WEIR
Get a key
Theme

How to Find Influencers on Twitter and Check Them (2026)

A data method for how to find influencers on Twitter: shortlist by bio, rank by engagement per view with a tested Python script, then drop the fakes.


Published · Updated · 7 min read

Cover image: Followers aren't reach

To find influencers on Twitter, search bios for your niche, rank the accounts by engagement per view, and drop the ones that look fake.

This guide is for brands and agencies picking creators on X (Twitter). You get a tested Python script that ranks your shortlist, real numbers for three tech creators, and the checks that catch fakes. We run Weir, the API it calls.

Here is the number to rank creators by, for one post:

def engagement_per_view(post):
    actions = sum(post[k] or 0 for k in ("likeCount", "replyCount", "retweetCount"))
    return actions / post["viewCount"] if post["viewCount"] else None
  1. Shortlist by bioNiche words, then a follower range.
  2. Pull recent posts20 original posts from 30 days.
  3. Score each creatorMedian engagement per view.
  4. Drop the fakesAbout screen, renames, bot signals.
  5. Reach out by handAnd label paid posts.

Start with the words your ideal creator would put in a bio: a niche, a job or a product type. X's developer docs say its user search finds accounts by "name, username, or content in their bio."

  • By hand: search the words on X and look through the accounts it finds. Our Twitter bio search guide walks through it.
  • By API: Weir's user search returns bios, followers and join dates, at 1 credit per account.
  • By topic: search posts with a minLikes floor, then note who wrote the ones that did well.
  • By List: a public X List built around your niche is a ready shortlist, and our List export guide pulls its members.

Keep the query to one or two words. In our bio search tests, a three-word query returned nothing.

For a Twitter micro influencer list, filter by followers. We call 10,000 to 100,000 micro here.

Step 2: Pull each candidate's recent posts

Engagement lives in posts. Weir's search pulls a creator's recent original posts with fromUsers, onlyOriginal and a 30-day since.

We ran that for three tech creators on 22 October 2026, at limit: 4 to keep it cheap. It cost 15 credits for 12 posts.

The first item for @MKBHD was a surprise, trimmed:

{
  "text": "RT @MKBHD: NEW VIDEO - iPhone Duo Review ...",
  "createdAt": "2026-10-21T19:17:47.000Z",
  "likeCount": 0,
  "viewCount": null,
  "retweetedPost": { "likeCount": 5973, "viewCount": 551522 }
}

It's MKBHD reposting his own video post. onlyOriginal drops replies, not reposts, and this one came back with 0 likes and no view count.

So skip any post with retweetedPost set, or it drags your average down.

Step 3: Score engagement per view, not per follower

Engagement per view is likes, replies and reposts divided by views. Take the median across a creator's posts, so one viral post doesn't decide the score.

Views count how often a post was seen; followers count who once hit follow. Here is how far apart they can be:

Median views per post, as a share of followers (22 October 2026)
@MKBHD, 6.2M followers13.3%
@ijustine, 1.8M followers4.9%
@UnboxTherapy, 2.4M followers3.2%

Unbox Therapy has more followers than iJustine, yet its typical post got fewer views.

A follower count alone would rank them the other way.

Step 4: Rank your Twitter influencer shortlist in Python

You need Python 3, requests and a free Weir key in WEIR_API_KEY. The first half fetches each creator's posts, skipping the last day while views still climb:

import os, statistics, time, requests
from datetime import datetime, timedelta, timezone

URL = "https://api.weirapp.io/v1/x/search"
HEADERS = {"Authorization": f"Bearer {os.environ['WEIR_API_KEY']}"}
NOW = datetime.now(timezone.utc)
stamp = lambda days: (NOW - timedelta(days=days)).strftime("%Y-%m-%dT%H:%M:%SZ")

def recent_posts(username):
    body = {"fromUsers": [username], "onlyOriginal": True, "limit": 20,
            "since": stamp(30), "until": stamp(1)}  # skip posts under a day old
    time.sleep(4)  # 15 searches a minute fits the free tier's rate limit
    r = requests.post(URL, json=body, headers=HEADERS, timeout=60)
    r.raise_for_status()  # a 402 means out of credits: stop
    return [p for p in r.json()["data"] if not p["retweetedPost"]]  # no reposts

The second half scores each creator and prints your ranking:

