What do you think is the most important metric to track in Google Search and why?

Goal: The goal of Google Search is to organize world's information and make it universally accessible.

User Segmentation:

- Users: Who use search to find information (students, researchers, casual searcher)

- Businesses - Who use search to drive traffic to their businesses (eCommerce, Local Businesses)

For this question, I'll choose to address the key metrics for the Users:

Pain Points / Use Cases:

- To find relevant information (better search results based on user intent)

- To find information efficiently (to be able to quickly understand what the links are leading to)

Based on the use cases of the users, we can define the most important metrics to track.

Metrics to Track:

%age of exits - how many people could not find any relevant search results and just left the page

CTRs - a higher CTR indicates higher relevance of the results

Number of results loaded - if the user couldn't find relevant results in top 10 and goes on to explore more results.. This again indicates irrelavant top results

Time spent on SERPs - to identify if the users are able to find relevant information quickly

Out of these, I would choose %age of Exits as the most important Metric as it summarizes all the other sub metrics and it has a direct impact on the business.

Would be happy to receive any feedback on this :)

How Search works

Go to google > search box > dropdown > click on autosuggets query OR keep typing 

> SEARCH button OR I'm feeling lucky (ignore. returns top result and not used heavily)

> search results

1. various pages (1 to N)
2. links in each page ( almost 8 to 10 results)

• User clicks on one link >  spend time > go back
• Ctrl + clicks multiple links and see each one > closes irrelvant one
• if user is satisfied then no further queries
• if not satisfied then similar queries in search again

Inputs for calculatig metrics 

1. time spent on each link for a query
2. number of links clicked for a query
3. Rank of each search result

Approach 1 

1. number of times queried - create group of similar looking queries
2. number of links clicked
3. summation of time spend on each link

(time1 + time2 + ... + timeN) / Number of links clicked / number of times queried

E.g. 

link 1 time  - 10 seconds and link 2 time is 50 seconds user queried only once

(10+50) / 2 / 1  = 60 / 2 = 30  

higher the score better.

Take reciprocal of it = 1/30 = .033

Score nearest to 0 is better 

Approach 2

We should also integrate ranking of links done by Google in the search

1. Take summation of (rank_i * time_i)
2. Divide the result by summation of (time1 to timeN) to get final score

Lower the score better it is (it means top ranked search results are given higher time)

Best Score achived using this method should be 1

But how would you compare 2 different queries with same score but different number of links clicked

• 1/standard deviation(rank1*time1,rank2*time2,rank3*time3,rank4*time4) 

The score near 0 is better as it means higher standard deviation. which means time spent on some pages is very high compared to other pages

Personally, I think second appraoch will serve as most important metric for Google Search