Forum Moderators: open
I've optimized a large, database-generated site whose CGI URLs have been converted to fake directories. The site contains information organized by state and then by city. The states are second level pages (ie, one click from the home page), and the cities are third level pages (two clicks from the home page).
The arrangement has worked fine, but many of the third level city pages are accessible only via pull-down menus on the state pages, which even Googlebot can't follow.
Several months ago, to help the spiders, I had text links added from the second level state pages to a half-dozen or so of the most Important City pages, and these city pages ranked well.
In an attempt to get all of the city pages at least spidered, we decided that, for each state, in addition to linking to these half-dozen most Important City pages, we'd create a mini state sitemap, a links page to the other city pages within the state.
We linked to this state sitemap from the appropriate state page. The newly linked-to city pages would actually be fourth level pages, if you counted html link clicks.
The Important City pages continued to have direct html links from the appropriate state pages, as well as from the new state sitemap.
Now, what's happened is that, since the last reindex, some of the Important City pages that had ranked well have now dropped from view. I can understand that if I'd put all of the city links (and there are a lot of them) on the state page, the conferred Page Rank to the Important Cities would have been dissipated among a bunch of links. That's not what we did, though. We added just one additional link to each state page.
I'm wondering, though, if somehow the sitemap pages might have had some dissipating effect. Some of them are quite large. There were other changes in the reindex that could have contributed... but I'm trying to sort out whether to yank the sitemaps.
I hope this is clear enough to follow. Rather than saying "upstream pages" in the post title, I might have more accurately said "parallel pages," but that might not have been clear either.
Also, any chance that you could have diluted your theming following the change?
I don't think that's the problem, Robert. Do the State pages rank as well as before?
Generally yes. There's a lot going on, though, so it's hard to isolate the factors (and unfortunately I haven't tracked every phrase for every month, since theoretically I'm finished with the project).
Looking at one of our main phrases, "widget information," both by itself, and also with State and City modifiers for one of our bellwether state/city combinations...
- the home page, for "widget information," without state or city modifiers, has gone down from a #5 to a #12.
- for "StateName widget information," the sample State page is now #4.
- one of the Important City pages in that state, for "CityName widget information," has dropped from #4 to #25. This is what's worrying me.
- for "StateName widgets," the State page that links to that City page has moved up from #3 to #2.
Possibly an important factor, one word has been added to the City page titles, spacing "CityName" further away from "Widget Information" by that one word. I've had this word removed, and hope we'll see these changes in the next update.
> word has been added to the City page titles, spacing "CityName" further away from "Widget Information" by that one word.
I would expect that to make a significant difference. Proximity is important in Google, even a stop word such as "in" can change things considerably.
Thanks... That's why I like to keep my titles short. I know that proximity is important to Google on the page as well. While people say that density isn't important, I find it amazing what tuning a page for proximity can do.
Bottom line, I think you're saying that the site-map city linking page is not a factor. Robber is suggesting that the Page Rank is "flowing" down to more pages... but it's my feeling that the Page Rank that is flowing from the city links page is essentially a split of the PR coming only from the one link to that page... and that it can't be draining the State page.
Robert, are any of the special city pages linked to from anywhere besides the main state pages? From anywhere at the root level?
for "StateName widgets," the State page that links to that City page has moved up from #3 to #2.
How does the link text pointing to the city page match up to what's on the state page and the phrasing used on the city page it's linking to?
And how are the individual city pages linked, like with each other?
I agree, it should have the same affect as adding another city to the State page would.
No. The root level only links to State pages. There are way too many City pages to link to from anywhere except the State pages... and even there we had to go to the site-map type arrangement to make all of the City pages accessible for spidering.
>>How does the link text pointing to the city page match up to what's on the state page and the phrasing used on the city page it's linking to?
The link text on the State page that's pointing to the City page only contains the city name. If you think about it, I couldn't have a half-dozen adjacent links all in the form "CityName widget information." The repetition of "widget information" would get me in trouble. In addition, I'm targeting more phrases on the City pages than simply "CityName widget information." It's really like "CityName widgets - red and blue widget information." I had to push the titles more than I wanted to, because this was our only shot at these combinations. And up till this month, they worked well.
When I set up the targets I checked Overture and Google AdWords to see which of their main targets were most commonly searched with states and which with cities, and set up the pages with these in mind.
I should mention that each bit of content I added to the site had to pass an editorial review, a marketing review, and a usability review. ;) I convinced them to make "Welcome" a graphic.
>>And how are the individual city pages linked, like with each other?
In terms of incoming links to a City page, the text links I described are the only ones. There are global outgoing navigation links as chosen by the usability committee, and a few optimizing concessions to me, but no cross-linking among these city pages. I don't think that cross-linking would in fact help these pages.
