Google assists SERP Click-Through Optimization
Big Mama Google in her ongoing campaign to keep her search index clean assists Webmasters with reports allowing click-trough optimization of a dozen or so pages per Web site. Google launched these reports a while ago, but most Webmasters didn't make the best use of them. Now that Vanessa has revealed her SEO secrets, lets discuss why and how Google helps increasing, improving, and targeting search engine traffic.
Google is not interested in gazillions of pages which rank high for (obscure) search terms but don't get clicked from the SERPs. This clutter tortures the crawler and indexer, and it wastes expensive resources the query engine could use to deliver better results to the searchers.
Unfortunately, legions of clueless SEOs work hard to increase mount clutter by providing their clients with weekly ranking reports, what leads to even more pages which rank for (potentially money making) search phrases but appear on the SERPs with such crappy titles and snippets that not even a searcher coming with an IQ slightly below a slice of bread clicks them.
High rankings don't pay the bills, converting traffic from SERPs on the other hand does. A nicely ranking page is an asset, which in most cases just needs a few minor tweaks to attract search engine users (mount clutter contains machine generated cookie-cutter pages too, but that's a completely other story).
For example unattended pages gaining their SERP position from anchor text of links pointing to them often have a crappy click through rate (CTR). Say you've a page about a particular aspect of green widgets, which applies to widgets of all colors. For some reason folks preferring red widgets like your piece and link to it with "red widgets" as anchor text. The page will rank fine for [red widgets], but since "red widgets" is not mentioned on the page this keyword phrase doesn't appear on the SERP's snippets, not to speak of the linked title. Search engine users seeking for information on red widgets don't click the link about green widgets, although it might be the best matching search result.
So here is the click-thru optimization process based on Google's query stats (it doesn't work with brand new sites nor more or less unindexed sites, because the data provided in Google's Webmaster Tools are available, reliable and quite accurate for somewhat established sites only):
Login, choose a site and go to query stats. In an ideal world you'll see two tables of rather identical keyword lists (all examples made up).
The "Top search queries" table on the left shows positions for search phrases on the SERPs, regardless whether these pages got clicks or not. The "Top search query clicks" table on the right shows which search terms got clicked most, and where the landing pages were positioned on their SERPs. If good keywords appear in the left table but not in the right one, you've CTR optimization potentials.
The "average top position" might differ from todays SERPs, and it might differ for particular keywords even if those appear in the same line in both tables. Positioning fluctuation depends on a couple of factors. First, the position is recorded at the run time of each search query during the last 7 days, and within seven days a page can jump up and down on the SERPs. Second, positioning on for example UK SERPs can differ from US SERPs, so an average 3rd position may be a utterly useless value, when a page ranks #1 in the UK and gets a fair amount of traffic from UK SERPs, but ranks #8 on US SERPs and searchers don't click it because the page is about a local event near Loch Nowhere in the highlands. Hence refine the reports by selecting your target markets in "location", and if necessary "search type" too. Third, if these stats are generated based on very few searches and even fewer click throughs, they are totally and utterly useless for optimization purposes.
Lets say you've got a site with a fair amount ofGoogle search engine traffic, the next step is identifying the landing pages involved (you get only 20 search queries, so the report covers only a fraction of your site's pages). Pull these data from your referrer stats, or extract SERP referrers from your logs to create a crosstab of search terms from Google's reports per landing page. Although the click data are from Google's SERPs, it might make sense to do this job with a broader scope, that is including referrers from all major search engines.
Now perform the searches for your 20 keyword phrases (just click on the keywords on the report) to check how your pages look at the SERPs. If particular landing pages trigger search results for more than one search term, extract them all. Then load your landing page, and view its source. Read your page first rendered in your browser, then check out semantic hints in the source code, for example ALT or TITLE text and stuff like that. Look at the anchor text of incoming links (you can use link stats and anchor text stats from Google, We Build Pages Tools, ...) and other ranking factors to understand why Google thinks this page is a good match for the search term. For each page, let the information sink before you change anything.
