Phil's blawg

I spent the last two weeks working on a dramatic change to BudgetSimple, which I finally was ready to get some feedback on.  I just started a users group for BudgetSimple, where customers can get an early peak at features, and give feedback.

So I started writing an email to these users about this new feature.  The first paragraph talked about why the change was made, the second one apologized for things they’d need to do, the third asked them if they understood the change, etc… 

As I was writing the letter I realized, this feature sucks.  Or at least in it’s current state.  It required effort from the users, it made me feel defensive, etc…  This feature I was proud of just 10 minutes earlier, now seemed like crap.  And it was.

Just the act of writing the letter though made me think of a much better way to implement it, that would cause less pain on the users and would be much better to explain the benefit.

Even if you don’t have a user group, write a letter to your users explaining your latest feature.  You don’t need to even send it, but I bet the act of putting the feature into words will give you need clarity.

Probably the question I dislike the most from investors: “What’s to keep someone from just creating this idea themselves?”  On the one hand, it’s a good question, if someone had a secret weapon, it certainly would make it harder for others to compete, but on the other hand, many successful companies do not have this.

Just to name a few off the top of my head that “someone could write a copy of in a week”:  37signal’s Basecamp, Facebook, AirBnB, LinkedIn, Angie’s List, Kickstarter, Groupon, Yelp, Pinterest, Etsy, eBay, SalesForce, GoDaddy, SurveyMonkey, craigslist, Tumblr, Twitter.  (And this is just tech companies, consider traditional retail like Wendy’s and Best Buy and it’s even more straight forward).

Yes, any CMU senior could probably write a copy of any of the companies features above in a week, some in a weekend or an evening (ignoring the technical issues many have to deal with at mass scale). And yet it would sound foolish to say “Oh, I’ll write a copy of Twitter, and take them down”.  Because the most valuable thing they had is execution, which then led to network effects and essentially being a “household name”.

To compete with any of the companies above, you can’t simply copy the features and expect to dethrone them.  Some (but not all) were the first into their space, and have created an almost unstoppable brand.   In many cases, there were similar products, with better features competing against them, and yet they won.  So it seems like a smarter question to ask is: “How will you become unassailable?” or “How will you make your brand synonymous with this space?” 

In the wake of the terrible tragedy in Newtown, lots of people have been trying to score points for their favorite side of a decades old debate in America: gun policy.  As someone who has no strong opinion on this issue, as I neither want to own guns nor live in fear of them, I’ve mostly stayed out of this debate.  But misleading statistics passed around the Internet this weekend have really bothered me.  I just wanted to put out what I feel are the more accurate statistics to compare, and you can make your own decisions from this unbiased source.

The most common one that bothers me is an Apples to Oranges comparison, that is people will often quote things like: “In 2009, the United States had 9,146 deaths by guns, while the UK only had 18”.  On the face of it, that statistic accomplishes it’s goal, holy shit we have 508 times as many murders then the UK!  Even if you normalize for the first problem, that the United States is 5 times larger then the UK in population, you still only get around 100 deaths in the UK, meaning a 91 times difference.  

Unfortunately, this is a false comparison.  It sounds like it makes sense to compare gun deaths to gun deaths when the debate is about removing guns, however I think the actual goal is to reduce deaths.  If you replace a gun death with a knife death, I don’t think anyone will feel much better.  Newtown is just as tragic if it was a maniac with grenades and a katana.  Any statistics in this debate should be using ABSOLUTE homicide numbers of any method.  Using this method, the UK in 2009 had 724 homicides compared to 14,159 in the USA.  Normalizing for population, that would give the USA a homicide rate of only 3.7 times that of the UK (still a difference worth noting, but not quite as shocking as the 508 times number above).  The USA basically has a non-gun related murder for every two gun related murders.  The UK on the other hand has 40 non-gun related murders for every gun related murder (using 2009 data). 

I’ve also seen some scatter plots of gun ownership per capita and homicides per capita, that suspiciously left off many countries.  

