Archive for December, 2006

Search or “pay to play”

December 17, 2006

To drive customers to our site and create brand and product awareness, we wanted to be included in the major search engines and to advertise both online and offline.

There is plenty of SEO (search engine optimization) research explaining how Google, Yahoo and MSN crawl the web to find and index sites. We have read a lot of it, and we advise “buyer beware”.  Some SEO companies profess to know the Google search algorithm and for either an upfront fee or a monthly charge will “optimize” your site for search listings. A few of these SEO companies promise (or even “guarantee”) a top 5 search result for your site (meaning your site will show up in the first 5 listings when someone searches for a keyword like “coffee” or “fondue”). Website owners complain because they are not listed for several months despite tremendous efforts to optimize their site for search listings. In response, SEO experts claim that Google “sandboxes” (excludes from search results) new websites for 6-9 months.  Many sites set up so-called link farms that contain nothing but links from other sites to their site because search engines will improve the page rank (which position in search results a site will appear on) if the site has a lot of incoming (and outgoing) links. Exchanging links with other sites is not an option for us because we do not want our site to contain dozens of links and banner ads, nor do we want to engage in the dubious practice of setting up other sites as link farms that would point to our main site.

In order for us to achieve our short-term goal of a search listing on the first three pages with the major search engines, we had to focus on website best practices, including: good site layout, easy to navigate (no broken navigation paths/links internal to the site), relevant and constantly updated content, well-structured home page, simple file structure with appropriate naming convention, sufficient (but not overstuffing) keywords in titles and text, some incoming links, clean and lean code, tags on images, a file (robots.txt) specifying for the search bots what they should and should not crawl and index.

In the short term we decided to participate in “sponsored” search until we would appear in the organic search results.  Sponsored search is the practice of bidding on and paying for relevant keywords and waiting until customers click on our ad that appears adjacent to search results for that keyword. In November, our first month of operation, we spent about $500 on sponsored search at around $0.50 per click on average per keyword (or about 1000 clicks). Given the click-through rate was somewhere around 2%, that represents 50,000 searches for our chosen keywords in one month while our sponsored search was “turned on” (there were days/times we turned it off because we did not believe we would attract the target demographic or visits would be low quality).  Here’s what we got: for the Google Adwords program (sponsored search), average pageviews per visit was 1.75, compared to 3.8 p/v for Google organic search (more than twice as high), and 4.15 p/v for direct visits (users who accessed our URL directly).  The Google Adwords referred visits are not what one would consider quality visits.  And the Google Adwords program generated few sales for us.

During November and December, we improved content, added features and optimized the website to make it more “search engine friendly”.  Several search bots index our site daily (including Google, Inktomi/Yahoo, and MSN) and we monitor our progress closely.  As of the date of this post, Google does not rank us in the top 1000 search results for several of our keywords, including espresso cups, cappuccino cups and fondue pots.  MSN ranks us in the top 5 for cappuccino cups, top 10 for espresso cups, and top 12 for fondue pots. Even when adjusting for the fact that MSN indexes fewer sites than Google (345,000 search results for cappuccino cups vs. Google’s 1.3 million), it still does not explain why MSN ranks us and Google does not.

So should we be patient and wait until we move up in the Google search results, or do we continue to spend on the Google Adwords program knowing that it doesn’t bring quality visits, product sales or even long-term customer relationships?



December 15, 2006

Building a business is not possible without the help and advice from people we like to call subject matter experts. One of the many rewards of building and running a business is to be able to meet people who are subject matter experts and freely offer their vast wealth of experience and expertise to a new business like ours. We have been fortunate, and are very appreciative, of the many people who are helping us build something very exciting! In a series of posts we would like to acknowledge a few of our friends…

We have been fans of (CG) long before we launched In fact, we frequently visited their online community and read many of their articles, product reviews and forums to learn about espresso and coffee, growers, roasters, baristas, espresso and coffee enthusiasts, equipment manufacturers and distributors, and so on. Although there are several other good online coffee communities, we believe CG is by far the largest, most active and highest quality. Mark Prince, its senior editor and uber-coffee-geek, brings to mind Robert Parker and Hugh Johnson (apologies to the wine aficionados for borrowing Messrs. Parker and Johnson) in terms of his knowledge, credibility, passion, vision and influence.  He and his team have created a coffee community numbering more than 500,000!

