If you aren't vigilant about sending them reminders, coupons, and special offers, and asking them to refer you to friends, and offering frequent shopper programs and so on, then of course they aren't going to necessarily come back, and every sale you make is going to cost you the same high price you needed to get the first sale from them.

The lifetime value of the customer assumes you will make a reasonable effort to continue to get the custeomr to buy from you- which is proven to be much more likely after they have bought once, however you still have to work at it.

The advantage is, once they've bought once, you have their name and can contact them directly, which should cost you close to nothing compared to attracting someone who has never bought before.

I target different facets of my company site on different pages. Each page dumps the targeted user info into seperate databases. I have been recruiting links for each different page seperately to boost page rank.

I will swear to the isolated increase in page rank on the targeted pages. My most linked-to page is PR7 while, my home page and other pages range between 4-6.

The best part of all this is the effect on your targeted keywords. I am #1 in Google for over a dozen keywords and phrases that link directly to pages within my site. I would never approach that trying to fit all my KWs onto one heavily linked page.

See your site as several sites combined under one main home page and, recruit targeted links for each individually.

Bar graphs
Web Browsers (GRML)
Web Browsers (GRML)
Web Browsers (GRML)

Bar graphs
Web Browsers (GRML)
Web Browsers (GRML)
Web Browsers (GRML)

Bar graphs
Web Browsers (GRML)
Web Browsers (GRML)
Web Browsers (GRML)

By all means use a program to figure out potential link sources, but don't send out anything without first personalising it so it doesn't look like a template based email.

Let me tell you a story...

I see a number of untargetted & unpersonalised link requests in my job, they get deleted just as fast as "generic" pharmacy spams, if its a particularly bad day they may get thrown through spamcop in an attempt to reduce the amount of daily spam. At home I see relatively few, although the obvious cookie-cutter spam ones always get spamcop'd.

Two weeks ago I saw one which was being flagged as spam (DNS blacklisted) and was set for auto-delete, however since they'd personalised it I sat down and read it properly because it piqued my curiousity, finally I decided that I'd keep it & replied. You want to know why I did this? Because it appeared that someone had sat down, had a look at my site and taken the time to write me an email which could convince me why linking out to them was worthwhile based on the content I'd created.

Personalisation is about more than just Dear Joe Public at the top of the email, it's about spending as much time creating the request as you expect me to spend reading & replying to it - personalise the content and there's a reasonable chance that people like me will make it to the end of the email without either deleting it or reporting it as spam.

The best compromise I've found between the convenience of robots.txt and the specificity of the meta tag approach is to "Allow" the top-level pages that link to obscure content in robots.txt, and place robots meta "noindex,nofollow" (or just "nofollow") tags on those. Then the lower-level pages below them can be protected by robots.txt just in case a rogue link is found.

Technically, Google is not violating robots.txt, since it never spiders Disallowed pages. But since it never spiders them, it won't see any robots meta tags which may be present, either.

Took a few months of testing to figure out how to keep pages obscure, but it does work

They call it "artificial" (as opposed to "natural") because they detect a pattern of interlinking sites (a "bad neighborhood"), which are all MUCH more popular from each other than they are from the rest of the universe. Google has seen trillions of links, and they say "this ain't natural!"

It is therefore almost certainly an "artifice" of deceptive SERP perpetration, hence "artificial," created by a single entity masquerading as all the members of a closed mutual admiration society....and almost certainly deserves the contempt it receives from the rest of the universe.

It is NOT the same thing as a "Free-for-all" link farm, which collects spam from a lots of different entities, and it is NOT quite the same thing (although algorithmically, it will look very similar) as any of the Link-Exchange-for-Luzers programs.

And, of course, it's NOT the same thing as "hidden links", which is a different kind of deception altogether, and which Google tries to detect by altogether different methods.

It is very simple. If you have a website that nobody will EVER care about visiting, because they could have seen the exact same products and prices at any number of other places if you had only happened to die and rot before they did their search, then the only way you are going to get incoming links is to make them yourself, thus committing "artificial linking."

And deserving the oblivion that Google tries (not always successfully) to give.

It's not a penalty at all. It's a correct evaluation of the true link popularity (based on all independent links) of the site. It's just that oblivion is hell on promoters.

In-site cross-links
Cross-links, in this context, are links WITHIN the same site.
Link to on-topic, quality content across your site. If a page is about food, make sure it links to the apples and veggies web page. Specifically, with Google, on-topic cross-linking is very important. It helps to increase the linking power of other web pages in your website. 

Do NOT have an "all-star" page that out-performs the rest of your site. You want 50 pages producing 1 referral per day. You do NOT want 1 page producing 50 referrals a day. If you do find a page drastically out-producing the rest of the site, you need to off-load some of that to other pages. Do this with heavy cross-linking. It's the old, share the wealth thing.

Put it Online
Don't use virtual hosting. Use a stand alone ip.
Be sure the site is "crawlable" by a spider. All pages should link to more than one other page on your site. And, the link should be no more than 2 levels deep from root. Link the topic vertically, as much as possible, back to root. A menu present on every page should link to your sites main "topic index" pages (doorway pages and logical navigation system should lead to real content).

Submit
Submit the home page to directories. Now, this is the hard part. Forget about submissions for the next six months. That's right. Submit and forget.

Logging and Tracking
Get a quality webiste logger/tracker that does justice to inbound referrals, using log files (don't use a lame graphic counter - you need the real deal). If your host doesn't support referrers, back up and get a new host. You can't run a modern site without full referrals, available 24x7x365 in real time.

Perhaps, the best approach depends upon the overall topic. Or, another approach is using the amount of content in the site. Or, use both. It seems some topics lend themselves to more of a pyramid structure, where 2nd levels are possibly closer in meaning to each other. If so, cross-linking more is beneficial. On the other hand, if the 2nd levels were further apart in meaning, this would be more of a solar system approach.

For example, dog and cat are 2 levels of an animal website. This is a solar-system approach. They are different from a scientific standpoint. More importantly, the searches are not related.  On a household pet website, dog and cat are more closely related. Hence, this is a pyramid approach.

Doing data entry using Pioneer Report MDI (GRML web browsers) supports standard features for editing. Data entry is allowed for each column of a result. There is data entry support for cut, copy, and paste. Data entry supports inserting and deleting of individual or multiple results.

One data entry feature lacking is bulk copy and pasting of selected results. In data entry, it is easier to copy a set of results from one view and pasting it in another than inserting each result item, separately.

Creating dynamic forms is a design goal for all GRML file and web browsers. Currently, all file and web browsers create input controls in a sequential manner. The first input control read from a GRML file or web page is the first input control created in the browsers form area. Within each form, the edit controls are created as a group. Then, the combo (drop-down) boxes are created. Finally, the submit button is created.

Headlines MDI (GRML web browsers)
Tree MDI (GRML web browsers)
Pioneer Report MDI (GRML web browsers)

Other GRML blogs of interest are...

GRML (web browsers) Blogspot
GRML (web browsers) Blog City
GRML (web browsers) Blog Drive
GRML (web browsers) Blog Lines

These blogs cover releases of GRML file and web browsers. All releases support CSV and delimited formats. GRML has capabilities and features not used by CSV and delimited formats. When reading a GRML file or web page using these features, they are ignored when saving.