Over 4500 interventions to improve Google Search in 2020
Google’s mission is to “organize information from all over the world and make it accessible and useful globally”, and to achieve this ambitious goal it is necessary a continuous and constant work to improve the search system: and so, if a few years ago 3200 interventions were needed to improve the engine, the continuous evolution of the Web and the needs of users have led to a further increase in this work, culminating in over 4500 changes made in 2020.
Google’s improvement work in 2020
The news comes directly from Danny Sullivan, Public Liaison for Search, who wrote an article on The Keyword blog to present the results of this Google Search update work, oriented to respond to the evolving needs and expectations of people using Google.
To be precise, in 2020 alone Google performed:
- 887 launches in search results.
- 323 real-time traffic experiments.
- 605 tests on search quality
- 937 experiments side by side
How Google enhancements work and what they do
The nearly 5,000 launches – 40 percent more than in 2018 and ten times more than in 2010, when Google declared a rate of about one change per day – concern ranking interventions (such as broad core updates), user interface changes and much more. The numbers show that the company has accelerated efforts on this front over the years and is increasing the speed with which it updates Google Search.
Before being actually published and shown to users, every single change proposed for the Search is subjected to a strict evaluation procedure, which serves to carefully analyze data and metrics from various experiments to determine whether it really improves the service for people (otherwise it is not launched). And so, in the face of more than 600 thousand experiments, only (so to speak) 4887 have actually resulted in improvements to the Research.
Real-time traffic experiments are used to test how real people interact with a certain feature before launching it for everyone. Google looks at a long list of metrics – for example what people click on, how many searches have been completed and if some have been canceled, how long it took people to click on a result and so on – and use these results to assess whether user involvement with the new functionality can ensure that changes improve the relevance and usefulness of the results.
Search quality tests are carried out thanks to