Kalzumeus Software

So on my lunch break I was browsing the Internet and chanced upon Matt Cutt’s blog, and read the legal papers out of mild curiosity.  I noticed a throw-away comment buried in there, got a little miffed, and decided to do some minor venting on the Internet.  And I finished my workday, went out to bid fairwell to a friend leaving the country, and when I get back to my computer I’m on the front page of WordPress.  Google certainly inspires some passion in people, what can I say.  I understand that (and, if you’ve been following my posts about AdWords, you know I’m in general a fan), but there’s limits to my love of everything.  Except chocolate ice cream and cute kittens.

Anyhow, back to comfortable anonymity and the pressing business of saving teachers time and money.  Speaking of AdWords, mine seems borked today.  I’m showing a conversion rate of 0%, which is highly statistically improbable (somewhere on the order of 3% if you assume my past performance is indicative of expected results — statistics & probability was my favorite course in high school).  Even more strange, when I go to Analytics it shows 0 traffic for the last 3 days, which is crazily borked — yesterday I had normal stats up to the usual reporting delay.  Ahh well, kinks in the system, I’m reasonably sure they’ll get cleared up soon.  On the plus side, before Analytics started swallowing its tail I noticed that at least one person was able to successfully download the demo and check for updates.

The background: Google was sued recently regarding their efforts to prevent click-fraud in AdWords. It was a class-action suit, which basically means that there are a large number of people who were “harmed” by the tortious action at issue and that some lawyer has taken it upon themselves to sue on behalf of all of the ones who don’t opt out. Class action suits are a huge scam but that is another matter altogether.

Google attempted to settle the suit. In the process, the would have to contact class members (the people who have theoretically lost money due to fradulent clicks), and they hired a firm which specializes in this sort of work. So far so good. And that firm zealously tried to contact class members in a variety of ways, including through snail mail and email. So far so good.

Now, we all know the problems with getting mail to large numbers of people. Mail addresses changed, people go on vacation, challenge-response systems are engaged, what have you. The firm zealously tried to correct for all of these, by investigating new email addresses, tracking people down after vacation, clicking through the “I am a human” tests, etc. So far so good.
Now, what is the other main way for a mail delivery to fail? Spam filters. Now, remember, as a class member you haven’t opted-in to the lawsuit or the settlement. You might not even think you’ve been harmed by the action at issue, or you have no desire to waste your time for what is typically a sliver of a credit (the attourneys, of course, get 25%-33% of millions — in this case attourney fees will probably go above $20 million). So you might understandably not want to really talk to someone wanting to talk to you about the lawsuit. In this case, service from an agent of Google’s to tell you about your rights regarding the lawsuit is spam. You didn’t ask for it, you don’t want it, and it has a commercial purpose (they’re being paid to get the email to you, and the email is sent to divide up a pot of money — although unlike most spam its not your money).

So, as can be expected, lots of these advertisers have Gmail accounts. And what did Google do? It checked them. Google algorithmically peaked at all the accounts on the list their agent had developed which they had access to, to see if the mail was marked spam or not. There were 75,000 accounts in which it was marked spam, and an unknown (larger) amount of accounts must have been compromised to get that statistic.

Unhinged rantings of a conspiracy nut? Well, no. Google’s lawyers bragged about this in a recent document they filed to the court regarding the settlement (which is tied up in legal wrangling). In relevant part (page 13 of the pdf of the document which Matt Cutts provided on his blog while responding to concerns about click fraud):

Gilardi [ed: the firm Google was using to contact people] also re-sent 74,591 email notices to intended recipients whose addresses ended in “gmail.com” and “googlemail.com”, and for whom Google had information that the first email notice had been directed to the recipient’s spam folder. (italics mine)

Google is apparently hunky-dory with this. Its essential for the Google lawyers to demonstate that their notices stand up to certain legal requirements regarding legitimately trying to notify class members (note that its completely non-essential to go peeking). Google brags on page twelve:

[T]here is no question that Google complied with the notice procedures ordered by this court. In fact, Google did more than was required to provide the best notice practicable. (italics mine)

I’m sorry Google, I just don’t remember telling you you could go peeking at the mail, even to “provide the best notice practicable”. As a matter of fact, given that I know you’ll be storing it for life I actually bothered to read that privacy policy of yours. Lets see, where was it… aha.

Information sharing

Google only shares personal information with other companies or individuals outside of Google in the following limited circumstances:

• We have your consent. We require opt-in consent for the sharing of any sensitive personal information.
• We provide such information to our subsidiaries, affiliated companies or other trusted businesses or persons for the purpose of processing personal information on our behalf. We require that these parties agree to process such information based on our instructions and in compliance with this Policy and any other appropriate confidentiality and security measures.
• We have a good faith belief that access, use, preservation or disclosure of such information is reasonably necessary to (a) satisfy any applicable law, regulation, legal process or enforceable governmental request, (b) enforce applicable Terms of Service, including investigation of potential violations thereof, (c) detect, prevent, or otherwise address fraud, security or technical issues, or (d) protect against imminent harm to the rights, property or safety of Google, its users or the public as required or permitted by law.