I created my own with all countries of the world plotted.  With gun ownership per capita (per 100) as the X axis, and homicides per capita (per 100k) as the Y Axis.  So the higher up you go, the more people are getting killed, and the further right you go, the more guns people own.

As you can see, the USA is a crazy outlier on the X-axis, owning far more guns per person then any other country.  On the Y-Axis, El Salvador has far more murders.  I’ve seen lots of people “crop” this so it doesn’t show the countries above the 5, so that USA looks like it has the most murders by gun ownership, but based on this chart alone, it doesn’t look like there is a clear correlation between gun ownership and homicide.  It is a valid point that comparing El Salvador and it’s non-functional police force with the USA isn’t fair (although I’d say Brazil or Uruguay are probably more fair), so lets do this graph again, but this time only with richer countries, I’ll pick the top 75 (using GDP per capita).

These are mostly westernized countries with modern economies and governments, although as you can see, the ones with high murder are ones that many would still consider “developing”, so let’s narrow it to just the top 25 by GDP

You can see the USA in this group of richest of rich countries, has about 75% more homicides then the worst of it’s rich friends, but 200-700% more guns per person.  (It’s clear that increasing GDP has a big effect on reducing homicide!)

Finally, this next graph is the most telling of our problem, I looked at “rampage killings” that resulted in more then 10 deaths since 1990 (this is excluding terrorist attacks).

You can see the USA leads the way by a disgusting margin.  That said, the rest of the countries are a mix of GDP, religions, cultures, social welfare, gun control, etc… and they’ve all had rampage killings in the past 20 years.  We have a serious problem and need some kind of change, but it doesn’t seem clear that any one change we can do would eliminate madmen killing innocents entirely.

My sources: 2009 Firearm vs non-firearm homicide, GDP per capita, Murder rate per capita, Gun ownership per capita

1. Respond to a customer email with a personal reply
2. Do so within 24 hours.

Sounds simple?  Follow these simple steps and you’ll be in the top 1% of consumer services.

Stay tuned for the next tip, “How to get a rabid fan-base of customers” (Preview: Respond to a customer email within an hour)

With people saying Hostess failed because of lack of sales and a more health conscience public, I just wanted to remind everyone of these pictures I took during Snomageddon 2010.  It was the first time I had seen a serious shortage of groceries as roads were blocked, and this is what we saw at the grocery store

Compare to the fruit

(This is the East Liberty Giant Eagle off Shady Ave)

Maybe this is something that’s well known in the industry, but I just started using it so I thought I’d share.

Whenever I reach out to reporters or potential business partners, I always write a personal email (thus I don’t use MailChimp).  The problem with personal emails is they are harder to track your marketing success.  How do you know if someone actually checked out the link you sent them?  They didn’t reply to your email, but let’s face it, people are not good at replying to emails. Wouldn’t it be useful to know if they checked out your site and hated it, or just never clicked the link in the first place? Enter the Google Analytics UTM tag.  It’s the best way of figuring out who is clicking your links, and how deep they explore.

There are a few keys to this.  First, use an organization strategy that makes sense.  For example, I use the Campaign parameter to track the type of marketing, so “press” for media contacts, “partners” for biz dev, “internal” for emails generated by my app, etc…  I then use the Source parameter to track the specific media outlet/biz partner.  So if you received a press release from me, the link would look like this:

http://www.BudgetSimple.com?utm_campaign=press&utm_source=nytimes&utm_medium=email

It’s important that you hide these ugly parameters in a better looking link.  You can use either a URL Shortener  but I find these are less trustworthy for people to randomly click, so I usually make an HTML link, so the text is the URL but the HREF is the UTM tag.

Now you can look in Google Analytics, under Traffic Sources->Sources->Campaigns and see who is clicking your links!  Because these links are so specific, you can also tell how engaged they were (ie, if they bounced they probably had a poor impression of your site).  This will also help you taylor future emails if you need to follow up.  

It’s also important to note that this will cookie the visitor of the link (so even if they return to the site another way, you’ll still know where they came from), so don’t click your own links!