If the page is not exactly a traffic generator for other targeted keywords, you can optimize it with regard to a better CTR for the keyword(s) it ranks for. Basically that means use the keyword(s) naturally on all page areas where it makes sense, and provide each occurence with a context which hopefully makes it into the SERP's snippet.
Make up a few natural sentences a searcher might have in mind when searching for your keyword(s). Write them down. Order them by their ability to fit the current page text in a natural way. Bear in mind that with personalized search Google could have scanned the searcher's brain to add different contexts to the search query, so don't concentrate too much on the keyword phrase alone, but on short sentences containing both the keyword(s), respectively their synonyms, and a sensible context as well.
There is no magic number like "use the keywords 5 times to get a #3 spot" or "7 occurences of a keyword gain you a #1 ranking". Optimal keyword density is a myth, so just apply common sense by not annoying human readers. One readable sentence containing the keyword(s) might suffice. Also, emphasizing keywords (EM/I, STRONG/B, eye catching colors ...) makes sense because it helps catching the attention of scanning visitors, but don't over-emphasize because that looks crappy. The same goes for H2/H3/... headings. Structure your copy, but don't write in headlines. When you emphasize a word or phrase in (bold) red, then don't do that consistently but only in the most important sentence(s) of your page, and better only on the first visible screen of a longer page.
Work in your keyword+context laden sentences, but -again!- do it in a natural way. You're writing for humans, not for algos which at this point already know what your page is all about and rank it properly. If your fine tuning gains you a better ranking that's fine, but the goal is catching the attention of searchers reading (in most cases just skimming) your page title and a machine generated snippet on a search result page. Convince the algo to use your inserted sentence(s) in the snippet, not keyword lists from navigation elements or so.
Write a sensible summary of the page's content, not more than 200-250 characters, and put that into the description meta tag. Do not copy the first paragraph or other text from the page. Write the summary from scratch instead, and mention the targeted keyword(s). The first paragraph on the page can exceed the length of the meta description to deliver an overview of the page's message, and it should provide the same information, preferably in the first sentence, but don't make it longish.
Check the TITLE tag in HEAD: when it is truncated on the SERP then shorten it so that the keyword becomes visible, perhaps move the keyword(s) to the beginning, or create a neat page title around the keyword(s). Do title changes very carefully, because the title is an important ranking factor and your changes could result in a ranking drop. Some CMSs change the URL without notice on changes of the title text, and you certainly don't want to touch the URL at this point.
Make sure that the page title appears on the page too. Putting the TITLE tag's content (or a slight variation) in a H1 element in BODY cannot hurt. If you for some weird reasons don't use H-elements, then at least format it prominently (bold, different color but not red, bigger font size ...).
If the page performs nice with a couple money terms and just has a crappy CTR for a particular keyword it ranks for, you can just add a link pointing to a (new) page optimized for that keyword(s), with the keyword(s) in the anchor text, preferably embedded in a readable sentence within the content (long enough to fill two lines under the linked title on the SERP), to improve the snippet. Adding a (prominent) link to a related topic should not impact rankings for other keywords too much, but the keywords submitted by searchers should appear in the snippet a short while after the next crawl. In such cases better don't change the title, at least not now. If the page gained its ranking solely from anchor text of inbound links, putting the search term on the page can give it a nice boost.
Make sure you get an alert when Ms. Googlebot fetches the changed pages, and check out the SERPs and Google's click stats a few days later. After a while you'll get a pretty good idea of how Google creates snippets, and which snippets perform best on the SERPs. Repeat until success.
Related post: Google Quality Scores for Natural Search Optimization by Chris Silver Smith
Google is not interested in gazillions of pages which rank high for (obscure) search terms but don't get clicked from the SERPs. This clutter tortures the crawler and indexer, and it wastes expensive resources the query engine could use to deliver better results to the searchers.
Unfortunately, legions of clueless SEOs work hard to increase mount clutter by providing their clients with weekly ranking reports, what leads to even more pages which rank for (potentially money making) search phrases but appear on the SERPs with such crappy titles and snippets that not even a searcher coming with an IQ slightly below a slice of bread clicks them.