Hmm, thats what I remember: opt-in consent for all disclosures of private data. I think the contents of my inbox is pretty darn private. So that ones out. You’ve already explained in your own words that the peeping was more than the court required, so excuse #3 is out. So what about #2: were you “processing information on [Google’s] behalf”? If you were, then this exemption swallows the entirety of the policy policy!

I’m less than happy, and now seriously wondering if all those business documents I’ve got floating around my Gmail inbox are going to end up in the hands of your lawyers without so much as a by-your-leave if your lawyers, in their sole discretion, think its for my own good strategically a good idea to get Google out of a lawsuit.

Do no evil, indeed.

[Edit: Fixed spelling mistakes and bolded some juicy bits.]

I just learned that JOS regular Phil makes software for chimney sweeps.  If the prospect of this doesn’t make you smile, you have neither soul nor taste in Disney movies.

Chim-chiminey chim-chiminey chim chim charoo, I does what I likes and I likes what I do…  

Chim-chiminey chim-chiminey chim chim, cheree, a sweeper’s as lucky as lucky can be. 

I made some major changes to Bingo Card Creator last, oh, Thursday I think it was, and then built and uploaded.  Unfortunately, this is largely a manual process and I missed the critical “copy the new installer from the build directory to the web directory before sync’ing the local copy of my website and the FTP server”.  Which means everyone has been getting the old version.  D.  O.  H.  Well, no sense crying over spilled milk.

I just had this brainstorm and wanted to share.  It should be live in my program by, oh, the end of today:

Checking for updates is a wonderful thing to enable, both because it makes sure your customers have the best-and-brightest version of your product available, and because it lets you see on your web logs “Ahah, that install actually succeeded”.  I wanted to have it ready for my first release (there have been at least 10 mini-releases since then, all with the same version number) but couldn’t figure out a way to do it simply.

Enter the drop-dead simple version checking solution: make a directory on your website (I’d put it in robots.txt too with an exclusion, as its going to have duplicate content and you don’t want it to be a search result anyway).  Populate it with a bunch of HTML files corresponding to your version numbers (e.g. v1.0.htm, v1.01a.htm, beta_release.htm, whatever you want).  All the ones but the “latest” one say “You need to update your software.  Here’s how:”, with your favorite tracking code embedded.  The latest one says “Yep, your software is up to date”, has the tracking code, and maybe gives a bit of advice or something (hey, why miss an opportunity to sell to people).  Every time you add a new version, you change the version number in the executable/resource file/wherever, rename your up-to-date page to the new number, and put an out-of-date page in place of the old page.

BAM.  This takes only 10 seconds per update, requires no additional programming, and can be done in 100% static HTML (no need to query a CGI script or anything).  Then you put a Check for Updates item in your Help menu, and perhaps pop up a window on program execution the first time its executed (“Thank you for using X.  Would you like to check for updates now?”) and perhaps every 2 weeks or a month after the program is installed.  Its not quite as seemless to the user as, say, Firefox’s automatic updating, but it’s a heck of a lot simpler for you.

I decided to switch my payment processing from Payloadz to www.e-junkie.com .  I’ll be keeping Payloadz around for their web store link but driving all the traffic from my site to e-junkie.  There are three major reasons:

Automatic Redirection: Its absolutely crucial to me that I be able to redirect customers back to my site after the transaction is completed.  Currently, I just want to capture the fact that they’ve made a purchase with Google Analytics.  However, I’m thinking of eventually displaying their registration code in-line in the page and dropping them directly at a quick-start guide.  e-junkie made this trivially simple to accomplish.  Payloadz… not so much.  I had to hack up a form myself for each link I wanted this behavior on, which was not an option considering that form would have to be in my site navigation on every page.

Rational Pricing: e-junkie is $5 a month, regardless of how much I sell.  Payloadz is… strange.  For sales under $100 in any consecutive 30 days I pay nothing.  If I get above $100, I have to pay $15, and that continues to $500.  At which point I have to pay $29.  And so on and so forth.  Frankly, I don’t know where my monthly sales are going to stabilize and don’t want them stabilizing at a point where I routinely pay 12-15% of sales to somebody doing, well, essentially not a heck of a lot for me.

Customer experience: Payloadz let me add my own text to emails to my customer, but I couldn’t change their default text, which was not applicable to my product and does not match my tone elsewhere.  e-junkie let me write my own email template, soup to nuts.  I also don’t have to have their corporate name present anywhere visible to the user at any point in the process, which I rather like because neither Payloadz nor e-junkie screams “responsible businessman of the sort I am glad to pay to get educational materials from”, which is sort of what I’m going for.

