The Essence of Influencer Seeding
Companies exist to generate profit. And profit is created by assigning value beyond the cost of
raw materialsโin other words, by producing good products or good servicesโand
selling them to others.
Researchers and developers are the ones who create the product in its ideal form, while
marketing
communicates the productโs existence to the market, offers direct experiences (product usage) or indirect experiences (product/branding design),
and leaves a compelling impression that encourages purchase or repurchase.
And one way to amplify that impression through the voices of influential people is
influencer seeding.
Seeding literally means โto plant seeds.โ
You send the product (the โseedโ) to influencers and expect them to voluntarily create content,
allowing organic viral effects to โgrowโ on their own.
Traditional marketing explains the flow using B2B (business-to-business) and
B2C (business-to-consumer), but
influencer collaboration occupies a very unique space.
Itโs not selling directly to individual consumers like old-fashioned door-to-door insurance sales,
nor is it selling to corporations. Itโs an entirely distinct area of marketing.
TikTok Influencer Seeding Work
There are two major factors that I personally find important in seeding work:
- Whether the person is likely to accept a collaboration offer
(especially for unpaid collaborations, large creators often decline)
- Whether their content can generate enough user engagement for the brand
(recent view trends, content type, and brand alignment all matter)
Paid collaborations generally have higher acceptance rates because compensation is involved.
But for unpaid seeding, even if the creator receives the product for free,
making a video is still additional labor,
and it influences their imageโso acceptance is much lower than people think.
Personal Struggles
It has already been a year and a half since I started this job.
My workflow begins with collecting influencer page links in bulk through Google keyword searches.
Then I use a program I built to remove duplicates from previous collaborations,
and I manually open each link to evaluate the content quality.
With this process, I can make a list of 50 creators in 4 hours, or up to 80 if I work very efficiently.
After sending DMs, I typically receive positive responses from 1โ2 people, sometimes 3โ4.
Itโs basically me carrying out the kind of work that often gives people the impression of
โUghโฆ another advertisement message.โ
But what can I sayโthatโs the nature of marketing.
Looking at this workflow, there are definitely parts that seem automatable. But reality isnโt that simple.
What Can Be Fully Automated
First, collecting influencer page links is technically automatable and can be done without putting any burden on TikTokโs servers.
By attaching site:tiktok.com to relevant Google keywords,
expanding all results under the โShort Videosโ tab,
and using the Chrome extension Link Grabber,
you can extract all links visible on the page at once.
This process extracts links from HTML that is already loaded in the browser, so it does not generate additional traffic.
In other words, it does not burden TikTokโs servers.
Itโs simply re-collecting information that Google has already crawled and displayed.
After that, cleaning the collected links (https://www.tiktok.com/@user/videos/...)
and removing duplicates with existing lists
can also be automated safely and legally using simple Python scripts.
The Real Issue: Reading Influencer Profile Data (=Traffic Generation)
But what marketers truly need is not just the โlink.โ
Itโs the actual data that loads when you open the profileโ
follower count, recent video frequency, content type, view/like trends, contact links, etc.
To check this information, you ultimately must access TikTokโs servers,
and that is when request traffic occurs.
If TikTok detects abnormal patterns (too many requests, too fast, repeated loading),
it triggers temporary rate limits or access blocks
to prevent automation.
Which means full automation is extremely limited,
and speeding up requests results in instant restrictions.
(Even opening multiple profile links simultaneously can get you throttled.)
Because of this, applying automation end-to-end is difficult,
and it naturally raises the question of โHow far is technically permissible?โ
I enjoy this work and genuinely want to do it better with more efficiency,
but the technical limitations are very clear, and it can get frustrating.
So in my next article, I plan to explore whether this problem can be fundamentally solved
(Is marketing doomed to be repetitive manual labor?),
and how existing influencer-analysis platforms
collect such massive amounts of data.