High rankings don't pay the bills, converting traffic from SERPs on the other hand does. A nicely ranking page is an asset, which in most cases just needs a few minor tweaks to attract search engine users (mount clutter contains machine generated cookie-cutter pages too, but that's a completely other story).
For example unattended pages gaining their SERP position from anchor text of links pointing to them often have a crappy click through rate (CTR). Say you've a page about a particular aspect of green widgets, which applies to widgets of all colors. For some reason folks preferring red widgets like your piece and link to it with "red widgets" as anchor text. The page will rank fine for [red widgets], but since "red widgets" is not mentioned on the page this keyword phrase doesn't appear on the SERP's snippets, not to speak of the linked title. Search engine users seeking for information on red widgets don't click the link about green widgets, although it might be the best matching search result.
So here is the click-thru optimization process based on Google's query stats (it doesn't work with brand new sites nor more or less unindexed sites, because the data provided in Google's Webmaster Tools are available, reliable and quite accurate for somewhat established sites only):
Login, choose a site and go to query stats. In an ideal world you'll see two tables of rather identical keyword lists (all examples made up).
| Top search queries | Avg. Pos. | Top SERP clicks | Avg. Pos. |
|---|---|---|---|
| 1. web site design | 5 | 1. web site design | 4 |
| 2. google consulting | 4 | 2. seo consulting | 5 |
| 3. seo consulting | 3 | 3. google consulting | 2 |
| 4. web site structures | 2 | 4. internal links | 3 |
| 5. internal linkage | 1 | 5. web site structure | 3 |
| 6. crawlability | 3 | 6. crawlability | 5 |
The "Top search queries" table on the left shows positions for search phrases on the SERPs, regardless whether these pages got clicks or not. The "Top search query clicks" table on the right shows which search terms got clicked most, and where the landing pages were positioned on their SERPs. If good keywords appear in the left table but not in the right one, you've CTR optimization potentials.
The "average top position" might differ from todays SERPs, and it might differ for particular keywords even if those appear in the same line in both tables. Positioning fluctuation depends on a couple of factors. First, the position is recorded at the run time of each search query during the last 7 days, and within seven days a page can jump up and down on the SERPs. Second, positioning on for example UK SERPs can differ from US SERPs, so an average 3rd position may be a utterly useless value, when a page ranks #1 in the UK and gets a fair amount of traffic from UK SERPs, but ranks #8 on US SERPs and searchers don't click it because the page is about a local event near Loch Nowhere in the highlands. Hence refine the reports by selecting your target markets in "location", and if necessary "search type" too. Third, if these stats are generated based on very few searches and even fewer click throughs, they are totally and utterly useless for optimization purposes.
Lets say you've got a site with a fair amount of
Now perform the searches for your 20 keyword phrases (just click on the keywords on the report) to check how your pages look at the SERPs. If particular landing pages trigger search results for more than one search term, extract them all. Then load your landing page, and view its source. Read your page first rendered in your browser, then check out semantic hints in the source code, for example ALT or TITLE text and stuff like that. Look at the anchor text of incoming links (you can use link stats and anchor text stats from Google, We Build Pages Tools, ...) and other ranking factors to understand why Google thinks this page is a good match for the search term. For each page, let the information sink before you change anything.
If the page is not exactly a traffic generator for other targeted keywords, you can optimize it with regard to a better CTR for the keyword(s) it ranks for. Basically that means use the keyword(s) naturally on all page areas where it makes sense, and provide each occurence with a context which hopefully makes it into the SERP's snippet.
Make up a few natural sentences a searcher might have in mind when searching for your keyword(s). Write them down. Order them by their ability to fit the current page text in a natural way. Bear in mind that with personalized search Google could have scanned the searcher's brain to add different contexts to the search query, so don't concentrate too much on the keyword phrase alone, but on short sentences containing both the keyword(s), respectively their synonyms, and a sensible context as well.