I would still heartily recommend Payloadz to people selling, say, $5 e-books on eBay.  And eventually I’ll stop paying e-junkie and roll my own IPN solution.  Perhaps.  I figure that will take me 2 hours, and $5 a month is not really worth 2 hours of my time…

Discounting the one-time startup costs (dang international fax that cost $17 or so) I’m currently profitable.  Basically, I’ve sold enough in July (by the 25th) to cover all costs for the month of July and also absorb that unfortunate decision to try selling on eBay.  If I get one more sale during the month, I will be strictly-speaking profitable.

My conversion rate of downloads to purchasers is a little hard to guesttimate since I got a massive surge of downloads from pirates last week.  If you treat all the pirates as potential customers, I have a conversion rate of just over 1%.  If you exclude all downloads from people referred by my weblog (JoSers who are not exactly in the market for elementary reading software) and by the pirates, my conversion rate is, and this is a back of the envelope calculation, about 5%.

My plan is, within the next week, to cancel my account with Overture and triple my budget on Google to $90 per month.  That would put my marginal costs at about $95 a month ($110 if sales exceed $100, and I rather expect that they will), which is slightly less than what I get from four sales.  I think thats easily achievable, considering that four sales was my initial goal for August and I’m already halfway there with a third of the advertising in a month where my target market is very not-motivated to buy.

I think I’ll celebrate with a milkshake.  Hmm, do any accountants in the audience have an opinion on whether I need to count that as an expense? :)

If you’re like me, you can’t afford at the moment to actually absorb a click from everyone who wants to click on your AdWords ad.  So, at some point in the day, you go over your budget and then its no more ads for you.  You can use this unfortunate circumstance to your advantage.  Go to your favorite ad group, check all the keywords, and click Edit Keyword Settings.  Now turn Position Preferences on, and back out to the keyword list screen.  Set your preference to exclude some ad positioning slots, preferably the top ones, and watch your CPC go down.  You’ll probably still run out of money during the day, but you’ll get more clicks for it.

Why does this work?  Well, by default, Google bids to win on your behalf, where winning is defined as “as high a position in the ad rankings as I can accomplish given my maxiumum bid”.  Suppose, for sake of argument, that I’m bidding on a particular keyword with my maximum bid being $.25.  I have competitors at $.20, $.10, and $.08.  By default (and assuming similar CTRs for all ads), Google will give me the top spot and charge me $.21.  If I’m capped at $1/day, that means five clicks and I’m done.  Some days you might get that in two hours.  However, if you either decrease your maximum bid by hand or put your position preference at 2-10 (which forces Google to bid down from your maximum bid until you’re not in the top spot), you’ll get the 2nd spot at $.11.  Allowing you to afford almost twice as many clicks per day.

I’ve been doing this for about 5 days now and I rather like it.  Lots of things are constantly changing in my AdWords account, but I’ve cut my CPC for my top keywords literally by 70%.

Before: my maximum bid was .25, and my average CPC hovered between .21 and .17 depending on the day (I’ve been getting it towards the lower end since Google’s recent quality-score update).   After the quality score update, I noticed that I was averaging position 1.2… and running out of money within 3 hours of my ads turning on (I have a 10%ish CTR for my best targeted ads).  Not good!

After: same .25 bid, about 8% CTR… and I’m paying nine cents each, for about a 2.8 average position.  I’m still getting my budget’s worth of clicks every day, execpt there’s a lot more of them.  My cost per trial download has declined from $.50 to $.30.

I finally fixed my layout woes.  After I had diagnosed the problem (bloody IE and its bloody standards noncompliance *grumble grumble* do you have any idea how long I defended you against Firefox *grumble grumble* how about showing a little loyalty here) I was able to fix the whole problem by passing my entire web directory through an XTML cleaner-upper.  Which reminds me that this project would absolutely not have happened without the following.  I hereby pledge $20 to each of them “in the indefinite future, when I have the money to spend”, because I’ve gotten that much use and more out of them and I am feeling in a profoundly appreciative mood to people who save me time at the moment.

1. NVU – The last web editor I’ll ever use.  It doesn’t have everything I need, but it has 99% of it, and it beats the heck out of notepad.
2. HTMLTrim/Tidy (half to each — HTMLTrim is a GUI interface to Tidy without which I would have had to actually WORK at getting the problem fixed) — The aforementioned automated pretty-print and HTML scrubbing program.  Saved me from having to write gawk scripts to do it for me, which is probably worth four hours of my time.
3. Eclipse – My IDE of choice for this project.  Someday I’ll figure out how to do GUI design the on you, too.  In the meantime, you’re almost as cool as NetBeans and, in my personal experience, more stable.
4. Filezilla – A bog-standard FTP client, except free.
5. Paint.net — Photoshop, on a budget of “free”, for people who have less artistic talent than an avante garde artiste who outsources his creative efforts to housepets.  (A hearty thanks to the person who suggested it to me.)  As to why I don’t use GIMP: I just don’t bloody understand HOW to start using that application.  Paint.net is literally as easy to use as MSPaint, my former image editor of choice.