When we were ready to go live with our business and website, we contacted CG to explore how we could join their community. CG encouraged us to market our espresso and coffee-related items on their site and within a few weeks invited us to participate in their holiday fund raiser and silent auction to benefit the Coffee Kids: Grounds for Hope charity. We mention this because it demonstrates the values and power of an online community (CG) and the recognition by participants in the coffee industry that they must give back to the larger coffee community (through charities like Coffee Kids) to nurture, sustain and grow that coffee community. CG managed to assemble an impressive list of 30 sponsors for the silent auction, which kicked off Thanksgiving week and will end mid-January. contributed a set of 2 orange espresso cups with saucers and 2 light blue espresso cups with saucers (each set costs $56 plus shipping). The espresso cups are handcrafted in Switzerland, available in 8 funky colors, accented in platinum and signed by the artisan. Bidding for the orange set closed at $75 and the light blue set is being auctioned by 7 pm on Dec 16. Although our contribution is modest, reflecting the fact that we are the new kid on the block, we are nevertheless proud to be able to participate with CG and the entire community. And we thank Mark Prince and CG for inviting us!

Since joining the CG community, we have met several interesting people who have offered their advice and help to us. We’ll introduce them in an upcoming post.

Outsourcing webhosting

December 15, 2006

We built our own development and testing environment by installing and running XAMPP (which includes an Apache webserver, PHP 5 server scripting language, MySQL database) on a Windows XP PC.  Total cost = $0.

As mentioned in a previous post, we decided to out-source hosting of our website. Webhosting has become a commodity service with feature sets being largely the same across the dozens of service providers. What really differentiates webhosting companies is price, performance and service levels.  It was important that our webhosting company own the data centers and not act as a reseller for someone else, invest in state-of-the-art hosting infrastructure, be reasonably priced ($10 or less per month), and have 24/7 live, competent technical support.

We selected Bluehost located in Orem, Utah as our webhosting provider.  Bluehost’s main challenges have been managing incredible growth (including that of Hostmonster and Fastdomain, two affliates) and not investing in new capacity and more support engineers to support the fast pace of that growth.  While we were developing and testing the site this summer and early fall, Bluehost’s performance deteriorated and we almost moved the website to another webhost.  Just before launching the website, we upgraded from a shared IP to a dedicated IP (an additional $30 per year). That change, combined with Bluehost resolving their performance issues, has improved performance to an acceptable level.

Now we wish that Bluehost would become more customer centric (less feature-driven) and improve its communication with customers.  One example would be to consult with customers about proposed changes before implementing these. Now we recognize that not all customers will agree with proposed changes, but if one applies the 80/20 rule, that should be good enough. Consulting with customers could easily be done through a blog (their CEO has one but it is not interactive) or their user forum. It would be a smarter approach to business to assess customer impact and reaction to major changes first rather than react once customers express their views and possibly move their business somewhere else.

Catering to different user behavior

December 15, 2006

One of the key design objectives for the website was to provide several different ways for users to browse/search the site. Although at present our site is not large, it is important to recognize that different users will navigate a website in different ways and therefore the site should cater to different user behavior. Equally important is to make any page on the site easily and quickly accessible. We did that in several ways:

  • Horizontal navigation bar with two levels
  • Contextual links embedded in text
  • List related items in right side column – items are clickable
  • Integrate a site search engine above the navigation bar
  • Minimize number of clicks to reach any page (max. 3)
  • Optimize page load speeds
  • We are using several free “web statistics” tools to track and learn about user behavior to continuously improve the user experience (our site does not collect any personal information about users unless explicitly stated, e.g. to enable the shopping cart). One of these tools, Google Analytics, allows website owners to track and analyze how users enter, navigate and exit the site, what content they view, how long they spend on the site, whether they return, and if they transact on the site. Note: Google Analytics is absolutely free, and can be used with or without Google Adwords or any other marketing/advertising campaign. It does require you to copy and paste a code snippet on every page you wish to track. Goggle Analytics is a fantastic analytical tool, but our assessment is that it is a works-in-progress (Google doesn’t label it “beta” but probably should) and some of the stats deviate more than expected from Awstats and Webalizer.

