Cover image for I Built a "Fastest Pizza Near Me" Tool with the UberEats API and Python

Benjamin Flores Vega

"Which pizza place near me is actually good and fast tonight?"

UberEats doesn't really answer that. It sorts by relevance, promotions, and whatever the algorithm feels like — not by "best combination of rating and delivery speed," which is what I actually want at 8pm on a Tuesday.

So I built a tiny tool that answers it properly, using the UberEats Stores Search by Location and Keyword actor on Apify and about 40 lines of Python.

The idea

Pull every store matching a keyword at a given address — name, rating, estimated delivery time — then rank them myself instead of trusting the app's default order. A place with a 4.8 rating and 45-minute delivery isn't necessarily better than a 4.5-rated spot that shows up in 20.

Walkthrough

Install the client:

pip install apify-client

Fetch the data:

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")

run_input = {
    "locations": ["85 Marsh St, Newark, NJ 07114, USA"],
    "search_keyword": "pizza",
    "max_search_results": 30,
}

run = client.actor("datacach/ubereats-stores-search-by-location-and-keyword").call(run_input=run_input)
items = list(client.dataset(run["defaultDatasetId"]).iterate_items())

Each store comes back with estimated_time_to_delivery as a string like "20 min". Parse it and score each place — rating matters, but delivery time should pull the score down the longer it takes:

import re

def parse_minutes(text):
    match = re.search(r"\d+", text or "")
    return int(match.group()) if match else 60  # unknown time = penalize heavily

def score(store):
    rating = store.get("rating") or 0
    minutes = parse_minutes(store.get("estimated_time_to_delivery"))
    return rating - 0.03 * minutes  # tune this weight to taste

ranked = sorted(items, key=score, reverse=True)

for store in ranked[:5]:
    print(f"{store['name']:<30} ★{store.get('rating', '?'):<5} {store.get('estimated_time_to_delivery', '?')}")

That's it — a ranked top 5 that actually balances "good" against "fast," instead of whatever order the app decided to show you.

Make it a CLI

Wrap it so you can run it for any address/keyword combo:

import sys

if __name__ == "__main__":
    address, keyword = sys.argv[1], sys.argv[2]
    # ...call the actor with {"locations": [address], "search_keyword": keyword}...

python fastest_near_me.py "123 Main St, Newark, NJ, USA" "sushi"

Ideas to extend it

  • Compare cuisines: run it for "pizza", "sushi", and "burgers" at the same address and see what's actually fastest tonight across the board.
  • Compare neighborhoods: run the same keyword across a few addresses to find which part of town has the best delivery options.
  • Tune the score: weight rating_count in too — a 4.8 from 3 reviews is less trustworthy than a 4.5 from 500.

Try it

There's a free plan (1 location, 10 results per run) if you want to try this on your own address before scaling it up.

If you tweak the scoring formula into something better, I'd love to see it — drop it in the comments.