>>I agree, it should have the same affect as adding another city to the State page would.<<
Thanks. I appreciate the confirmation.
>>Properly worded breadcrumb trails should be a pretty good way to echo back some of the pagerank, and, more importantly, put keywords into anchor text.<<
Again, with the City pages, only the city names were possible. All of the City pages are linked back to their State pages.
I think about that very often. "[big]CityName[/big]widget information" is one possibility. Is it wrong to do that? I wouldn't think so, as each link is properly describing its destination. Some people suggest that <A title="description of resource" href="..."> is recognised by Google as if it was link text. I haven't seen it tested.
> I had to push the titles more than I wanted to, because this was our only shot at these combinations. And up till this month, they worked well.
Some time ago I changed from trying to get word combinations in titles (and body text and link text) to trying to get phrases in titles (and body text and link text). I firmly believe that it makes a large difference.
calum - I'm not sure I understand what you're suggesting here... Putting a bold font in your link? How does this relate to the question of repetition of "widget information"?
>>Some people suggest that <A title="description of resource" href="..."> is recognised by Google as if it was link text. I haven't seen it tested.<<
I'd like to see a whole discussion of the title attribute, but that's another thread. I'd assume that there would still be a question of close repetition, though, if each of the title attributes of many CityName links also included "widget information," assuming Google would treat repetitions in these attributes as it would treat repetitions in other text.
>>Some time ago I changed from trying to get word combinations in titles (and body text and link text) to trying to get phrases in titles (and body text and link text). I firmly believe that it makes a large difference.<<
I think you're right that this would be ideal, and that it's a good principle in general. On this site, though, there'd be no way to create multiple City pages, or even multiple State pages, for each unique phrase.
I did searches for some of the City pages that only come off of the site map page, where I'd expect almost complete dissipation of Page Rank going to these. On the obscure cities, all of the targeted title phrases give excellent rankings.
As the cities get larger and more competitive, it's necessary to search for closer matches to the phrasing on the titles for the pages to rank. I'm hoping that the next update will show that removing that extra word we put in between "CityName" and "widget information" boosts our rankings... not an exact match, but at least a closer one.
On the very large cities, like Los Angeles, for example, I can't get rankings on the City pages even with very close matches. These will most definitely need external links coming into these pages to rank.
External links so far have been mainly to the home page, the States pages, and to some internal pages. The better the linkage, the more the searches can vary from the title order.
On the State pages that are really well linked, some of the modifying words in a search don't even have to be in the title, as long as they appear on the page. The titles are still important on these higher PR pages, though, in telling the algo basically what the page is about.
Repetition of a target word in a title is still something I'm not comfortable with. I used to think that you couldn't repeat a word in a title at all. Now I'm feeling that Google, at least, will allow repetitions in the title, perhaps closer reps even than they allow on a page.
One of the things I'm seeing, incidentally, is that for "widget information" by itself, is that Google's database size on this phrase is varying greatly... and we're further down with the bigger base than we are with the smaller one.
Just that if this looks too repetitive:
CityA widget information
CityB widget information
CityC widget information
...
then this looks a little better, IMO:
[big]CityA[/big] widget information
[big]CityB[/big] widget information
[big]CityC[/big] widget information
...
In practice, "widget information" could be in a lighter colour, maybe under CityName instead of next to it.
Browse states for widget information: StateA, StateB, StateC
I don't know how much Google considers link context and thus don't know whether getting "widget information" nearby would help.
The other thing that I don't think we've mentioned here is that nearby text is believed to count; whether it counts as much as link text, I don't know.
What I mean by this is that if you have too many repetitions of a target word close together, the engine looks upon this as an attempt to spam and demotes you in the serps. They would do the same for excessive density. Whether you call this a penalty is, I suppose, a semantic question. Essentially, too many reps is not satisfying the algo. I assume what I'm thinking of, whatever you call it, would be automatically applied.
On page body text, I usually try to keep 4 or 5 words in between repetitions of my target words (not always possible), and not to have my density excessive.
<Repetition discussion carried on in the Avoiding excessive repetition in global text links [webmasterworld.com] thread.>
[edited by: ciml at 12:09 pm (utc) on July 20, 2002]
[edit reason] Link Added [/edit]
To pass on some follow-up information... Google has just started listed the City pages with the one added word removed from the title, and the pages are back to ranking as normal. This suggests that you shouldn't split target phrases by too much in your titles, and that, as calum suggests, the ideal title situation is an exact phrase match.
Also, it's interesting to note that Google made this change prior to their monthly update, further suggesting that they are constantly adjusting the index, at least on high PR sites, which this one is.
The titles and generally good rankings on these pages have held for a couple of days now, but today I'm seeing all sorts of fluctations, not major changes, in the rankings, along with some surprisingly large changes in database size... even more than the ongoing kind of stuff I've been getting used to in the last couple of months.
