How to Find X Influencers for Your App Without Guessing

A founder sent me a spreadsheet in March. Sixty rows, sorted by follower count descending, top account somewhere north of 400k. He wanted to know which ones we could book for his scheduling app. I opened the first four profiles, and three of them hadn't posted anything but quote-tweets of other people's threads in six weeks. The fourth was posting daily and pulling maybe eleven replies on a good one.

He'd sorted the wrong column. Follower count is the number X shows you first, so it's the number people build lists around, and it's the one that tells you the least about whether a post will send anybody to your App Store page. I've watched a 12k-follower account move more signups than a 300k account on the same week for the same product. So here's the actual process, operators included.

How do you find X influencers for your app?

Search X for the problem your app solves, not your category, using operators like min_faves:150 and -filter:replies to surface posts that traveled. Read the authors, not the posts. Build a list of 20 to 30, then cut it to 5 by checking reply quality, posting cadence, and whether their audience is buyers.

That's the compressed version. The rest of this is how each step actually goes.

Where I start: operators, not lists of "top creators"

Every "Top 50 X Influencers for SaaS" listicle is scraped from follower counts and it's stale by the time it ranks. I don't use them. I use X's advanced search, which most people have opened twice and never learned, and its operator set.

The searches I run for a new app, in order:

"my calendar is a mess" min_faves:100 -filter:replies lang:en
(scheduling OR calendar) (app OR tool) min_faves:200 -filter:replies since:2026-05-01
"switched from" (Calendly OR Cron) min_faves:50
url:producthunt.com (scheduling OR calendar) min_faves:100
list:1234567890123456789 min_faves:100

Note what's happening. The first search is for the complaint, phrased the way a human would type it at 11pm. The second is category language, which finds fewer people but the right ones. The third finds switching moments, which are gold, because whoever posts "switched from X to Y and here's why" has an audience that treats them as a filter for tools. The fourth uses url: to find people who post about launches at all. The fifth runs the same threshold inside a List, which I'll get to.

min_faves is the load-bearing operator. It sorts for posts that traveled rather than posts that exist. Start at 100, and if you get forty results, raise it. If you get four, drop it to 40 and accept more noise.

Then I ignore the posts and open the authors. The post is evidence that this person can make something move. The profile tells me whether they can do it again next Tuesday.

Four lists, and why lists beat search after week one

Search is for discovery. Lists are for watching, and watching is where the real read happens, because a profile is a highlight reel and a timeline is a habit.

I keep four private X Lists per active campaign:

• Candidates. Anyone who cleared the 60-second read. Usually 25 to 40 people.
• Watching. People I'm unsure about. I let them sit for ten days and see what their median post looks like, not their best one.
• Adjacent. Creators in a neighboring niche whose audience overlaps. For a scheduling app, that's productivity people, but it's also freelancers-and-invoicing people, because both audiences are self-employed and time-poor.
• Burned. Anyone who took a deal from a competitor in the last 90 days, anyone who ghosted, anyone whose sponsored post read like a press release. This list saves me more time than the other three combined.

The Watching list is the one founders skip and then regret. A profile can look excellent on a Tuesday because you happened to land on the week they posted their best thread of the year.

Reading a profile in 60 seconds

I do this maybe forty times a week and it's genuinely a minute, because I look at five zones in a fixed order and stop at the first one that fails.

Zone 4 deserves more than a caption. Take a post from the middle of their timeline, one that clearly didn't blow up, and read the first fifteen replies. You're checking whether the replies are sentences from people with real timelines, or whether they're "🔥🔥" and "well said" from accounts that reply to forty people an hour. A creator whose average post pulls nine thoughtful replies is worth more to a paid campaign than one whose average post pulls two hundred emoji.

On zone 5: read their last sponsored post out loud. If it sounds like the brand wrote it, their audience heard that too, and yours will be the same. The people I book hardest are the ones who argue with the brief. Arvid Kahl pushes back on framing constantly, and that's a good sign, not a difficult client.

The three numbers I check before I reply

Only three. Everything else is decoration.

Link-post survival is the one nobody checks and it's the one that has cost me the most money. I ran a campaign in 2025 where the creator's organic posts routinely cleared four figures in likes, and every post they'd ever made with an outbound link sat at a fraction of that. We shipped anyway because the fit was so good. It underperformed the two smaller creators in the same campaign by a wide margin, and I should have caught it in the profile read.

Here's what the reply ratio looks like across four profile types. These four are illustrative composites, not real accounts and not Wiral data, drawn to show the shape of the thing.

What disqualifies a creator instantly

No debate, no second look, back to search.

The disclosure one isn't me being fussy. The FTC's endorsement guides put the obligation on the advertiser too, so a creator who hides paid relationships is a liability you're renting. Also, and this matters more commercially, their audience usually already suspects. Trust that's been spent quietly doesn't come back.

Engagement pods are the one I get argued with about. A pod is a group chat where twenty accounts agree to reply to each other within minutes of posting. You can spot one in thirty seconds: open three of their posts and look at who replied first. Same names, every time, and the replies are content-free.

The DM, and the part founders always get wrong

Once someone clears the five, this is what I send. Copy it.

hey [name] - the [specific post, quoted in 6 words] one from last week
is why I'm here. we built [app], it does [one sentence, no adjectives].

it's genuinely relevant to [their specific audience descriptor], and
if it isn't, tell me and I'll go away.

paid, not a "free access in exchange for a post" thing. what's your
rate for one post plus one reply from you in the thread? happy to
work to your format, I'm not sending you copy to read out.

Three things in there are doing work. Quoting a real post proves you read the timeline. Naming the rate question first respects that this is their job. And "I'm not sending you copy to read out" removes the fear that kills most replies, which is that the creator thinks you'll make them sound like a brand.

The part founders get wrong: they want fifty of these. Ten good creators is a campaign you can actually brief, read, and adjust mid-flight. Fifty is a mail merge, and creators can smell a mail merge through the screen.

Where this stops working

Under about 100k followers, everything above holds. Above that, the reply ratio compresses because huge accounts get replies from people who follow the reply-guy strategy rather than the account, and the number drifts toward noise. For large accounts I switch to a harder test: I ask for a screenshot of the link clicks on their last two link posts. Most will send it. The ones who won't are usually telling you something.

The other limit is niche depth. If your app serves a small technical audience, the pool of creators who genuinely have that audience might be nine people worldwide, and no amount of clever operator syntax will manufacture a tenth. In that case, stop sourcing and start budgeting for the nine.

I still think the biggest edge available to a small team right now is the willingness to spend two hours reading timelines before spending a dollar. Almost nobody does it, which is exactly why it works.

If you'd rather skip the two hours, that's our job. Browse the creator network to see profiles and rate-card context, or send us your product and we'll come back with the five names we'd actually book. Creators can apply here.