Back to the Campfire Blog

How Google Measures Success: Return on Search Page

Written by: Noah Learner Tags: critical thinking, analytics, future of seo

Published: Jul 3, 2025

I spent a lot of time at Google IO having my mind blown by some of the best SEOs in the game, and direct access to some of the public-facing reps at Google.

Jean-Christophe Chouinard believes that Google measures the success of search pages (SERPs) differently than websites do.

We all care about our individual websites' clicks and CTR, while Google cares about performance for an entire set of results or search journey. 

And we should care primarily about conversions, and the actions that lead up to it, because that's where revenue comes from and means we're building a product or service that people are buying. 

Having our content crawled, indexed, ranked, seen, and clicked all have to happen before a conversion, and we measure our investment primarily by the cost to produce it and the length of time to break even on that cost and then start returning a profit from that page. 

So I started thinking about Google's perspective...

Listen to my conversation with host Jon Clark on Page 2 Podcast about ROSP:

How would Google view successful search?

This took me down a bit of a wormhole to explore this from Google's perspective.

I have to thank Cindy Krum for reminding me to look at things from Google's business goals in mind.

A quick note: I use We and Us in the following section as if I'm in Google's shoes.

If I worked at Google, what KPIs would I look at?

• Revenue driving Clicks [AD clicks]/impressions
• Time on page - how long did user's interact with my product
• # of actions taken: scroll, button clicks, page stickiness
• Comparison of clicks vs what we expect by position - using this to determine reranking in future based on performance by location on page
• Daily Active Users
• Monthly Active Users
• Revenue / User
• Clicks / User
• Pogo sticking
• Cost to produce a SERP
• ROSP - Return on Search Page
• ROSPB - Return on Search Page Buckets

The last two bullet points are now living rent free in my head.

ROSP or Return on Search Page

Return on Search Page is:

Revenue Driven / Cost to produce a Search Page

I coudn't help but wonder if Google engineers tweak the results in order to achieve a specific multiple for ROSP – for example, I used to target a 12:1 Return on ad spend (ROAS) with bike shop ads. 

And what stops it from trying to achieve a higher and higher ROSP?

Especially if it's already crossed the creepy line?

The creepy line refers to a quote by then-CEO Eric Schmidt all the way back from 2011 that promised that Google wouldn't do anything a regular person would consider creepy – yet they were already tracking our every move at the time:

"There's what I call the 'creepy line,' and the Google policy about a lot of these things is to get right up to the creepy line but not cross it," Eric Schmidt said in a not-at-all creepy way

No, I think it would be a bucket of engagement metrics thout would each need to be in a specific section of a curve of values; each metric would need to be in a given range.

ROSP Framework

Goal: Keep targeted engagement metrics in specific ranges while accomplishing maximum ROSP.

What would Google's implementation strategy look like?

• Start with bare AIOs to establish a baseline for user metrics
• Layer in rich results + Ads
• Compare to what we know/expect from past experience with Ai Overviews and our historical knowledge of user engagements
• Rinse and repeat

Accomplishing ROSP means different conditions need to be met for different components:

For users

• Users must trust answers
• Users must have a delightful search experience
• We must provide useful results
• We must provide fast results
• We must provide accurate results

Google infrastructure

• Reduce production costs of SERPs
• Reduce compute power needed
• Reduce storage costs
• Increase caching
• Increase % of queries that get mapped to cached results
• Increase insights - use Chrome data more, get more people using Gemini (and every other Search surface) to get more user Data
• Reduce LLM model training / production costs
• Reduce crawling costs

Content producers

• Get Content Producers to engage with us, providing the inputs that feed our outputs
• Show producers they matter with:
  • Search Central events
  • Search Relations team helping on socials
  • Provide a Podcast to help website owners
  • Provide Videos to help website owners
  • Provide helpful docs to help web professionals + site owners
  • Seem devoted to helpful, unique content

UX decisions

Increase screen size of features likely to increase exposure to revenue generation:

• Bigger taller ads
• Product grids (paid + organic)
• AI Overviews with tiny links
• AI Images + AI Overviews have links that trigger more searches

As you create content for long-tail queries and invest in PR outreach to gain traction in LLMs, remember that Google has a dog in the return-on-search-investment fight too.