AI Overviews on mobile vs desktop: 40 searches compared

If you track AI Overviews or rankings, you have probably seen a rule of thumb that mobile and desktop results are “different”. How different, and in what ways? We ran the same 40 searches on both devices to find out.

The results come from the Google Search API, run on 16 October 2026 with country set to us. The 40 searches are the same list we used in our AI Overview tracker study: ten how-to, ten product, eight local business, six brand and six news or entity searches. We ran each once on mobile, then compared with a desktop run of the same list from the same day. It is a small sample we chose, so the figures describe it and nothing wider.

AI Overviews: the same rate on both devices

An AI Overview showed for 25 of the 40 searches on mobile and 25 on desktop. The counts were identical in every group:

Under the same totals, four searches differed. nike and premier league table had an overview on mobile only. facebook login and bitcoin price had one on desktop only. The other 23 overviews appeared on both.

That is a small difference, and it needs a caution. When we ran the same 40 searches twice on desktop, two searches changed between the runs: cheap flights to london lost its overview and tesla gained one. So four differences between devices sits close to the two we see between two desktop runs. With this sample, we cannot separate a device effect on whether an overview appears from the normal run-to-run movement.

The cited sources were mostly the same

For the 23 searches that had an overview on both devices, we compared the domains each cited:

• 13 of 23 had exactly the same set of domains on mobile and desktop.
• On average, 80% of the domains cited on either device were cited on both.
• The number of sources was almost the same: 7.7 on mobile against 7.8 on desktop.

The ten that differed were how to boil eggs, how to lower blood pressure, what is inflation, why is the sky blue, how to change a tire, how to learn python, best vpn, noise cancelling headphones, amazon and tesla. In most of them the two devices shared most of their sources. The widest gaps were best vpn, with 3 domains in common out of 9 cited in total, and amazon, with 2 out of 7.

One edge case is worth handling in code. For how to learn python, the mobile response had an AI Overview with no sources at all, while desktop cited one. An overview can come back with an empty sources list, so do not assume the list has an item.

Mobile shows more organic results

The number of organic results was higher on mobile in every group:

Across the 40 mobile searches the count ranged from 5 to 13. We did not test why. If you track rankings, it matters for pagination: a “page 1” on mobile holds about two more results than one on desktop. Count results, as in our rank tracking guide, instead of assuming a number per page.

People Also Ask and the knowledge panel

People Also Ask appeared for 30 of the 40 searches on mobile and 33 on desktop. That gap is as small as the movement between our two desktop runs, which were 29 and 33, so we would not read anything into it. The knowledge panel showed for 6 searches on both devices.

The local pack was missing on mobile

For all 8 local business searches, local_pack was empty in the mobile responses, and desktop returned a local pack for every one of them. The mobile responses had more organic results for these searches, about 10.4 on average, and two entries in those lists were google.com/localservices pages.

We do not know whether Google showed a local block on mobile that the response does not capture, so we are not claiming that Google has no local pack on mobile. What we can say is that in our test the local_pack field was empty on mobile. If you need local business data, use the Google Local API, which returns the businesses directly. In our tests it returned about 20 per page, and it has no device setting. We cover tracking it in local rank tracking.

The code

Put your key in an environment variable:

export CRAWLERDATA_API_KEY="YOUR_API_KEY"

This function runs one search on both devices and summarises each:

import os
import time
from urllib.parse import urlparse

import requests

API_URL = "https://api.crawlerdata.com/v1/jobs"
HEADERS = {"Authorization": f"Bearer {os.environ['CRAWLERDATA_API_KEY']}"}


def google_search(query, country="us", device="desktop", retries=4):
    body = {
        "endpoint": "google-search",
        "params": {"query": query, "country": country, "device": device},
        "queue": "live",
    }
    for attempt in range(retries):
        resp = requests.post(API_URL, json=body, headers=HEADERS, timeout=180)
        if resp.status_code == 429:
            time.sleep(int(resp.headers.get("retry-after", 10)))
            continue
        resp.raise_for_status()
        data = resp.json()
        if data["job"]["status"] == "completed":
            return data["result"]
        if "retry" in (data["job"]["error"] or "").lower():  # "busy" and "temporarily blocked" both say retry
            time.sleep(5 * (attempt + 1))
            continue
        raise RuntimeError(data["job"]["error"])
    raise RuntimeError("Search failed after retries")


def domain(link):
    return (urlparse(link).hostname or "").removeprefix("www.")


def summarize(result):
    aio = result["ai_overview"]
    return {
        "organic": len(result["organic_results"]),
        "ai_overview": aio is not None,
        "cited_domains": {domain(s["link"]) for s in aio["sources"]} if aio else set(),
        "people_also_ask": len(result["people_also_ask"]),
        "local_pack": len(result["local_pack"]),
    }


def compare_devices(query, country="us"):
    desktop = summarize(google_search(query, country, "desktop"))
    mobile = summarize(google_search(query, country, "mobile"))
    shared = desktop["cited_domains"] & mobile["cited_domains"]
    return {"desktop": desktop, "mobile": mobile, "cited_in_both": len(shared)}

For best vpn and pizza in brooklyn ny it printed:

best vpn
  desktop  organic  9 | overview True  | cited  6 domains | PAA 4 | local pack 0
  mobile   organic 11 | overview True  | cited  7 domains | PAA 0 | local pack 0
  cited in both: 6
pizza in brooklyn ny
  desktop  organic  9 | overview False | cited  0 domains | PAA 4 | local pack 3
  mobile   organic 14 | overview False | cited  0 domains | PAA 4 | local pack 0
  cited in both: 0

The pizza in brooklyn ny run shows the local pack gap in one search: three businesses on desktop, none on mobile, and 14 organic results on mobile against 9 on desktop. Also note best vpn: on this run People Also Ask was present on desktop and absent on mobile, which is the kind of single-search difference a larger count smooths out.

What to do with this

• Track the device that your audience uses. If most of your visitors are on phones, a desktop rank is the wrong number. Store the device with every check, as the mobile vs desktop guide explains.
• Do not hard-code a results-per-page. It was 7.6 on desktop and 9.5 on mobile here.
• Handle an overview with no sources. Check that sources has items before you read it.
• Use the Local API for local data. It does not rely on the local pack in a search response.
• Read single checks as samples. Overview presence moved between two desktop runs, so judge a device gap over many checks, not one.

What it costs

Every search is 1 credit, on every queue and device. Starter is $1.00 per 1,000 requests and the largest pack is $0.30 per 1,000, see pricing.

Failed requests are not charged.

Next steps

To look at a single search before building anything, our free AI Overview checker runs without an account and shows the AI Overview, its sources and the top results for the country and device you pick.

You now have numbers on how AI Overviews, organic results and the local pack compare between mobile and desktop for 40 searches, and a function that compares any search on both. Which sites AI Overviews cite covers the sources, and the Google Search API page lists every parameter with a full example response.