DEV Community

Cleanrows
Cleanrows

Posted on Originally published at henryklimes.github.io

Check which companies run Google Ads, and export competitor ads in Python

Google has a public Ads Transparency Center, but no generally available API for it. The official Transparency Center API only covers ads served in the EU/EEA, and only for approved callers. If you want to score a list of leads ("do they already spend on Google Ads?") or export a competitor's ads with run dates, you are left clicking around by hand, or building a scraper against a page that was not designed for it.

In this post I'll show the approach I use now: a hosted Actor on Apify, Google Ads Transparency API – Competitor Ads & Advertiser Check, called from Python with apify-client. I built it, so I'll also be straight about what it does not return. Every output below is from a real run on 17 November 2026 that cost $0.01.

What we want

  1. A yes/no plus a size signal for each company: is it advertising, roughly how many ads Google shows, and when the latest one ran.
  2. The ads themselves, as rows: format (text, image, video), first and last shown dates, days running, a creative image or preview, and a link back to the Transparency Center.
  3. Brand names and domains, not only advertiser IDs. Google's own search box does not match a plain brand keyword.

Setup

pip install apify-client
export APIFY_TOKEN=...   # from console.apify.com → Settings → API & Integrations
Enter fullscreen mode Exit fullscreen mode

A free Apify account includes monthly platform credit. The Actor charges $0.50 per 1,000 ads and $2 per 1,000 company checks, with no start fee. Not-found and failed lookups are free.

1. Check which companies run Google Ads, then export a few ads

import os
from apify_client import ApifyClient

client = ApifyClient(os.environ["APIFY_TOKEN"])

run = client.actor("henryklimes/google-ads-transparency").call(
    run_input={
        "queries": ["hubspot.com", "zoho.com", "Canva"],
        "region": "US",
        "maxAdsPerQuery": 4,
        "shownInLastDays": 30,
        "includeSummary": True,
    },
    max_total_charge_usd=0.01,  # hard cap for this run
)

rows = list(client.dataset(run.default_dataset_id).iterate_items())

for r in rows:
    if r["type"] == "summary":
        print(
            r["query"],
            r["is_advertising"],
            r["ads_total_label"],
            r["latest_ad_shown"],
            [a["name"] for a in r["advertisers"][:3]],
        )
Enter fullscreen mode Exit fullscreen mode

Output from the live run (3 summary rows and 8 ads, about 20 seconds):

hubspot.com True 400-500 2026-11-17T21:01:38Z ['Hubspot, Inc.', 'HubSpot', 'Mileage Quest LLC']
zoho.com    True 1000-2000 2026-11-17T21:08:50Z ['Zoho Corporation Pvt. Ltd.']
Canva       True 1000-2000 2026-11-17T21:07:07Z ['Canva, Pty Ltd', 'CANVA US, INC.', 'Dobler Management Company, Inc.']
Enter fullscreen mode Exit fullscreen mode

All three were advertising in the US in the last 30 days, and each had an ad shown the same day as the run (times are UTC; that is 22:01, 22:08 and 22:07 in Amsterdam, CET). Google publishes ad totals as ranges (400-500, 1000-2000); small counts are exact.

The $0.01 cap is what stopped the run after 8 ads (3 checks × $0.002 + 8 ads × $0.0005 = $0.01). HubSpot got a summary row but no ad rows. Zoho and Canva returned 4 ads each:

query     advertiser                   format  first_shown  last_shown  days
zoho.com  Zoho Corporation Pvt. Ltd.   TEXT    2023-11-24   2026-11-17  1085
zoho.com  Zoho Corporation Pvt. Ltd.   IMAGE   2026-11-11   2026-11-17     7
zoho.com  Zoho Corporation Pvt. Ltd.   TEXT    2021-10-25   2026-11-17  1423
zoho.com  Zoho Corporation Pvt. Ltd.   IMAGE   2026-11-11   2026-11-17     7
Canva     Canva, Pty Ltd               TEXT    2023-01-28   2026-11-17  1379
Canva     Canva, Pty Ltd               TEXT    2023-10-14   2026-11-17  1121
Canva     Canva, Pty Ltd               TEXT    2026-05-22   2026-11-17   181
Canva     CANVA US, INC.               IMAGE   2026-11-05   2026-11-17    13
Enter fullscreen mode Exit fullscreen mode

A couple of Zoho and Canva text ads have been running for three or four years. That is the kind of thing a swipe file or a "what is still working for this competitor?" check is for. Each ad row also has image_url or preview_url, and an ad_url that opens the creative in the Transparency Center.

To check a lead list without paying for creatives, set maxAdsPerQuery to 0. You get one summary row per domain. I ran that on five SaaS domains (HubSpot, Pipedrive, Zoho, monday.com, Asana) for $0.01; all five were advertising.

2. Brand names, and why a domain is not the same as one advertiser

Canva is a brand name, not a domain and not an AR… advertiser ID. Google's own search does not resolve that. The Actor matches the name to advertiser accounts, or falls back to the brand's website.

That fallback matters. A domain search returns every advertiser whose ads point at that domain, which can include resellers, agencies and partners. The HubSpot row above is the example: next to Hubspot, Inc. you also get names like Mileage Quest LLC. Use an advertiser ID when you want one account only. All ads of an advertiser, last 90 days is the ready-made version of that.

Limits, honestly

  • Ad copy text is not extracted. Image ads and many text ads include an image_url. Video and some responsive ads only have Google's preview page, plus the ad_url.
  • Google's official API is EU/EEA only, and not generally available. This Actor reads the same public data the Transparency Center website shows, for any country.
  • Ad counts are ranges, not exact totals, except when the number is small.
  • Domain and brand searches are wider than one company, as in the HubSpot and Canva rows above.
  • The data is whatever Google is publishing at the moment you run it. No login, no cookies, no proxy in normal use.

Try it

If a brand does not resolve, the row comes back as not_found (free) with Google's closest advertisers, and you can rerun with an ID. I read the issues on the Actor page.

Disclosure: I built this Actor; the post was written with AI assistance from real runs on 17 November 2026.

Top comments (0)