Mac Poker posted the transcript of an interview with yours truly.
That’s my personal good news for the day. My personal bad news: I got banned from a Poker forum I joined today. They didn’t seem to like me joining in a conversation about Poker Copilot, as the developer. And I thought I was treading softly…
Last month, Google users found http://pokercopilot.com/ through 237 unique search phrases. After removing the ones that directly specify “Poker Copilot” in a range of spellings, there are about 230 phrases. The top phrase – “poker tracking software mac” – sent me only 2.5% of my Google traffic.
Which makes me think that the conventional approach to search engine optimisation is flawed. That approach is to choose a few highly relevant keywords, and do everything possible to rank highly for them. But it ignores the reality: that websites receive Google traffic through an ever-changing, highly varied long list of keywords.
What are the ramifications of this?
It’s well-known that search traffic follows a “long-tail” distribution. Amazingly, once I remove the top phrases, all of which directly specify “Poker Copilot”, what I’m left with also follows the long-tail distribution.
For Google Analytics users, Juice Analytics have produced Concentrate, a tool to help make sense of the long-tail of Google keywords.
For the last few months I’ve given people who purchase Poker Copilot a survey with exactly one question:
“Where did you hear about Poker Copilot?”
I collated the answers for the 100 most recent purchases:
The problem is, I don’t exactly know how to interpret these results. Does the dominance of Google in this chart mean that:
Of the various forums in the survey, 2+2 dominates. I search 2+2 almost every day for any mention of Poker Copilot. I announce major new releases there. Does the data indicate that I should spend even more time on 2+2, or that I’ve been neglecting other forums?
I have this problem in general with the mounds of data that I collect from various sources: how to interpret and use it.
A final observation from this graph for Mac developers: the Apple downloads site is responsible for a huge percentage of my downloads but very little of my sales.
The 3-bet frequency stat on the dashboard was way, way wrong on the update I released yesterday. It was showing the number of times you made a 3-bet, and not the frequency. Fortunately it was correct (as far as I can tell) in the HUD.
I’ve just released a new update that fixes this problem.
This is another stability/bug-fix release with some incremental improvements. This include:
You can download from http://pokercopilot.com/. Enjoy!
My first rule of stackoverflow.com:
A review of Poker Copilot in French has been posted.
I don’t speak French, so I don’t know if this is a positive or negative review!
For the last few weeks, Poker Copilot asks new users whether we can track their usage:
I’ve now got enough usage data to make an attractive chart showing the most popular Poker Copilot features:
Look at the “Player Name Filter”, in last place. This little-used feature gives several users grief. It is only useful for the tiny group of Poker Copilot users who play online under more than one nickname. So I’m taking it away from the Filter toolbar. As of the next update, the “Player Name” selection is accessible via a menu. And it defaults to “All” players, which means all nicknames you play under.
Interesting is that the only heavily used charts are custom charts and the bankroll charts.
In the next update of Poker Copilot, there’s a revised calculation for 3-bet. It used to be
(3-bet on any street)/(hands played) * 100
Now it is
(3-bet preflop)/(3-bet preflop opportunities) * 100
It gives a vastly different number. Much higher.
On the Business of Software forum, I asked:
What are proven strategies for 1-person companies to keep up with the daily deluge of user e-mails?
Here’s a summary of the excellent responses:
Change the Product based on Common Requests
Examine common requests and consider how you can change the product to eliminate them.
Have a Forum
Customers will often look in the forum to see if someone else has asked the same question.
Test-Driven Development
Have a battery of ever-growing automated unit tests, so repeated failures of the same type are
uncommon.
Beta-testing
Have a good pool of beta testers who can point out the obvious stuff which is all too
easy to miss – that’s invaluable.
Use Defect-Tracking Software/Customer Support Software
Having a good system, such as FogBugz, in place to respond to emails helps.
Virtual Personal Assistant
A virtual personal assistant can offer Tier 1 user support.
Have a FAQ
A FAQ is very important. Whenever you get the same request or question more than 2
or 3 times add it to the help file.
Raise the Price
Low price = high support costs. If you can sell 1 copy for $39.95, or 2 for $19.95…which will cost you less down the road?
Optimise Workflow
Check email only twice a day.
Optimise User Interface
People suck at following directions. Computers do not. Optimize accordingly. Spend extra time on coding and UI to have the computer do more of the work.
Make it harder to fail and easier to recover from failure
Example possible failure: people might confuse your registration key with a transaction number, so you prefix the reg key with a unique string and pre-fill that on the form. This sharply cuts down on the number of people who fill in your Paypal transaction ID in the field, find it doesn’t work, and then immediately mail support.
Use Snippets for Common Emails
Snippets are very helpful in getting a reply out quickly.
Relentless, continuous improvement…
…in program, website text, and support emails as you learn what isn’t working.
Thanks to all whose text I have shamelessly reshaped. Let me know if you want acknowledgement via a link to your site.