ICP Feed Scraping

ICP feed scraping is industry slang for systematically observing the personalized TikTok or Instagram feed of an ideal customer profile (ICP) in order to see which creators the ranking system is pushing. In careful product language it is feed-as-avatar observation: real phones, a defined persona, public content, a list of who keeps showing up.

The word “scraping” here does not mean unauthorized access, credential stuffing, or ripping private data. It means you are not relying on a static directory. You are reading a live, personalized public feed the way the customer would.

Why it matters to an indie iOS or SaaS founder

Hashtag search and influencer databases answer “who exists in this tag?” Your growth problem is “who is my user actually watching tonight?” Those sets overlap less than vendors imply. A climbing-app founder who exports every account tagged #climb will drown in gyms, travel pages, and one-off trip posts. A session that behaves like a sport climber in their twenties will keep surfacing the same three coaches and the same injured-finger rant account. That recurrence is the signal.

Doing this yourself has costs. It burns evenings. It contaminates your personal algorithm. It produces a notes app graveyard. Founders then conclude the niche “has no influencers,” which usually means they scrolled as themselves.

ICP-aligned feed observation is also how you avoid paying a macro who used the right keyword once. If they never appear in the avatar’s session, they are not distribution into that avatar, no matter what the media kit says.

How to do it without turning into a sketchy crawler

Use accounts and devices you are allowed to operate. Warm a session with interests, language, and geography that match the user. Do not impersonate a real human being or use stolen identities.

Watch like a user, log like an operator. For each recurring creator: handle, why they appeared, whether they already discuss apps or only the adjacent problem, public email or form.

Repeat sessions. One twenty-minute scroll is a mood. Several sessions on more than one device is a pattern. Personalization needs time.

Stay in public surfaces. For You, Explore, Reels. Do not automate your way through logins you do not own. Do not collect non-public emails. Business inquiry addresses that creators publish are fair game; harvested private address books are not.

Translate the list into outreach, not a warehouse. The value is who to contact this month, not a 50,000-row scrape you will never email.

How JumpTrend approaches it

JumpTrend’s avatar algorithm is this method as a product. You define the avatar (niche, age, problem). JumpTrend scrolls TikTok and Instagram on real phones it operates. You get trending creators in that feed. Every plan includes the scroll. Scout ($1,500/month) delivers the list. Outreach and Contract add inbox and paperwork.

Positioning against hashtag tools is explicit: those tools search what people tagged; JumpTrend scrolls what the customer is served. That is ICP feed observation, not a 350-million-profile index.

JumpTrend does not ask for your personal TikTok login and does not promise that a listed creator will convert. It promises the matching work: relevant people from the live feed. If your lawyers hear “scraping” and panic, use the phrase on the homepage instead. we scroll their algorithm on real phones, because that is the actual design.

What to log from a session (and nothing else)

Handle, platform, why it appeared (topic), recurrence count, whether they already demo apps, public contact, a one-line “would my user know them.” That is a matching log. It is not a dossier.

Do not store unrelated personal data. Do not record minors as targets. Do not automate follows/likes to “warm” a session in ways that violate platform rules. A useful avatar phone watches, waits, and notes repeaters. A sketchy crawler tries to drain the firehose. JumpTrend’s design is the first. If a vendor brags about the second, leave.