    We also would like to explain our thoughts about site search. You may question site search for a small site, but we thought that if someone came to our site looking for red cappuccino cups or yellow dinner plates, they may be more inclined to type the search phrase into a search engine rather than using the navigation bar. Or they may be looking for a cheese fondue recipe or how to make a caffe latte, in which case they may again prefer to search. Our decision to incorporate site search was also based on implementation cost (license fee plus time to integrate and configure). We found Wrensoft’s Zoom search engine to be the answer. A free version (with no advertising) is available for small websites up to 50 pages. For larger websites, you can purchase either the standard version ($49) professional version ($99) or the enterprise version ($299). Zoom is easy to install, configure and maintain (ours was up and running in about 30 minutes) and runs on javascript, PHP, ASP or CGI. It is capable of full text searching of static and dynamic files (including PDF and Flash). And it supports foreign language search. You can modify the CSS and seamlessly integrate Zoom into your site (e.g., add to the navigation bar). The search algorithm is open and configurable for the experienced programmer. The search results page is configurable, and we included a quick shopping cart at the top of the search results page. Wrensoft has just released version 5.0 of Zoom. All in all a great package!

    From prototype to online store

    December 14, 2006

    We developed a prototype website (without shopping cart and database integration) using Apache/PHP/MySQL running on Windows XP on a Dell PC. You can download XAMPP from Apache Friends. There are several versions that run on Linux, Windows, Solaris and Mac OS. Installing the package was simple and quick, and compared to downloading and configuring each of the applications/database separately, was a huge timesaver! We accomplished a number of objectives with the prototype, including approaching banks for financing (the prototype together with the business plan was part of our “roadshow”). Having a working prototype definitely made a difference in our discussions with banks.  We successfully raised funds from a local bank to finance the import of Terra Keramik’s handcrafted tableware from Switzerland.

    There are dozens of shopping cart options available and it is important that one does his/her homework. We selected WebAssist’s eCart for the following reasons:

  • supports multiple server-side languages (including PHP 4 and 5)
  • integrates with several databases (including MySQL)
  • compatible with all major payment gateways (both local and remote checkout)
  • available as a Dreamweaver extension
  • good tutorials for both static and dynamic implementations
  • includes several prepaid support incidents
  • good user support forum
  • several complementary WA packages available
  • We actually purchased WebAssist’s eCommerce package, which includes eCart, database search, client/server-side validation, email from the website, cookies toolkit, etc. and we found is a great value at $399 (includes 5 pre-paid support incidents). The support incidents were useful because we wanted to customize the code and the tutorials were too generic for our purpose. Two areas where we would like WebAssist to improve is their integration with AuthorizeNet (probably the largest payment gateway for processing online/offline credit card payments). We struggled a little with Web Assist’s integration with AuthorizeNet’s AIM method (local checkout), and we recommend that WebAssist improve its tutorial and Authorize Net provide better, up-to-date API documentation. To overcome the integration issues, we used Sound Commerce’s AuthNet PHP script ($49) to integrate eCart with AuthorizeNet. We also had trouble with WebAssist’s email application auto-generating a customer confirmation email on successful customer payment/checkout. Using one of the support incidents, we worked with a WebAssist engineer to resolve the email issue to our satisfaction.

    Do it yourself

    December 12, 2006

    For anyone out there wondering – can you learn how to design and develop a good eCommerce website – the answer is a definite YES!

    We suggest you start with a web development tool (also referred to as an HTML editor) such as Macromedia Dreamweaver that has both a design view (WYSISWG) and a code view. Although there are several popular web development tools available, Dreamweaver is considered by many the industry standard and therefore integrates well with other software applications and extensions. To get a jump-start, we attended a 3-day web development course (about $1400) at Motion Over Time in New York City to learn how to build a simple website.

    Initially, we designed the layout and content of our site using the design view of Dreamweaver, but quickly “graduated” to the code view and hand-coded most of the site (including re-writing much of what we had done in design view to separate the structure, content and presentation of the site). The latter step may not be necessary for smaller sites but it does improve the maintenance and performance of a larger website. In the process we embraced cascading style sheets (CSS) to separate both style and structure from the content and behavior of the site. There are a vast number of books and articles on CSS, and we recommend both A List Apart and books written by Eric Meyer.

    We chose another industry standard, Adobe Photoshop, to work with static images. This is a powerful tool that may be more than what you need, but we found that some of the more simpler alternatives have limiting functionality. And like Dreamweaver, Photoshop integrates well with other applications (especially Adobe products) and there are numerous good tutorials and “how to’s” available. One of these we recommend is the book Photoshop CS for the Web Hands-On Training from We opted to buy this book and learn ourselves rather than taking a “classroom” type of course.

