KitchenPC

So, I bank with Fidelity.com and use one of their credit cards for the majority of my purchases.  Their credit cards are issued by FIA Card Services (MBNA), which was of course acquired by Bank of America in 2006; my current theory is it’s actually impossible not to have a credit card owned by Bank of America.  FIA’s online payment system is a textbook example of usability gone wrong.

Gripe Number 1

Rather than a normal human-being style logon with a username and password box, they’ve decided to divide the logon process into two pages.  On the first page, you enter your username.  You then click “Sign-in” and get a brand new page where you enter your password.  Of course, the password textbox is not given focus by default and it’s nowhere near the top of the tab order, so you have to grab your mouse and click on it.  Paul Fitts is turning over in his ergonomically perfected coffin I’m sure.  After typing in your password, you can press enter to logon.  Nowhere do they have the ability to save your login info, retain your session, or do anything that might be helpful to minimize this ridiculously protracted logon ritual.  Pretty much the only thing I ever do on this site is pay my credit card balance in full.  Being without a job and living off savings, I have no reason to keep a positive credit card balance.  I’m allowed three online payments per month, so I have to juggle them around strategically as not to incur any unnecessary finance charges.  The online payment experience is also designed to make use of as many unnecessary steps as possible, which leads me to gripe number two.

Gripe Number 2

From the home page (after their non-sensical logon rigmarole), you have to go through the following steps to make an online payment:

1. Click which credit card you want to drill into (even though I only have one)
2. Click Pay Online
3. Go to a page that shows absolutely no new information, and click Pay This Account

You’re now at the page where you can pay your credit card balance, and presented with a Payment Date, which defaults to the date at the way end of your billing cycle.  It’s not as if I’m here to maybe make a payment today, or anything.  So now I have to click on this and change the date to today’s date.  The payment amount textbox is, of course, blank, so hopefully you have a great memory for numbers or you wrote down what your credit card balance was 3 pages back.  No where on this page does it say either the minimum payment due or the current balance.

I can, of course, see why they do these things.  FIA would rather I make my payments as late as possible, insuring more charges will have passed their grace period.  Not knowing what my balance is will provide a higher likelihood of me not paying my card in full.

So enough with this, I decided I just want to automatically pay my balance every 14 days.  Luckily, they have an “Auto-Pay” feature that should do exactly that, right?  Wrong.  This would be gripe three.

Gripe Number 3

The auto-pay feature allows you to pay your bill automatically either bi-weekly, monthly, bi-monthly, or semi-annually.  It has one box that says “Payment amount $:” where you can enter in the dollar amount to pay.  I decided to type in $5,000 here since my bi-weekly credit card charges would never exceed this amount.  The logical programmer side of me figured if my balance was under $5,000, they would simply pay the current balance on the card.  WRONG!

A few weeks later I decided to check on this.  When I logged on, I saw that my credit card bill was now negative $4,100!  They had swiped five grand from my checking account!  There were three more payments setup staggered every other week for another five grand a piece.

So let me get this straight, FIA.  To use your auto-pay feature, you have to magically know what your balance will be?  And if it’s under, they simply take the money anyway and leave you with a negative credit card balance?  It was then I realized this feature has absolutely nothing to do with customer convenience, it is simply designed to deter anyone from paying off a credit card, and hopefully leave people automatically making minimum payments on their exorbitant credit card balances for as long as possible.  I have no choice but to remember to logon every few weeks, jump through these hoops and pay off my card in full.  At this point, I’m thinking I should just go back to using a debit card.

The lesson here?  When designing a UI, don’t be evil.  You’ll just annoy bloggers who will write about your gratuitous usability sleaze.

Back in January, I was approached by a company called Nuiti Labs who is the maker of a popular Firefox plugin called Firesay.  Firesay enables users to control their web browser with their voice just by saying things such as “Open Tab” or “Go To Facebook.”  While these sorts of things have been around as far back as I can remember (yeesh, I think they had these programs for Windows 3.1!), they’ve never really seemed to catch on.  I suppose people are just too used to using a mouse or keyboard to interact with computers.  However, suppose there was a scenario where it really did make more sense to use voice controlled technology; that’s what Ehud Halberstam from Nuiti wanted to talk with me about over coffee in Kirkland one afternoon.