Although the title sounds anti-Obama, let me first be clear that I’m pretty much apolitical and don’t believe the President actually can affect the stock market via policy.  I do believe they can affect it via sentiment, like if the President said “We’re in some deep shit”, the stock market would probably panic, but obviously a President would never do that when the market is up.

I came to this observation listening to the debates.  President Obama and his supporters often say he inherited a bad economy, which is true because the stock market and housing bubble crashed at the end of George W Bush’s 2nd term.  I remember GWB and his supporters claimed the same thing when he came to office, which was true, the stock market crashed and the dotcom bubble collapsed in Bill Clinton’s 2nd term in office.  Which got me thinking… does a two term President cause a stock market crash?

Let’s look back at two term Presidents:

George W Bush - *Crash in 2008* at the end of his 2nd term

Bill Clinton - *Crash in 2000* at the end of his 2nd term

Ronald Reagan - *Crash in 1987* at the end of his 2nd term

Richard Nixon - *Severe depression in 1972*, in the middle (but what ended up being the end!) of his 2nd term

Before this, the trend doesn’t hold, while stock prices declined at points during Lyndon Johnson’s 2nd term, and maybe we can argue he didn’t have a full first term,  the stock market did great during Eisenhower’s 2nd term, and of course the worse crash of all time in 1929 was during the first term of Herbert Hoover.

So this could all be a coincidence, as the sample size is only 40 years.  All of the crashes in the past 40 years have been a result of bubbles collapsing, and in the case of 1972, an oil embargo, but I wonder if the bubbles collapse partly because the same person has been in office, and is guaranteed not to return?

Even more interesting, the first term of ANY elected President (so Ford’s first and only term doesn’t count) seems to be a period of great or at least steady stock market appreciation.  So short the market if Obama is re-elected and long (but only 4 years!) if Romney is elected?

*Updated 10/24/12 with a response from WP-Engine at the bottom*

My friend John posted a glowing recommendation about WP-Engine the other day.  I see a lot of people praising WP-Engine, so maybe I just got the short end of the stick, but I promised him I’d talk about why I found it so lacking.

First the back story.  Last year I was working for a consulting company that has a very popular blog.  Unfortunately, like many small companies, there was not a dedicated IT person for infrastructure, and their blog was hosted on a local ISP with a shared server, and so there were constantly problems.  WordPress would get hacked, the database would get corrupt, etc…  So I jumped in and was excited to suggest WP-Engine, which I heard so much about and which promised to take the management away from us and solve all of these problems.

WP-Engine costs an order of magnitude more then crappy shared hosting, so it really had to provide a lot more value to justify this, but I figured it was easily worth it, as downtime could cost the company business.

Unfortunately the problems started as soon as we began our switch.  WP-Engine asks you how you want to domain configured (www or non-www) and even though I had set it up to be www, when we switched our DNS to WP-Engine they had configured it wrong and there was a big error on our site.  No worries, problems happen… so I called WP-Engine.  That’s when I learned they have NO PHONE SUPPORT!  My site was down and they were asking me to hangup and email them to open a support ticket.  There was another option for “Critical issues” that you could page them, so this is what I did.  It rang and rang (probably trying to page people) and eventually I got someone who was out at lunch who promised to call someone at the office and get it fixed, which they did (in probably under an hour).

So that wasn’t so bad.  It was disheartening learning we wouldn’t have phone support, when even the crappy ISP we moved from had this, but with their 100%(!!) uptime guarantee, I hoped we wouldn’t need it.

We had setup Pingdom to tell us when the site was down for at least 5 minutes when we were with our previous provider, and soon after signing up for WP-Engine we started getting outages.  They would typically be just little ones (5-10 minutes) but enough to register on Pingdom.  Other times we’d load our blog and get an error, only to refresh a minute later and it would be back (meaning there were clearly outages that Pingdom wasn’t picking up on).  

What really burned though was after a particularly noticeable outage we tried calling WP-Engine, again using their paging system, but this time no one picked up (this was during business hours).  The next day we got an email from WP-Engine (I was expecting a mea culpa) but it was instead telling us that they had to bump us up to the Business tier because of our traffic levels.  This was really insult to injury, but also confusing.

