What Panda and Hummingbird Mean for SEO

Two Google updates have changed the rules of search, and they are frequently discussed together despite doing almost opposite things. Panda was a filter that removed low quality pages from contention. Hummingbird was a rebuild of how the engine interprets a query in the first place. Confusing them leads to the wrong response, so it is worth separating them properly.

Panda: a quality filter, not a penalty

Panda arrived in February 2011 and targeted thin, duplicated, and low value pages, particularly content farms producing large volumes of shallow articles to capture search traffic. Google described it as affecting 11.8% of queries, which is enormous for a single change.

The critical detail, and the one most site owners got wrong: Panda operated at the site level, not the page level. A section of weak pages could suppress the rankings of genuinely good pages elsewhere on the same domain. This is why the correct response was often deletion rather than optimization. Removing or consolidating hundreds of thin pages improved the performance of the pages that remained, which feels deeply counterintuitive if you believe more pages means more traffic.

Hummingbird: a new engine, not a new filter

Hummingbird, announced in September 2013, was different in kind. It was a replacement of the core query processing system, described at the time as the biggest overhaul since Caffeine in 2009, and it affected roughly 90% of searches. Almost nobody noticed, which tells you what it did: it was not removing anyone, it was understanding queries better.

The shift is from matching strings to interpreting meaning. A search for "where can I fix my phone screen near me" is not usefully served by finding pages containing those exact words in that order. It requires understanding that the user wants a local repair business, that "phone screen" implies a specific service, and that "near me" is a location signal rather than literal text to match.

This matters increasingly because of how people are starting to search. Voice input produces longer, more conversational queries than typing does. An engine that matches keywords handles those badly. An engine that parses intent handles them well.

What the two updates mean together

1. Inventory every indexed page and mark each one: does this fully answer something a real person searched for?
2. For pages that fail, decide between improving, merging into a stronger page, or removing. Merging is usually the best of the three.
3. Stop writing to a keyword density target. It was never a real ranking factor and after Hummingbird it actively reads as unnatural.
4. Structure pages around questions and the intent behind them, using the words your customers actually use rather than the words your industry uses internally.
5. Cover the surrounding questions on the same page, since the engine now understands topical relationships rather than isolated phrases.

Valter Brandt

Chief Marketing Officer

Valter Brandt is the Chief Marketing Officer of ThisCom, working with clients across the United States and Europe. He has led marketing strategy through the major shifts in social advertising, mobile, content marketing, programmatic media, and marketing automation.

All articles by Valter Brandt →