filter:replies | Strong original posts, with their replies to others removed |
| from:veltrix min_faves:50 filter:links | Strong posts that sent people somewhere, usually their own site |
Set the threshold from the account itself. Look at a handful of recent posts, note a typical like count, and start at roughly three times that. If the query returns hundreds of posts, raise it; if it returns five, lower it. You want twenty to forty candidates.
2. Why should I rank by the account median rather than raw likes?
Because raw likes rank posts by the size of the audience that day, not by how good the post was. A multiple of the account median views and engagement shows which posts beat that account normal.
The general method, median multiples within one account, is covered in the cross-platform guide to finding top posts. On X it gets one improvement: views are public, so you can compute two multiples per post instead of one.
For each candidate post, record:
- Views as a multiple of the account median views. This is how far X distributed it.
- Engagement per view compared with the account median rate. This is how well it converted the people who saw it.
- The date and format, so you can group winners later.
The two multiples separate different kinds of success. High distribution with an ordinary rate usually means the algorithm or a large account pushed it. An ordinary view count with a very high rate means the existing audience loved it, which is often the more copyable result.
3. How do I tell a hit from a pile-on?
Compare replies and quotes with likes. When replies approach or exceed likes, the post was argued with, not applauded. Read the replies before counting it as a success, because on X controversy inflates every count.
| Pattern | Usual meaning | Check |
|---|---|---|
| Likes far above replies, steady reposts | Genuine approval | Skim ten replies to confirm the tone |
| Replies close to or above likes | Disagreement or complaints (being "ratioed") | Read the top replies; log it as a controversy, not a hit |
| Quotes high relative to reposts | People are commenting on it to their own followers, often critically | Open the quote posts and read the framing |
| Bookmarks high relative to likes | Reference content people want to keep: guides, lists, threads | Note the format; these often convert best later |
| Views far above median, everything else ordinary | Wide distribution with little response, possibly promoted | Keep it out of the organic list |
Bookmarks deserve more attention than they get. A post that many people saved is one they expect to use, and X shows that count publicly on every post. A competitor whose most bookmarked posts are how-to threads has found the content their audience values enough to keep.
4. What about promoted tweets?
You cannot confirm from outside whether an X post was promoted. Treat a post whose views jump far above the median while replies, reposts and bookmarks stay near normal as possibly promoted, and keep it in a separate list.
Meta publishes an ad library that confirms boosts on Facebook and Instagram. X does not publish anything comparable outside a limited EU repository, so on X you are working from engagement shape alone. The boosted post detection guide explains the shape in detail; the short version is that bought views do not bring proportional replies and reposts.
Do not throw these posts away. If a competitor pays to distribute a post, that post states what they most want people to hear. Keep it as a positioning signal, just not as a content benchmark.
5. What is the full routine, step by step?
Query a quarter of posts above a like threshold, record the two multiples, flag ratio anomalies and possible promotion, then group what remains by format and theme. The deliverable is a pattern, not a leaderboard.
- Compute the account medians first. From thirty recent original posts, take the median views and the median public engagement rate. Everything later is compared with these two numbers.
- Run a dated threshold search. Search from:handle min_faves:N since: and until: for one quarter, with -filter:replies. Adjust N until twenty to forty candidates remain.
- Record the two multiples. For each candidate, write views over median views and its rate over the median rate, plus format and theme in a few words.
- Flag ratios and possible promotion. Mark posts where replies approach likes, where quotes dominate, or where views jumped without matching interaction. Move them to a separate list with a note.
- Group the rest by format and theme. Count how many winners share a format (thread, video, image, text) and a theme. A pattern across five posts is a finding; one big post is an anecdote.
6. What should I do with the list?
Write one or two sentences a planner can schedule, such as which format and topic repeatedly beat the median for two competitors and are missing from your calendar. Recheck the pattern next quarter.
Compare the top-post patterns across two or three competitors. If practical threads outperform for all of them, that is a category preference and a gap if you are not publishing any. If only one competitor wins with a format, it may be their audience rather than the format.
Oppira tracks competitor X posts daily and surfaces their strongest posts for the period alongside the same view from Facebook, Instagram and LinkedIn, so the quarterly search becomes a chart you check rather than a query you rebuild.
Key Takeaways
Search beats scrolling
from:handle min_faves:N with since: and until: surfaces a quarter of top posts in one query.
Set thresholds from the account
Start around three times a typical like count and adjust until twenty to forty candidates remain.
Two multiples per post
Views over median shows distribution; rate over median shows how well it converted viewers.
Replies near likes means argument
A post with replies approaching likes was contested, so read it before calling it a hit.
Bookmarks flag reference content
High bookmarks relative to likes point to guides and threads people wanted to keep.
Promotion is unconfirmable on X
Separate posts with views far above normal and flat interaction rather than benchmarking against them.
Frequently Asked Questions
Sources
- How to use advanced search X Help Center, September 2026.Official reference for advanced search fields, including engagement minimums and date ranges.
- View counts X Help Center, September 2026.Notes that older posts do not have view counts.
- Bookmark counts X Help Center, September 2026.
Explore More
Related analyses, benchmarks, and industry insights
Related Guides
Glossary Terms
Turn reading into a reaction
Oppira watches your competitors, keeps your playbook current, and drafts the response. See your market clearly by tomorrow.