I originally signed up for the $99 tier because it says up to 100,000 visits per month.  Google Analytics was showing ~50k visits per month so I figured this was plenty of room to grow.  WP-Engine was showing that I had something like 150k visits though!  They are not very clear on how they count a visitor, but I can only assume it must be IP based, so you will get penalized if you are crawled frequently by Google (which we were).  Ugh, $250 a month now when we were paying $30 per year?  Getting hard to justify my decision considering we had more downtime so far, but it looks like the Business Plan comes with phone support, so that sounds great.  I emailed back to WP-Engine and said:

“OK, we’re on the Business plan, now how do I get that phone support, there is a secret number right?”.  

“No, just call our main number and then press the option to page someone”

“Uh, I’ve done that and sometimes got no one”

“What? When?”  (He then proceeded to ask me for specific times and dates, which I had, and after much back and forth said they were working on making phone support better and gave us a credit for our downtime)

The bottom line for me was that a paging system is not phone support, especially where sometimes there is no answer!

We continued to have downtime issues with WP-Engine (they seemed to be a DDOS target).  Here’s how the typical response would go:

“Oh man the site is down again” *Calls WP-Engine, hits ‘2’ to page their emergency support… no answer*

“Fuck” *Emails WP-Engine to open a ticket, no response after 20 min*

Then I’d contact them by Twitter, where they would almost always respond instantly!  Usually some low level person would say they’d look at my ticket, then I’d get a response like “It works for us, is your Internet up?”…. “Yes, I’ve checked this from a variety of locations”  *no response….site still down*

Anyway, it went on like this several times (they actually blocked one of my co-workers Twitter accounts, WTF!).  This is the number one reason I can’t recommend WP-Engine.  You can’t sell yourself as a professional hosting solution and not have phone support.  They would also just randomly disable our plugins causing the site to break in other ways “This plugin isn’t supported now”.  Their metering of the site is shady, and their outages were way too frequent.

They also offered a CDN as a premium feature.  In our testing, the CDN was usually SLOWER then not using it.  Sometimes daily backups didn’t occur… just a whole manner of crap.

I’m also very dubious of their client list.  One time their ISP had a major failure and all WP-Engine sites were down, but they claim FourSquare as a client, and their blog was up?  Maybe there is some elite tier of servers that gets geo redundancy.  Their uptime status page was also not helpful.  I’m talking about a 5-7 hour outage on a business day.  The only reason I figured out what happened was I also use Linode, and one of my slices was affected by the same outage, but their status page was great about keeping us up to date on the downtime.

On the positive side, they do actually do a very good job about fixing hacks.  We got hacked once when we were with them and their experts cleaned it up and fixed it quickly (once we got ahold of them!).  So there is promise in the idea of a managed WordPress provider.

Maybe they’ve gotten better, I understand startup growing pains, but with something as critical as website hosting, I wouldn’t recommend starting with someone still trying to get their act together.

*Here’s a response from WP-Engine (Austin W. Gunter - Brand Ambassador @austingunter)*

I’m the community and content marketing guy for WP Engine. I wanted to write a quick note in response to your post about your experience hosting with us: http://jamespanderson.tumblr.com/post/34114537491/why-i-cant-recommend-wp-engine

First off, thanks for writing the feedback about WP Engine. As a managed hosting company, we hold ourselves to rather high standards, both in terms of technological performance, but perhaps more importantly in terms of the service we provide via our support channels. That said, based on your post, you had a less than stellar experience with us.  That’s not really ok in my book.  While every company makes mistakes, let me go ahead and apologize for the trouble you all had. I hope that you’ve found a hosting home that you’re very happy with in the meantime.

There’s a few things you mentioned in the post that I wanted to own up to, as it were. First of all, offering 100% uptime was a failed experiment that we ran. Even Amazon S3, which by most accounts is the model for uptime has 99.9% uptime in their SLA. We’ve since remedied this in all our messaging and are setting proper expectations.