This Israeli-based company was hard at work on their next version of Firesay, which would support the Google Chrome web-browser.  This product “spin-off” would be branded as “Firesay In-Page” and was no longer a plugin just to control the browser, but was an entire API that allowed the website itself to harness this technology.  Web sites could now provide a whole new experience to their users by offering voice controlled functionality and text-to-speech feedback.  Ehud was quite interested in exploring the possibilities of kitchen related scenarios, and stumbled across KitchenPC during his investigation.

A partnership was definitely mutually beneficial.  Nuiti would have a great recipe site to showcase their product, providing a real use case and not just theoretical examples for marketing purposes.  KitchenPC would get to “ride their PR wave” and hopefully get a flood of traffic through their product launch.  Though “people who install browser plugins” is hardly my target demographic, I figured I’d at least get some of the techie chef crowd interested.

We spent the last couple months working on the required KitchenPC changes, and I’m happy to finally be able to announce today the launch of a Firesay enabled KitchenPC!

Firesay hands-free web browsing for KitchenPC

Kitchen Mode feature for KitchenPC

The Firesay enabled KitchenPC experience is quite simple.  Once you have the browser plugin installed, you can load any recipe on the site.  You’ll now notice a little Firesay command widget at the bottom of the screen whenever Firesay is listening for commands.  It’ll display a list of available commands for whatever context you’re in.  While in the recipe viewer, you can say “Firesay, Start Cooking.”  KitchenPC will then shift into a “full screen” theater mode that is optimized for kitchen use.  Only the information you need is displayed, and the fonts are huge so you can read the screen from ten feet away or so.

KitchenPC is able to break down recipes into steps automatically, and the first step is displayed on the screen and read out loud using a synthesized voice.  The user can say “Firesay, Next Step” to advance to the next step, or “Firesay, Repeat Step” if they miss something and don’t want to have to glance up at the screen.

I’ve done a lot of testing and the voice commands actually work surprisingly well!  I can be anywhere in the room and still have great results.  In fact, one time I stepped out of the room into the hallway and it still worked.

What I most like about this partnership is this is the first feature (hopefully of many to come) that really starts to target my initial vision of bringing the power of the personal computer into the kitchen.  Though not many users have PCs set up in the kitchen, hopefully this sort of innovation will start to change that.  For now, I can see this feature being used on laptops and tablets while cooks are busy chopping up onions or searing steak.  According to Ehud, there will soon be versions of Firesay that run on iPhones and iPads too.

You can learn more about hands-free operation on KitchenPC by clicking here.  I’d love to hear your opinions on this feature, or any ideas on how I can make it more useful as well!

Tuesday evening, I decided to check out an event called eDate.  eDate is somewhat like speed dating, but pairs up entrepreneurs with potential law firms, marketing gurus, investment specialists, technologists, and more.  You get ten minutes each with two experts of your choice, and also get to mingle with all the attendees in a non-formal setting as well.

I went with a friend of mine and his business partner who are also doing a startup and we all had a fantastic time.  The event was very well organized and had free snacks and drinks, including wine.  I decided to meet with a legal expert and a marketing firm.  I’ve had a lot of questions around setting up a formal corporate entity, how shares work, transferring the KitchenPC intellectual property, and potentially being set up so I can bring in outside funding.  Coincidently, I recognized one of the law firms (Ashbaugh Beal) in attendance and decided to meet with them.  Rick Beal personally helped me out quite a bit several years ago with a rather large insurance claim involving wind, two trees, my house, and gravity.  I figured I could at least go and name drop.

Regarding legal work, it’s something that eventually needs to be done to make any company “official.”  KitchenPC is an LLC in Washington State, but I’ve always filed “No Activity” and not really used it for anything.  What I’ve learned since is there’s quite a bit of controversy in the field around deferring payments for cash-starved startups as well as firms taking equity stakes in these companies.  Traditionally, it would be considered unethical and a conflict of interest for a legal firm to own shares in a startup they represent.  For example, if Ashbaugh Beal owned shares in KitchenPC, and one of their other clients is my biggest competitor, the advice they gave that client might be biased by the fact they have an interest in my success as well.  Other law firms (mostly the big ones in the bay area who are used to dealing with dot-coms) say this is a bunch of bologna and have no problem doing these sorts of arrangement.  Ashbaugh Beal is among the former group, however I’m hoping if I were to still work with them we could work out some sort of deferred payment plan.

I also met with a guy named Adam who works at Odd Dog Media.  I wanted to chat with him about cheap ways to get users on to my site.  So far, my user acquisition strategy has been to “get lucky” by getting blog mentions and other press, which causes random spikes every so often.  Other strategies I’ve found, such as ad campaigns, Facebook marketing, YouTube videos, etc all seem too expensive per user acquisition and perhaps not a very efficient use of money.  I really liked Adam and would love tot meet up with him again.  He had all sorts of great ideas for the site, including developing an embeddable widget that other bloggers can embed in their HTML to show recipes and meal plans for my site.  He says he and his team get together on Fridays for beer and to toss around ideas, and invited me to join in one of these weeks.  I’m hoping he’ll let me take him up on that offer.

I also decided to bring along several full-page glossy print-outs of my website (big thanks to my friend Ken, who will of course never read this, for having those done for me!)  This turned out to be a great idea and everyone loved seeing them.  Describing your website is one thing, but actually being able to show what it looks like really helps you connect people to the idea.  As they say, a picture is worth a thousand words.  I received nothing but positive comments on the design.  I’ll definitely be taking these with me to future meet-ups as well.

Since I got spots one and three (there were eight rounds), I was finished early.  However, due to some of the experts not being fully booked, I was able to grab several openings which provided me with two free dates!  I chatted with a finance guy who talked to me a bit about taxes and how to write-off some of my debt, and also with Perkins Coie, another well-known Seattle law firm.  Perkins is huge in the startup world, and has absolutely no problem with taking shares in a startup in exchange for legal services.  Typically, they charge about $2,500 for setting up a corp in Deleware, and then take between 0.75% and 1.5% of your company in return for around $30,000 in deferred legal services.  The payback is tied in to your first round of funding.  This deal sounded rather typical from what I’ve heard.  Big law firms are also known to help open doors for you.

Even though I stuffed my wallet with a huge stack of KitchenPC business cards before I left, I still managed to run out.  One thing that came as a surprise is several people who I talked to had heard of my site, so apparently I’m getting more well known at least in the entrepreneur community.  Hooray for very slow progress!

After spending quite a bit of effort working on SEO lately, I’ve noticed an unforeseen problem that this optimization has created.  With more and more users finding my site through search engine queries for recipes, the entry point into the site has shifted.  Normally, we think of the home page as being the first page a user sees when they reach your site.  Thus, the home page will capture the user’s attention within seconds, look the prettiest, and really sell the product.  It’s the home page that you do A/B testing on and allot most of your initial dev time to.

As a content site, a lot of users will find my site through one of the 10,000+ recipe links that I put out there on search engines.  A user may search for “chicken recipes” and come across this little gem.  This page, in this case, would be the first impression the user gets of KitchenPC.  The problem with this page was it just wasn’t designed to really sell KitchenPC as a product.  The user has no idea what my site is, except for a place where recipes live.

If you’re a long time reader, you know that these recipe “permalinks” were a feature I added a bit after launch to promote recipes on the site without requiring a user to have to create an account and logon.  It also gave Google something to crawl since Google can’t reach the secure pages.  The home page, back then, was nothing more than a “please sign up!” message and a quick blurb about what the site does.  I never had the money for a clever video or the time to do a site tutorial, so I figured I could attract users through recipe links.  This worked relatively well, however the static recipe page design was built to be completely static content without any of the real KitchenPC features.  It only displayed the raw recipe data, unlike the site’s “popup” recipe viewer which had all sorts of great functionality like “Create shopping list”, “Adjust serving size”, and “Add To Calendar.”  Someone looking at this static recipe page would have no idea that these features existed, and very few users would know they had to logon, go dig up the same recipe using search, and then they could work with the recipe using the meal planner tools.

Some of the new features on the static recipe pages

Thus, I’ve spent the last couple nights porting a large chunk of the “rich” KitchenPC recipe viewer features over to this page.  The result is near feature parity between the popup recipe viewer and the static recipe viewer.  As of 20 minutes ago, the following features are now available on the static recipe pages:

• Print
• Email Recipe
• Create Shopping List
• Add to Cookbook
• Add to Calendar
• View/Add Recipe Comments
• Adjust Serving Size

I’m still a bit short of complete parity.  For example, you can’t yet rate a recipe on this page (I need to refactor a bunch of the rating code before I can implement this correctly, but hopefully soon.)  Creating a shopping list will also overwrite your existing shopping list without warning, rather than prompting you if there’s existing items like the popup version does.  This is because I don’t load the user’s shopping list into memory on this page.  The page also doesn’t provide the ability to remove the recipe from the cookbook if it’s already there.  However, these minor issues will be fixed in due time.  The point of this work was to advertise to new visitors what the site can do.

Suffice to say, when new users stumble across the site through search engines, Facebook Likes, and Twitter posts, they’ll definitely notice KitchenPC provides a lot more meal planning tools that they wouldn’t otherwise know existed.  Hopefully this will result in a few more signups in the coming weeks!

KitchenPC attempts to construct the perfect shopping list by taking into account the various forms an ingredient may be used across multiple recipes and converting those amounts into the form said ingredient would usually be purchased in at a grocery store.  This works fairly well in theory, as your shopping list will say “all-purpose flour: 9oz” and as you browse the baking aisle at the store, all the various bags of flour will indicate their size in ounces.  If the shopping list expressed flour in cups, you’d really have no idea how much flour you need and just buy a large bag just in case.  This problem would be compounded if you were cooking large portions of recipes and needed like 465 cups or something.

The flour example “sells” nicely because almost every recipe uses flour in a volumetric form and almost every flour brand sells flour by weight.  Actually, pretty much everything that isn’t a liquid is sold in weight as the volume can shift and settle during transit.  However, it became clear that every ingredient in the database isn’t so cut and dry.  For example, tomatoes are also sold by weight and will be weighed by the cashier at checkout.  Your grocery store doesn’t really care if you buy a bunch of small tomatoes or a few large tomatoes, the total weight is what will be ringed up.  A KitchenPC user, however, might not like shopping for “13.5oz of tomatoes” though.  Imagine a scenario where a user adds a recipe that calls for 3 tomatoes, and another recipe that calls for 2 tomatoes.  What they would probably enjoy seeing on their shopping list is “tomatoes: 5”.  With the current implementation, what they’ll instead see is “tomatoes: 22oz” – Well great, now they have to either guess how many tomatoes that is, or use those hanging scales at the grocery store like some little old lady.  Ok, to be honest I’ve never actually witnessed anyone using those scales but I imagine they’re only used by little old ladies who are also the same ones holding up the line with eight million coupons to give to the cashier.

That’s no good.

The solution?  Well, this sounds like another opportunity to use my new and improved (actually just improved) form conversion engine that went online a few days ago.  This code gives me the opportunity to convert pretty much anything in the database to anything else I want.  Two modifications were made to the shopping list.

Tooltip for an ingredient in the shopping list widget

First, when you hover over an ingredient in your shopping list, KitchenPC attempts to express it in other helpful ways.  For example, if the ingredient is expressed in weight and there’s a default volumetric form available, it will show that conversion in the tooltip.  If the ingredient is expressed in whole units (such as bananas,) KitchenPC will show the actual weight of bananas you require.  This would hopefully prevent someone from not buying enough bananas for recipes that call for “mashed bananas” because the bananas they bought were all too small.

Second, the printed version of the shopping list has a new column for these estimates as well.  Hopefully, this will make reading the printed shopping list at the store much easier.

I actually would like to go a step further with this design to make it even more powerful.  One feature I have in mind is a rich “hover-over” panel that pops up when you move the mouse over an ingredient anywhere on the site, similar to hovering over someone’s name on Facebook.  This panel would show convenient equivalent amount information, as well as a picture of the ingredient (pulled in from Google’s Image Search API) and quick links to find other recipes that use that ingredient.

Printed shopping list

Another improvement this new code allows me to do is start to cater the unit types in the shopping list to really target what the user wants rather than cater to programmatic limitations.  The real reason that ingredients such as “red tomatoes” are listed in weight by default is the system was designed to eventually integrate in with online vendors.  The shopping list schema needed weight information because that’s what a third party vendor would require within the order information.  This technical requirement is now less than necessary as weight conversions can be done on the fly.  In theory, I can connect to a third party vendor and “order” tomatoes in either weight, or whole tomatoes; whatever the vendor accepts.  For this reason, I plan on actually changing many of the KitchenPC fruits and vegetables over to whole units when appropriate.

These new features are now live on the website for those who want to check them out, but keep in mind a lot of the database metadata isn’t really “optimized” yet to really take advantage of this ability, so some ingredients will be a bit off or not display estimated amounts.  Hopefully, one of these days I’ll get the chance to go fix up the database; or at least the more common ingredients.

Enjoy!