There is no magic number like "use the keywords 5 times to get a #3 spot" or "7 occurences of a keyword gain you a #1 ranking". Optimal keyword density is a myth, so just apply common sense by not annoying human readers. One readable sentence containing the keyword(s) might suffice. Also, emphasizing keywords (EM/I, STRONG/B, eye catching colors ...) makes sense because it helps catching the attention of scanning visitors, but don't over-emphasize because that looks crappy. The same goes for H2/H3/... headings. Structure your copy, but don't write in headlines. When you emphasize a word or phrase in (bold) red, then don't do that consistently but only in the most important sentence(s) of your page, and better only on the first visible screen of a longer page.
Work in your keyword+context laden sentences, but -again!- do it in a natural way. You're writing for humans, not for algos which at this point already know what your page is all about and rank it properly. If your fine tuning gains you a better ranking that's fine, but the goal is catching the attention of searchers reading (in most cases just skimming) your page title and a machine generated snippet on a search result page. Convince the algo to use your inserted sentence(s) in the snippet, not keyword lists from navigation elements or so.
Write a sensible summary of the page's content, not more than 200-250 characters, and put that into the description meta tag. Do not copy the first paragraph or other text from the page. Write the summary from scratch instead, and mention the targeted keyword(s). The first paragraph on the page can exceed the length of the meta description to deliver an overview of the page's message, and it should provide the same information, preferably in the first sentence, but don't make it longish.
Check the TITLE tag in HEAD: when it is truncated on the SERP then shorten it so that the keyword becomes visible, perhaps move the keyword(s) to the beginning, or create a neat page title around the keyword(s). Do title changes very carefully, because the title is an important ranking factor and your changes could result in a ranking drop. Some CMSs change the URL without notice on changes of the title text, and you certainly don't want to touch the URL at this point.
Make sure that the page title appears on the page too. Putting the TITLE tag's content (or a slight variation) in a H1 element in BODY cannot hurt. If you for some weird reasons don't use H-elements, then at least format it prominently (bold, different color but not red, bigger font size ...).
If the page performs nice with a couple money terms and just has a crappy CTR for a particular keyword it ranks for, you can just add a link pointing to a (new) page optimized for that keyword(s), with the keyword(s) in the anchor text, preferably embedded in a readable sentence within the content (long enough to fill two lines under the linked title on the SERP), to improve the snippet. Adding a (prominent) link to a related topic should not impact rankings for other keywords too much, but the keywords submitted by searchers should appear in the snippet a short while after the next crawl. In such cases better don't change the title, at least not now. If the page gained its ranking solely from anchor text of inbound links, putting the search term on the page can give it a nice boost.
Make sure you get an alert when Ms. Googlebot fetches the changed pages, and check out the SERPs and Google's click stats a few days later. After a while you'll get a pretty good idea of how Google creates snippets, and which snippets perform best on the SERPs. Repeat until success.
Related post: Google Quality Scores for Natural Search Optimization by Chris Silver Smith
Labels: Anchor Text, CTR, Google, SEO, Webmaster Central
 Stumble It! |
 Post it to del.icio.us |
posted by Sebastian @ Wednesday, July 04, 2007 · PERMANENT LINK
-->
Subscribe to my blog
2 Comments:
At Wednesday, July 04, 2007,
JLH said…
That is an excellent step-by-step guide. The key to any successful experiment is the ability to change the variables and then observe the effects of that change. Google's webmaster stats gives an great feedback, as you mentioned. Without feedback, you're just changing things for change sake.
I like most have been doing this sort of analysis of referrels for a while, but without the benefit of knowing the search queries that wern't being clicked (those obviously don't show up in your logs).
Good work by you for discussing it so well, and good work by Google (Vanessa) for making it possible.
At Wednesday, July 04, 2007,
Sebastian said…
Thanks for the kind words. While passing our compliments to Vanessa and the Webmaster Central team, we should ask for an unlimited list fetchable via Web service for example, so that developing a tool supporting this process makes sense. I'd rather deal with a huge database table than with a tiny array of max. 20 elements.
Post a Comment
<< Home