Again, with the support you received, I’d be pretty upset too, so I can appreciate your blog post.  Your experience doesn’t live up to the high standards we set for ourselves, and it doesn’t live up to the normal customer experience at WP Engine.  You may not have received an apology from us, but please accept mine.  I’m sorry for the bad experience you had.  Rest assured that your time with us is not representative of the normal customer experience with us.

For the visitor count of your site, we’re actually rather clear on our site now about how we count visitors. You can check the pricing page, and I also wrote a blog post about how we count visits. Basically, as a host, we’ve got to serve everything that hits our server and this has to factor into our pricing in order to remain profitable, and also to ensure specific sites don’t consume disproportionate amounts of server resources compared to other sites that are paying the same amount each month.

For the CDN issues, I’d be very curious to know how you’re running those tests. We’ve seen various ways of running speed tests recently that are so inaccurate, yet some reputable developers were actually relying on them as if they were scientific.  

And for our client list, I assure you that we’re representing ourselves, and those brands accurately and appropriately. You’ve mentioned a lot of technical aspects of hosting, so you’ll understand that while some sites may be affected by downtime, others may not because they are served from different data centers. Or it could also be different clusters within the same datacenter. Basically, connectivity issues at one datacenter may not send all the servers there offline.

WP Engine has servers across the US, as well as the UK, and Asia. Depending on where your site is hosted, you may not be affected the way another client of WP Engine’s is.  This is consistent with other large-scale hosting companies that host enterprise-scale WordPress sites.

As a company, WP Engine has done a lot of growing this year. Not only in terms of customers, which is an important measure, but not the end-all be all. We’re maturing as a company, refining our ops, and adding more tech support. As with any growing company, there are challenges to scale, and when we make mistakes or hit bumps in the road, we make sure to own our errors and make sure that appropriate measures are taken so that problems we encounter once are resolved for good.

Phew.  All that to say, thank you for the blog post, Phil.  I appreciate the feedback. If you’re ever interested in giving WP Engine another shot, please let me know so I can make sure everything goes smoothly.

It drives me absolutely nuts, it’s like Alta Vista bad.  I even turn personalized results off, and yet for some reason Google wants to give me 10 search results from the same domain.   Open incognito window and I get wonderful normal Google Search. Tell me this is a bug, because otherwise I’d recommend searching in another browser at all times.

Signed In:

Incognito

Wait what?  INCREASED support requests?  Isn’t our goal at startups to reduce the support requests?  After all, support requests means time taken away from developing code right?

That used to be my general thinking.  When I was working on BudgetSimple part-time, my goal was to try to make it so almost everything was self-serve and simple so I didn’t have to deal with support requests during the middle of the day.  

One day though, I was looking at my conversion funnel, and noticed there was this significant portion of people who signed up for the service and never used it again.  Although the “Cancel Account” link asked people for feedback, most people didn’t even bother to cancel.  No amount of analytics or logging could help me figure out why these people were canceling, they just setup an account, played around with it for a while and never came back.

Trying to think how I could understand these users, I remembered back in the day, when I signed up for FogBugz and the next day received an email from Joel Spolsky asking me how my experience was going.  This was obviously an automated email, but even knowing that, it still felt personal and the reply address seemed to be Joel himself. So I decided to try this at BudgetSimple.

Our users are even less technical then FogBugz users, so I expected a very high response to this email.  Boy, I was not let down.  I scheduled a script that would email everyone the day after they signed up with a personalized, unfancy email that simply asked if they needed any help, and it came from me directly.  

Suddenly, my support emails went from 0-2 per day, to 10-12!  People who otherwise would have just never returned either gave me useful feedback: “Actually Phil, since you asked, it wasn’t working at all in my browser” or motivating praise: “Wow Phil, I was trying a couple solutions, but your personal attention will make me use your service for sure!”.

In short, I was able to re-engage users and give them a feeling of personal attention that is rare on the web.  I was also able to find tons of bugs and problems I would not have otherwise found, and the retention rate has increased, and slowly the support emails are decreasing.