Price Drop Alerts with a Shopping API

A price-drop alert is a four-step loop: poll a product query on a schedule, read the numeric extractedPrice from each listing, compare it to your stored baseline, and notify when it falls. The Shopping API supplies everything the loop needs in one call — price, currency, merchant, and a stable product URL to key the alert on.

This guide walks through the pattern with a tested Python script and honest cost math, so you know what a real price-watch system costs before you build it.

Table of Contents

• The price-drop loop
• Price data the API gives you
• Working Python: a price watcher
• Cost math for a real watchlist
• Honest limits
• FAQ

The price-drop loop

Every price-watch system, from a hobby script to an enterprise repricer, is the same four steps:

1. Poll. On a schedule, query each product you watch.
2. Normalize. Turn the display price into a number you can compare.
3. Compare. Against the last price you stored for that product.
4. Alert. When the current price drops below the baseline (or your threshold).

The hard part is never the loop — it is getting a clean, comparable price reliably. That is where the API earns its keep.

Price data the API gives you

Each listing in the Shopping API response includes both a display price string and a numeric extractedPrice. The numeric one is what your comparison logic should use:

{
  "title": "Sony WF-1000XM6",
  "price": "$299.99",        // display string
  "extractedPrice": 299.99,  // number, ready for math
  "currency": "USD",
  "merchant": "Best Buy",
  "url": "https://www.bestbuy.com/product/sony-wf1000xm6-..."
}

Keying alerts on url (the merchant product URL) keeps the same product stable across polls. Keying on the title alone is fragile — two merchants can sell "Nike Vomero 18" at different prices, and that is exactly the difference you want to catch.

Working Python: a price watcher

Here is a complete, runnable watcher. It keeps baselines in a JSON file, polls a watchlist, and prints an alert for every product whose price dropped since the last run. Wire the print to your email, Slack, or push channel.

Replace sk_live_your_key with your key from the dashboard.

import json
import os
import time
import requests

API_KEY = "sk_live_your_key"
BASE_URL = "https://apiserpent.com/api/shopping"
STATE_FILE = "price_baselines.json"

def fetch_prices(query):
    resp = requests.get(
        BASE_URL,
        params={"q": query, "engine": "yahoo", "num": 20},
        headers={"X-API-Key": API_KEY},
        timeout=60,
    )
    resp.raise_for_status()
    out = {}
    for item in resp.json()["results"]["shopping"]:
        price = item.get("extractedPrice")
        if price is None:                 # Yahoo didn't show a price this poll
            continue
        out[item["url"]] = {
            "title": item["title"],
            "price": price,
            "merchant": item.get("merchant"),
        }
    return out

def load_state():
    if os.path.exists(STATE_FILE):
        return json.load(open(STATE_FILE))
    return {}

def save_state(state):
    json.dump(state, open(STATE_FILE, "w"), indent=2)

WATCHLIST = ["nike running shoes", "sony headphones", "iphone 15 case"]

state = load_state()
for query in WATCHLIST:
    for url, listing in fetch_prices(query).items():
        previous = state.get(url, {}).get("price")
        current = listing["price"]
        if previous is not None and current < previous:
            drop = previous - current
            print(f"PRICE DROP ({drop:.2f}) {listing['title']} "
                  f"${current:.2f} was ${previous:.2f} @ {listing['merchant']}")
        state[url] = listing

save_state(state)
print(f"\nWatched {len(state)} products. Baselines saved to {STATE_FILE}.")

Run it once to seed baselines, then again later — any product whose extractedPrice fell prints a PRICE DROP line. The script skips listings with no price instead of treating null as a catastrophic drop to zero.

Cost math for a real watchlist

Here is what a price-watch system actually costs on the Shopping API, billed per 1,000 calls (no page multiplier, no subscription):

Tier	Per 1,000 calls	500 keywords, 1×/day	5,000 keywords, 1×/day
Default	$0.60	$0.30/day	$3.00/day
Growth	$0.06	$0.03/day	$0.30/day
Scale	$0.03	$0.015/day	$0.15/day

At the Scale tier a 500-keyword daily sweep is a cent and a half a day. Even hourly checks of a 200-product watchlist (4,800 calls/day) stay well under a dollar a day. See the pricing page for the current full table.

Honest limits

• No history built in. The API returns the current price; you store the baseline. Your database is the history.
• Null prices happen. A share of listings appear without a displayed price on a given poll. Skip and retry them next round.
• Short rail. One query returns roughly a dozen products. For a broad catalog, split it across many queries.
• Source freshness. Prices are as fresh as the product rail at read time. For true real-time marketplaces, a marketplace's own API or feed is the better tool.

FAQ