Ad observations vs Auction Insights and impression share: how to read them together

Key takeaways

• Auction Insights and impression share describe auctions your own ads were eligible for.

• External observations describe which ads appeared for a chosen search, location, device and time.

• The denominators differ, so the percentages will not match and should never be combined.

• Name the source next to every percentage in a report.

• Neither source shows a competitor's spend or the keyword it bought.

A monitoring report says a competitor appeared in 45% of your saved brand checks. Your Google Ads Auction Insights report shows the same competitor with a 20% overlap rate. Neither number is wrong. They answer different questions.

What Auction Insights measures

Google's Auction Insights report compares your performance with other advertisers in the same auctions. For Search campaigns, Google lists six metrics (Google Auction Insights):

• Impression share: impressions you received divided by the estimated impressions you were eligible to receive.

• Overlap rate: how often another advertiser's ad received an impression when yours also did.

• Outranking share: how often your ad ranked higher than another advertiser's, or showed when theirs did not.

• Position above rate: how often the other advertiser's ad was shown above yours.

• Top of page rate: how often an ad appeared above the unpaid results.

• Absolute top of page rate: how often an ad was the very first ad.

Shopping campaigns show only impression share, overlap rate and outranking share. Performance Max offers segmented data for Search and Shopping ads on the Search Network.

Three limits are easy to forget. Google says the report does not show insights when impression share is below 10%. It requires a minimum threshold of activity for the selected period. And a competitor's impression share in your report may differ from what that competitor sees in its own account, because your eligible impressions only partly overlap with theirs.

What impression share measures

Google defines impression share as the impressions your ad received divided by the total impressions it was eligible to receive. Lost impression share is split into budget (your ads did not show because the budget ran out) and rank (your ads did not show because of Ad Rank) (Google impression share definitions). The denominator is built from your eligibility: your keywords, your targeting, your schedule. It says nothing about auctions you were not eligible for.

Microsoft Advertising uses similar metrics

Microsoft Advertising reports a comparable set for Bing: impression share, overlap rate, position above rate (called AboveRate in the API), top of page rate, absolute top of page rate and outranking share (Microsoft Advertising AuctionInsightKpi). Microsoft describes top of page as the mainline position above organic results. The same caution applies: the numbers describe auctions from your account's point of view.

What an external ad observation measures

An observation records the result of one check: a specific search, run on one engine, from one location, on one device, at one moment. It captures which ads appeared, their text, displayed and destination URLs, and their order. That makes observations good at questions account metrics cannot answer:

• What exact copy and offer did a competitor, retailer or affiliate show on my brand search?

• Which advertisers appear on searches where my own ads were not eligible or not running?

• Where did each ad send people?

• Does the picture differ between Chicago and Seattle, or between mobile and desktop?

Observations also have clear limits. They do not reveal an advertiser's spend, bids, impression volume or the keyword it bought. Google's matching rules let an ad appear on searches beyond its literal keyword (Google keyword matching), so a monitored search is not proof of purchase. And a scheduled check samples the results at certain times; it does not see activity between checks.

Why the percentages differ: side by side

A small arithmetic example

Suppose, in an invented example, you ran 20 completed checks for "northstar boots" in one week: 10 on desktop and 10 on mobile, in Denver. A reseller, trail-deals.example, appeared in 9 of them. Its share of checks with an ad is 9 out of 20, or 45%.

Now suppose your own brand campaign was paused on mobile that week. In your Auction Insights report, the reseller can only overlap with your desktop impressions, and only in auctions where your campaign was eligible. Its overlap rate may be 20% or it may not appear at all if thresholds are not met. Both numbers can be accurate. Neither can be converted into the other by relabeling a column, and neither tells you the reseller's spend.

A reporting framework that keeps both honest

1. Name the source next to every percentage. Write "share of monitored checks" or "Google Ads overlap rate", never just "share".

2. Match scope where possible, and state it where not. Same dates, same device split, same market. If the account covers a country and monitoring covers three cities, say so.

3. Use account metrics for your own performance. Lost impression share by budget or rank belongs to your campaign decisions.

4. Use observations for what appeared. Quote the ad, the destination and the advertiser relationship.

5. Do not build a combined score. Two measures with different denominators averaged together mean nothing.

6. Treat changes as questions. If a competitor appears more often, first check whether you added locations, devices, keywords or frequency, and whether checks failed during the period.

When a trend moves

Suppose observations show a new coupon site appearing on brand searches, and your Auction Insights shows your top of page rate falling in the same weeks. That is worth investigating, but it is not proof that one caused the other. Look at actual examples, check your own bids, budgets and quality changes, and write down the next test that could separate the explanations.

How Ads Lighthouse labels its numbers

Ads Lighthouse is an observation tool. It runs scheduled and on-demand checks on Google and Bing for your monitored keywords across chosen locations and devices, and saves each ad with its context. In the paid results explorer, the Visibility column is defined as the share of observations in the current selection where a domain showed an ad. The product labels it as describing your sample, not Google impression share. Filters for device, operating system, country and domain change the selection and therefore the denominator, and the keyword and advertiser reports let you open the underlying saved ads. Exports to CSV or Excel keep the rows you are looking at, so the numbers you share match the screen.

Ads Lighthouse does not import Auction Insights into these reports. Keep your account data in Google Ads and Microsoft Advertising, and present the two side by side with their definitions. The paid results help article explains how the Visibility column is calculated, the PPC teams use case shows how observations fit next to account data, and the Google Ads and Microsoft Advertising (Bing) monitoring pages describe coverage per engine. The glossary collects the terms used in this guide.

See the product overview and pricing (from $49 per month, 7-day free trial). Related guides: what to put in a weekly brand search report, desktop, mobile and location coverage and a weekly competitor ad copy workflow.