    To add interactive behavior to our website, we learned javascript (a client-side scripting language that runs in your browser) and PHP (a server-side scripting language that runs on the server). PHP is one of the more popular languages because it is a simple, yet powerful object-oriented language that is open-source and integrates well with MySQL (fast, open-source relational database). Javascript is also a simple language, but should be used judiciously because many users will turn off javascript in their browsers and therefore DHTML (dynamic HTML) and CSS may be more than adequate alternatives.

    We solved an interesting challenge using javascript: how to represent that our products are available in 8 funky colors without displaying 8 large images for each product that would use extensive screen real estate. Our solution: show thumbnails (50×50 pixels) in different colors and enable user to roll over these thumbnails to see single, larger image (200×200 pixels) change color as the mouse moves across the thumbnails (done in javascript). Each thumbnail is also clickable to pop-up super-size images (500×500 pixels) in color. Check out the effect on our espresso cups page.

    The intersection of design and technology

    December 10, 2006

    Once we decided to create an online marketplace to distribute Terra Keramik’s handcrafted tableware, we asked ourselves several key questions:

    1. How do you sell a premium product that is handcrafted by an artisan, three-dimensional and finished in stunning, colorful glazes through the web, which is a two-dimensional medium that does not present well the physical properties of a product such as finish, color, weight, dimensions, etc.?
    2. What brand and product qualities do we want to communicate through our web design, content and features?
    3. What type of web presence do we want? Product-driven (i.e. catalog layout), content-driven, feature-rich, or a combination
    4. What about the technology platform? Open-source or proprietary (i.e. Microsoft)? Are performance, security, and scalability important? How about in-sourcing vs. out-sourcing of development and hosting?

    So here are our key design choices:

  • aesthetically attractive site to present handcrafted product
  • provide product specifications for experts and connoisseurs
  • simple, clean look-and-feel consistent with the product
  • intuitive navigation
  • relevant content that would support the product marketing
  • interactive user experience
  • transparent information about business terms and conditions
  • And here are our key technology choices:

  • open-source technology platform (Apache, Linux, PHP, MySQL) supported by large user communities
  • largely standards compliant (HTML, CSS) [we say largely because we strive to be but we had to make some compromises in the short term]
  • separate site structure, content and presentation
  • optimize site performance
  • use packages or widgets where it makes sense (WebAssist’s eCart, proxy2’s AdvancedPoll, Sound Commerce’s AuthNet integration)
  • local checkout payment gateway (we own the customer and his/her experience)
  • secure the site with a branded SSL certificate
  • Let’s come back to the question about out-sourcing development and web hosting.

    Out-sourcing hosting made sense given the commoditization of this service and therefore attractive feature/price relationship.

    But the decision to out-source development was not so simple. Several friends suggested we out-source development to India. We were concerned about finding someone who could translate our brand and product vision into a website that combined good design with good technology at a reasonable cost. Our experience is that organizations tend to have an inherent bias towards either design or technology, and it is rare to find someone who is able to balance the two. And manages to do so on a reasonable budget and schedule.

    So we decided to develop the website ourselves.

    But will it fly

    December 7, 2006

    Before investing a lot of time writing a business strategy and plan, we wanted to “market test” the product with subject matter experts (people in the interior design, home accessories, tabletop, and coffee accessories businesses) as well as with potential retailers. To do that we bought a small shipment of espresso cups, cappuccino cups and coffee cups from Terra Keramik of Switzerland.  It arrived in October 2005 (only a few broken pieces) and we promptly hit the road to solicit feedback on the product and our business plan.  We received a lot of positive feedback and encouragement!  People provided useful input about the design of the cups, the need for attractive packaging for a high-end product, the competitive nature of the tabletop business, and so on.

    In our conversations we also came to realize that retailers would expect a 30-40% margin to “resell” the handcrafted tableware, which made the business unattractive to us.  And we would have to work out issues concerning logistics and inventory, shelf space, warranties, and so forth.  Ugh!  So we started to think about a pure online marketplace and our business strategy evolved in that direction.

    Benefits of online strategy:

  • modest start-up costs
  • faster time-to-market
  • owning the customer and customer experience
  • control of brand strategy
  • combine products with relevant content
  • better economics
  • Challenges of online strategy:

  • customers would prefer to touch and hold the product
  • no customer base of an existing brick-and-mortar retailer
  • experiment with emerging online marketing methods
  • relatively high shipping cost of product
  • One of us had previous online experience, and we felt comfortable that it was a strategy worth pursuing, albeit not without risk.  Let’s see if it flies!