Why geotagged TikTok posts arrive without a geotag

The place is the query, not a column

Ask TikTok Location Scraper what is being posted in New York and it will tell you. Ask the resulting dataset where those posts came from and it has no idea.

The output is a standard post record: the id and caption, views, likes, comments, shares and bookmarks, the hashtags, a channel block with the creator’s follower counts, the video and audio metadata, and a postPage link. Every one of those fields describes the post. None describes the place.

On a single-location run that costs you nothing, because the answer is sitting in the input you submitted. It becomes a problem the moment you put a second URL in the list.

The URL is not what you would guess either

There is no separate place page on TikTok to copy a link from. A location is a tag URL carrying a flag:

https://www.tiktok.com/tag/newyork?location=true

Drop the ?location=true and you have an ordinary hashtag query, which returns posts using the word as a topic rather than posts filed against the address. Those are different sets and the difference is invisible in the output.

The fix is free

Billing is per location query at $0.006, with roughly the first twenty posts included. That is true whether the locations arrive in one run or ten. So splitting a sweep costs nothing and buys you a labelled dataset:

def place_url(slug):
    return f"https://www.tiktok.com/tag/{slug}?location=true"

rows = []
for slug in ("newyork", "miami", "austin"):
    run = client.actor("apidojo/tiktok-location-scraper").call(
        run_input={"startUrls": [place_url(slug)], "maxItems": 20}
    )
    for row in client.dataset(run.default_dataset_id).iterate_items():
        rows.append({**row, "place": slug})   # the field the actor cannot give you

One extra key, added at ingest, and the comparison you wanted becomes possible.

What the free page is worth

Because the allowance is per location rather than per run, breadth is subsidised here in a way depth is not. Twenty cities scanned at their first page is twenty query fees and no item charges at all: twelve cents for four hundred posts.

A thousand posts from a single city, by contrast, costs the same query fee plus nine hundred and eighty item charges. Same actor, same price list, two very different shapes of answer.

If the question is “what is happening across this region”, the cheap sweep is also the better instrument. If it is “what is happening in this one place”, pay for the depth and skip the labelling problem entirely.