Where Does Your Traffic Come From?

..and what you read in log analysis programs

Don't think I would put much credence on those numbers for lots of reasons.

Back in the "good old days" when we were all young and posted at WMW, have a look at this thread which is still as useful today as it was then.

Basically you can use log stats for tracking changes, but not giving you absolutes.

AwStats: One particular site with very good rankings in all SE's:

Type-in / Bookmark - 44%
Search engines - 50%
Links - 6%

I, too, like the type-in share to be high, but not too high as the SE traffic is also growth potential. This is a site that has been online for years - i generally find that the SE share decline as you build loyalty. Imho, AwStats is pretty good to identify SE traffic, although it misses a few niche engines (or Google partner sites) so that they get classified as links. But, there are thousands of those and you can't keep up with all.

I think the reason for the low traffic from links is that webmasters tend to link to the front page of a site, while Search Engines link to the deeper pages that are right on topic for a query.

>what log stats are most valuable to you? Unique visitors? Time spent on site? Etc.

first thing I want is a high % looking at more than one page - because we sell mostly high-ish value product I don't always expect actual conversions on the first visit but if they look around then they're interested. Visit length is always an ego-boost but means less in reality because a lot of visits are from people during office hours (Thursday afternoons prime-time) and they can sit one one page for 15 mins while they wander off and do something.

Also the number of people e-mailing a friend or going through the quotes process and/or abandoned baskets (depending on whats being sold) is always interesting.

Plus paths through the site sometimes has me utterly mystified, and has alerted us to some not-very-user-friendly navigation structure in the past, so it's always worth a glance.

Just a couple of points

For IE users you can see how many bookmark by requests for favicon.ico
Flash links do not pass referrer
It's very very difficult to get uniques unless you force login

AwStats does that - calculate favicon as percentage of IE. Too bad AdSense does not count visitors - it would be fun to compare your bookmark rate with your AdSense click rate.

I've read, and my experience seems to bear this out, that favicon.ico is no longer an accurate measure of bookmarks. The first time you visit a site, IE (and FireFox, it seems; and probably others) looks for favicon.ico and places it to the left of the address bar, whether or not you actually bookmark the site. Since this seems to be an easily verifiable fact, then why do people continue to use favicon.ico as a measure of bookmarks?

IE does not take the favicon and place it to the left of the address bar. At least none of the versions of IE that i have used (from version 3 and up).

You can, however, make it do so by dragging the little "e" across the URL and drop it to the right of it but you will have to do that for each favicon that you want displayed there.

Stores favicon this is why I said for IE - firefox downloads it every time so can not measure bookmarking

does anyone know of any stats software that is tackling the bookmark thing differently?

most of my users are IE at the moment, but FF is gaining ground every month.

I personally prefer client side tracking (on-page script) for marketing purpose. After all, I want to know what the users do not what the server do. Logfiles is for server monitoring, client side tracking is for user behaviour and marketing monitoring :)

I only use server side tracking (logfiles or network) if thats the only place I can get the info I need such as spidr tracking, favicon.ico calls etc but for daily use client side tracking does the jo better and faster.

But back on topic...
I work with so many sites that a general break down is just not possible. However, I never like when organic search trafic increae to more than 25%. 50% is just way too dangerous for a media channel you don't control, in my opinion. On most sites I aim for 10-20%.

The rest typically comes from other referrers (including affiliates), repeat users and other forms of marketing (including PPC-engines).

One of the things I really look at is what terms people type into my site search. If people are trying to find something on my site that is not there, and I can add it, I will add it. If there is something on my site and people have to search for it, it can point to a usability issue.

My visitors number is very important, but frankly I want them to do something on my site so I make money...so the biggest metric I watch is the conversion rate. Of the people who did make it to my site, what percentage of them actually did something. (Signed up as a lead, bought something or some other action.)

And, of those people who did something, how did they get to the site? If by search, what search terms did they use? If I can find out what is working, I can expand on it. OTOH, I also look for what is not working and try different things to improve that.

Was there an 'almost conversion'? Did someone start an action and not finish it? (Such as putting something in the cart but not buying.) Is there something I can do so I can make 'almost conversions' turn into complete conversions?

or else - why are we reading this site ....

I also tend to look at conversions, were the user was before they made the conversion, where they went after and how they got to my site in the first place. Also tracking repeat usage, helps to identify what avenues provides the most conversions and helps to better identify the various profiles of a site's users.

Many metrics tools out there do *not* measure conversion...it is my main frustration with so many web stats programs.

...my best ranking site has a loyal group of followers (bookmarks), but never really benefits from its ranking. It's a bit of a niche - I know. 80% bookmark traffic I bet (imagine how low my traffic there is - heh)

Of course, I do what grnidone does - more or less: monitor search terms I accidentally rank for, optimise for them some more.

The thing is that that site is not there for the money. To get a real "conversion" (haven't had one yet) I've gotta score a "soul" {insert evil smiley here} Pretty hard I tell ya.

any suggestions on free, low cost solutions for this? i've just started to play with google adwords conversion tracking stuff.

Mikkel's right - you simply need client side measurement (javascript/web bugs) to get the right stats for user behaviour, and that includes conversion tracking. There is so much that never makes it to the logs.

For bot behaviour, otoh, the logs are essential, so of course it's best to have both.

Client side alone can't do the job (particularly when javascript is turned off, heh) - I build tracking into the site code itself. You also need to start tracking outside of your site if using advertising, it's important to know which ads perform better for attracting the best customers (RFV/LTV).

Customer service stats are an often neglected area in ecommerce sites - some sales look profitable until you see that cluster of customers are a pain in the rear end. One guy I worked with became 100% more profitable by cutting his low end line of products completely.