If you have ever tried to answer "what does 1,000 Instagram followers cost?" by scraping a few SMM panels into a spreadsheet, you probably got a clean-looking answer that was wrong. We run a pipeline that reads the public price lists of 33 SMM panels every six hours, and the main lesson from it is that the question itself is malformed. There is no single price. There is a ladder, and most of the work is making sure the rungs on that ladder are actually the same product.
Every figure below is quoted as of the crawl on 21 Nov 2026, 20:46 UTC, taken from the published market pages. Nothing is extrapolated.
The method, in plain terms
The pipeline has four steps, none of them clever on their own:
- Crawl. SMM panels resell through an API and publish a service list: a name, a rate per 1,000 units, sometimes min/max quantities. We read those lists on a schedule.
- Normalize. Each raw service name ("IG Followers [REAL] [30 Days Refill] - Instant!!") is resolved into a canonical product: platform, service type, and the dimensions that change what you are buying.
- Compare like with like. A price is only compared against other panels selling the same canonical product. Each panel counts once, at its own cheapest listing of that product.
- Drop dormant rows. Listings priced at zero, or at figures nobody could pay, are removed before any comparison runs.
As of the crawl on 21 Nov 2026, 20:46 UTC, that meant 151,331 services priced across 33 panels, resolved into 26,006 comparable products, with 95% of services resolved automatically.
One product is a ladder, not a price
Take the plain variant "Instagram / Followers", the one carried by the most tracked panels. As of the crawl on 21 Nov 2026, 20:46 UTC, 31 panels sold it. The cheapest tracked rate was $0.21 per 1,000 and the median was $0.6, so the cheapest sat 65% below the median.
The full ladder is more interesting than either number. The second rung was $0.2125, the 16th and 17th were both $0.6, the 25th was $1, and the most expensive panel charged $6.624 per 1,000, which is +1,004% against the median. Same product, same unit, same snapshot.
YouTube comments are even wider. For "YouTube / Comments / Custom", 26 panels sold it as of the crawl on 21 Nov 2026, 20:46 UTC: cheapest $0.1394, median $3.0322 (95% below the median), and the top of the ladder at $62.21, or +1,952% against the median.
If you sample three panels and average them, you are measuring which three panels you happened to pick.
Why names must be normalized
The temptation is to group by platform and service type and call it done. That breaks immediately, because a refill window, a country or a quality word makes a different product at a different price. Here are the Instagram follower variants carried by the most tracked panels, as of the crawl on 21 Nov 2026, 20:46 UTC:
| Product | Panels | Cheapest tracked | Median | Below median |
|---|---|---|---|---|
| Instagram / Followers | 31 | $0.21 | $0.6 | -65% |
| Instagram / Followers / Real | 31 | $0.12 | $1.1 | -89% |
| Instagram / Followers / refill 30d | 27 | $0.55 | $1.79 | -69% |
| Instagram / Followers / Real / refill 30d | 26 | $0.34 | $1.36 | -75% |
| Instagram / Followers / HQ | 22 | $0.192 | $0.8615 | -78% |
| Instagram / Followers / refill 365d | 21 | $0.48 | $2.6 | -82% |
| Instagram / Followers / BR | 20 | $0.8869 | $2.0489 | -57% |
Notice the trap in the second row. The lowest price anywhere in the Instagram followers category was $0.12 per 1,000, but for "Real" followers, whose median was $1.1. A naive "cheapest Instagram followers" query would return $0.12 and compare it against the plain-follower median of $0.6. That is a different product, not the same thing at a better price.
The dimensions also have measurable costs. We measure them on matched pairs: the same product with and without the dimension, so the number says what the market charges for the dimension rather than that expensive products are expensive. Across all categories, as of the crawl on 21 Nov 2026, 20:46 UTC:
- a refill guarantee added +51% (median across 351 matched pairs),
- a named country added +134% (median across 844 matched pairs),
- a quality word (real, HQ, premium) added +48% (median across 468 matched pairs).
Inside Instagram followers alone, a named country added +265% across 138 matched pairs. Fold "BR" followers into the worldwide bucket and your median is polluted by a product with a different price ladder (median $2.0489 against $0.6).
Here is the normalization idea as a toy. This is illustrative pseudo-code, not our production code:
# Illustrative only: the idea, not the real resolver.
def canonical(raw_name, platform, service):
name = raw_name.lower()
dims = []
if "real" in name:
dims.append("Real")
elif "hq" in name:
dims.append("HQ")
refill = parse_refill_window(name) # e.g. "refill 30d", "refill 365d", or None
if refill:
dims.append(refill)
country = parse_country(name) # e.g. "BR", "US", or None for worldwide
if country:
dims.append(country)
return " / ".join([platform, service] + dims)
The hard part is the long tail of spellings, emoji and contradictory words in names, not the code shape.
Median vs cheapest: the spread per platform
Once products are comparable, you can ask how far the cheapest visible price sits below the middle of the market. Across every product that at least ten tracked panels sell, as of the crawl on 21 Nov 2026, 20:46 UTC:
- Twitter: 36 products, cheapest 84% below median
- Instagram: 129 products, 78%
- TikTok: 57 products, 78%
- Telegram: 37 products, 68%
- YouTube: 83 products, 61%
- Website: 73 products, 56%
The spread is widest on Twitter and narrowest on Website. A narrow spread is not necessarily a worse market; it usually means the providers behind the panels are the same few, so prices converge. The full table, including Facebook, Spotify and Threads, is on the market data page.
Dormant listings priced at zero
This was the least obvious finding and the one that most often breaks hand-made comparisons. Panels rarely delete a service they stopped supplying. They price it at zero, or at a number nobody could pay, and leave it on the list.
As of the crawl on 21 Nov 2026, 20:46 UTC, 142,981 listings carried a current price, and 532 of them (0.4%) were not prices at all: zero, or figures in the thousands of dollars per 1,000 units. That is a small share, but it lands exactly where you look first. Sort a scraped list ascending and the top row is usually a service that was switched off, not the cheapest supplier. We drop those rows before comparing, and we also treat an offer far below the median as dormant rather than as an opportunity. More detail is in the write-up on what SMM panel prices actually look like.
What this means if you compare panels
Practical rules, whether you are buying, reselling or building a scraper:
- Never quote one price for a product. Quote the cheapest and the median, and say how many panels you saw. "$0.21 cheapest, $0.6 median, 31 panels" is information; "$0.21" alone is not.
- Normalize before you aggregate. Refill window, country and quality words are separate products. Merging them gives you a median of things nobody sells together.
- Filter dormant rows first. Zero and absurd prices are not offers. If you skip this step, your "cheapest" column is a list of retired services.
- Count each seller once. A panel with many near-duplicate listings should not dominate the distribution; take its cheapest listing of the product.
- Remember "cheapest" is relative to what you track. Our figures are the lowest prices among the panels we read, not the floor of the whole market.
If you want to see a full ladder for one product, the Instagram followers page lists all 31 rungs from the same crawl.
Disclosure: PanelHawk sells these services itself, buying from the panels it tracks.
Top comments (0)