KEYS = ("likeCount", "replyCount", "retweetCount")

def score(posts):
    seen = [p for p in posts if p["viewCount"]]  # views can be null
    rates = [sum(p[k] or 0 for k in KEYS) / p["viewCount"] for p in seen]
    return {"rate": statistics.median(rates),
            "views": statistics.median(p["viewCount"] for p in seen),
            "followers": seen[0]["author"]["followersCount"]} if seen else None

shortlist = ["MKBHD", "ijustine", "UnboxTherapy"]
scores = {name: score(recent_posts(name)) for name in shortlist}
for name in sorted(filter(scores.get, scores), key=lambda n: scores[n]["rate"], reverse=True):
    s = scores[name]
    print(f"{name:13} {s['rate']:6.2%} per view", f"{s['views']:>9,.0f} views",
          f"{s['followers']:>10,} followers")

We fed it the three pages from Step 2:

ijustine       3.63% per view    87,384 views  1,787,564 followers
MKBHD          1.18% per view   818,847 views  6,150,324 followers
UnboxTherapy   0.53% per view    77,313 views  2,417,649 followers

Treat that as a demo of the math, not a verdict. Each creator had 3 or 4 posts, and iJustine's were under six hours old. With 20 settled posts each, your ranking holds up better.

If the script stops early:

  • A 402 means you're out of credits, and retrying won't help.
  • A 429 means you went over your rate limit, so wait a minute and rerun.
  • A name missing from the output had no original posts with views in that window.

Step 5: Check if the account and its followers are real

Before you write to anyone, check the account. X's "About this account" screen shows where X thinks it's based, how it connects and its renames.

Our call for Unbox Therapy cost 3 credits:

{
  "username": "UnboxTherapy",
  "createdAt": "2011-01-18T04:39:42.000Z",
  "aboutProfile": {
    "accountBasedIn": "Canada",
    "locationAccurate": true,
    "source": "Canada App Store",
    "usernameChanges": { "count": 0, "lastChangedAt": null }
  }
}

An old account, never renamed, based where it connects from: that's a clean result. These are worth a closer look:

  • A country that clashes with the bio, or locationAccurate: false. Our guide to account based in on X explains when it's wrong.
  • A recent rename on an old, popular account, which can mean it changed hands.
  • Bot-like numbers: thousands followed, few followers back, or a machine's posting rate. Run our Twitter bot checker for the full list.
  • Views far below followers, on post after post.

No single signal proves a fake. Several together are your reason to move on.

Step 6: Reach out by hand and label paid posts

Outreach is yours, since Weir only reads: write to each creator yourself. Keep your list to public creators and brands, never private people.

The FTC's influencer guide says to disclose "any financial, employment, personal, or family relationship" with a brand. It wants that disclosure "placed with the endorsement message itself."

Unbox Therapy's four posts show the habit. Each one said "#partner" or thanked the brand "for partnering on this video."

Rules differ by country, and this isn't legal advice.

What does Twitter influencer data cost per creator?

You pay per result, and failed calls cost nothing. For one creator, as of October 2026:

1credit per account found by bio search
21credits for 20 recent posts
3credits for the About screen
25credits per creator, about half a cent on the $5 pack

Your free credits each month check about 200 creators this way. X's own API charges $0.11 for a profile and the same posts. TwitterAPI.io lists $0.15 per 1,000 tweets, so its posts cost you less than ours.

When not to use Weir

  • You'd rather not write code. An influencer platform searches and scores for you; Weir hands you raw data.
  • Audience details matter most. Weir returns public posts and profiles, not the age or gender of a creator's audience.
  • Price per post comes first. TwitterAPI.io charges less per tweet, and its API can also post.
  • You need a contract with X. X's own API costs $0.005 per post read, as of October 2026.

Weir reads public data only and can't post or send messages. It isn't licensed by X, so access can change.

FAQ

How do I find micro influencers on Twitter?

Search one or two niche words by bio, by hand or by API. Keep accounts with 10,000 to 100,000 followers, then rank them by engagement per view. Check each finalist's About screen before you reach out.

Is there a free Twitter influencer finder?

X's own search is free to use by hand. For scoring, Weir's 5,000 free credits a month cover about 200 creators with the Step 4 script and an About check each.

What is a good engagement rate on Twitter?

There's no single standard, and guides measure it in different ways. Compare creators in your niche with one formula and one time window. Per view is fairer than per follower, since it counts how often the post was seen.

Sources

All checked on 22 October 2026:

Keep reading