Every week, the staff at Weatherbug meet to go over the past seven days' emails from users.
The techies want to make sure the download is as smooth as possible, ad sales are looking for input on sponsors, the content team want to make sure the layout and data presented are useful, etc.
A couple of years ago, just as subscription Web sites were starting to get a lot of public attention, Weatherbug's General Manager Andy Jedynak noticed a clear trend in customer email.
"One of the largest bodies of feedback was, 'I love Weatherbug, but I would be interested in getting a version without advertising.'"
Despite the recession, the Company's ad sales were fairly healthy (see Part I of this Case Study to learn why; link below). The team did not have a pressing economic need to turn to subscription.
Also Jedynak knew that unless you are prepared to go very forcibly in the paid direction and slam down a paid-only barrier over the vast majority of your offering, only a tiny percent of your users will convert to paying.
However, with 17.5 million users "only a tiny percent" might still be worth pursuing.CAMPAIGN
The team began with careful price-point research by surveying their own users.
Jedynak's rule is require an "in-tab" of at least 1,000 answers per survey. (The "out-tab" is the much larger number of people you send the survey to in order to get 1,000 to respond.)
He explains, "With 1,000 people who make decisions, it's a plus or minus 3% margin for error with a 95% confidence level. I'm very comfortable with that."
However, what people say they will pay and what they actually end up paying are often different. Next the team ran a test subscription offer campaign to a group of Weatherbug users. They tested two different annual price points ($39.95 and $19.95) and a month-to-month payment option.
Surprisingly, $19.95 won in terms of profitability. (Although a higher price point may mean fewer orders, it can often be so profitable it's a better offer than a lower price.) The month-to-month vs. annual results were pretty much even, but the costs of processing monthly orders made them less profitable.
Therefore in the end, the team decided to launch with $19.95 per year as their sole offer.
In October 24, 2001 the first offers began to role out. Jedynak's obsession with user-experience meant he did not have the offer sent to the whole file at once. "We rolled out in chunks. The first was just 10,000 to test the system and make sure our transaction processing and operations were working, and that email support was doing a good job."
He asked that customer service begin tracking in-bound email from paying users separately from others so the team could see what types of problems came up, and also measure any possible increased customer service costs against subscription revenues.
"We track and categorize every single email for free and pro users every week in bar charts and pie graphics. Plus there's a 'voice of the customer' section that tells you what numbers don't tell you. We built it in-house using brute force and a spreadsheet."
Next, the marketing team began testing various creative to see what would get the most conversions.
They based their tests on what was working for current advertisers: Being sure to always include a clear click command. Two tests in particular stood out in the range of experiments (link to samples of each below):
Test #1: Slick ad based on customer surveys
>From extensive survey data, Weatherbug marketers had learned the number one reason why users liked the service was it gave them a feeling of control, because they could plan their lives better with top-notch data on upcoming weather trends.
Marketing created a beautifully-produced ad showing a human face (people like to look at other people) and featuring language about control inspired directly from user comments. The colors were mostly blue, which is 80% of the population's favorite color.
Test #2: Folksy ad "from the programmers"
This ad is the opposite of the carefully created slick one. "We just slapped it up there," says Jedynak.
It looks like a simple typed note in black ink on a white background with a little bug icon at the top that looks more like a kids' stamp than a formal company logo.
The note is "from" Weatherbug programmers asking for subscription money so they can get some pizza.
Aside from measuring how various creative approaches like the two above did, the marketing team also measured how each ad worked based on the number of times an individual user had seen it.
Weatherbug now has more than 75,000 paying users and about 150-300 new paid accounts are added each day. Jedynak says, "Realistically, if we keep on improving what the Pro version has to offer, maybe a little extra special info that free users don't get, we'll cap off at 2-3% [conversion of free to paid.]"
The team discovered that it is only worth making a subscription offer about seven times to an individual.
"We find with newer users, if someone is going to convert, it's most likely in one of the first seven times. So we make sure we put a frequency cap on the ads. Once we send them half a dozen times we pretty much stop letting them know. We just keep a button on the surface called 'subscribe' they can click anytime they want. Mainly newer users click it."
While the customer service team learned paid users do tend to email more than others, it is not so much of a burden that the current team can not handle it. (Weatherbug now handles all inbound email with just two full-timers and one part-timer.)
The folksy ad has consistently been the biggest winner. Jedynak says, "I think it's got something to do with the way people perceive the brand. It's an endearing brand." Data:
Creative Slick Ad Folksy Ad
Impressions 504,839 792,859
Clicks 20,493 50,009
Click percent 4.06% 6.31%
Orders 111 565
Click to conversion .54% 1.13%
View to conversion .022% .071%
Gene Slyman, Weatherbug Marketing Manager says, "It's also worth noting the this conversion difference runs essentially across all creative we ran: wraps, pops, email and homepage ad. What also is noteworthy, the pizza [folksy] creative has seen at least 5-10 millions of impressions, and the conversion rate has only dropped slightly. That's amazingly consistent performance."
Here is a sample of both creatives:
Note: If you missed last week's issue, Part I of this Case Study which focused on Weatherbug's ad sales tactics is available here: