For almost a year now we have been building and refining what I believe to be the most complete and accurate analytics solution for mobile. During this period, two major mobile industry players have stepped up with products of their own. While I was concerned about the Bango product initially, I realized quickly that their product has about 10% of the benefits and capabilities of ours. While they have been able to use their name, reputation and cash to infer that they have a more robust product then they have, customers quickly find out that things are really a bit different. Adding a feature like “Unique Visitors”, something that should have been there from the start, and acting like it is some kind of brilliant new idea doesn’t fool anyone. But it got me thinking….
In my now 18 years of experience with the Internet I have clearly learned two things. The first one of which I need to get better at.
KEEP IT SIMPLE! - Next to my desk, stuck on the wall is torn out sheet of notebook paper that simply say “Keep It Simple”. When our minds start to come up with great ideas and features, I look at that sign, record the idea and remember that the goal is to get a working product out with minimal feature that can be easily upgraded on a consistent basis. That’s why we build web based applications. Well apparently I need to get better at this.
The other day we were on a call with one of the world’s largest mobile phone manufacturers. After an hour of talking about Mobilytics I received a response that I have heard now more than once.
“We can really appreciate the level of thought and detail you put into the development of this product.”
Well that’s nice to hear. What I am not so sure about though is whether this is a good thing or a bad thing. Is it a good thing that I can’t sleep at night if one single visitor to a mobile web site is not properly tracked? Is it a good thing that I don’t want the system to go down EVER? The truth of the matter is that we could have officially released Mobilytics back in January by only including the Dashboard. This was still 90% more data than has ever been available on mobile users. But was that the answer?
Well apparently it was! Just Look at Bango and AdMob.
I spent an hour in a webinar yesterday given by AdMob seeing their complete product. I’ve also been playing with Bango and have been through their webinar as well.
Half way through the AdMob webinar it hit me like a ton of bricks. Apparently keeping it simple does not just mean holding off on enhanced features. It means much more. There are a number of things that our competitors have done (or not done) that we could have done to get to market over 6 months ago. I want to point out a few of these.
1. Leave out the very feature (Unique Visitors) that makes the product necessary in the first place (BANGO) - We could have simply done away with Visitors and Pageviews and simply used Visits. It worked for Bango! This is of course the main reason for mobile analytics, and the one issue that has taken the most resources and time to develop accurately. Leaving it out was not an option.
2. Hide the inaccuracy of the data by not showing the detail needed to even know the data is inaccurate! (AdMob & Bango). This one dawned on me during the AdMob webinar. When we present our product, we are always asked about our percentage of accuracy. My answer is always, “we have no way of knowing”. As far as I am concerned, if we can tell somehow that the data is inaccurate, then we can build a system to do the same. That’s exactly what we have done!. If you can eyeball the raw data and see an issue, then our servers can. So the answer is that we are 100% as accurate as we can possibly be given the information and technology available. We have taken incredible steps to identify unique sessions and visitors. What I suddenly realized yesterday was that the reason we need to be so accurate is that we are showing our users the information that will make it obvious if it is not!
We show every single page view individually as well as every visitor, their path through the site, their phone, carrier, country and detailed information, as well as all other visits from that visitor. We show how much money that visitor has spent on the site, what their average monthly spending is, and what their conversion rate is. This is just the tip of the iceberg.
During the webinar it occurred to me that with Bango and AdMob there is no way to prove the information is inaccurate without accurate detail to compare to. And they don’t provide it!
When asked how they identify unique visitors, Omar, the CEO of AdMob danced a bit around the issue, mentioned something about http headers, and used the word “proprietary” to put the question to rest. Bango talks about their “unique identifier” whatever that is. We assign a unique id to every visitor. When they come back, we know it is a return visitor. Yes there are situations where this is not possible. That is because of the limitation of wireless. We all live by that constraint equally. Even Bango’s “unique identifier” will not work in these cases. You can’t ask the user for their phone number each time they visit the site.
So how would you know if the data was wrong? If the report says you have 765 visitors, who are you to question it? How would you know? Well in our case, we show you each visitor and what they did. It will be painfully obvious if the data is wrong. Should we have left this out? Apparently we should have.
3. Stick the word UNKNOWN or UNDEFINED in whenever we don’t know, rather than try and figure out WHY we don’t know. (AdMob & Bango) - UNKOWN?? This is more of the same. How can it be unknown? That’s just an excuse for “we haven’t figured out why we don’t know what this is.” Every visitor to a mobile web site is coming from an ISP in a Country using a Browser on a Phone. How can it be unknown? In our case there is only one case for unknown and that is when an error occurs during the identification of the handset. In that case we label it as “Pending ID”, and it is reconciled with a nightly update. While we might not have details of the capabilities of a phone model right away, we can give the phone a name rather than lump it in with the other unknowns. When details are available, they will be updated.
Interestingly, you accessed the link on a Blackberry using the default internet connection, rather than the specific WAP gateway, hence then unidentified entries.
What he is really saying is that I accessed the Internet through a Blackberry Enterprise server (like 90% of Blackberry users do), and they are not able to identify users that way. What a crock! First of all it should not matter how I access the Internet, you should still know I am a Blackberry. The UserAgent does not change based on how you access. Secondly, the issue with Blackberries is that they all show up from Canada which is where the Blackberry servers are. So here’s the difference in our products. Bango marks it as “unidentified”, while we spent considerable time building a Blackberry specific tracking module that can identify the real carrier and country. Ask them how they handle sites like Skweezer, the Google wireless transcoder, Mowser, Yahoo, Opera, Vodaphone transcoder and other exceptions. I already know the answer. Don’t show the details and nobody will question it.
Here’s the top 3 phones according to Bango for our test site:
Unclassified? Opera Mini? - These are not Handsets!
Here’s ours:(click to zoom):
These are Handsets! To show 10% of all visitors as “unclassified” is not acceptable by my standards. Even for a “Free” version.
And here’s AdMob’s Operator Report from their demo:
And ours (Click to Zoom):
Trading fancy animation and graphics for detail, accuracy and ease of use was not an option for us. Our screens show at least 50% more information in a single place than Bango and Admob and do so without cluttering up the page. AdMob has taken up the entire “above the fold” of the screen with a single graph that moves and grooves while showing one-tenth of the information we show in the same space. Their fancy navigation takes up the left side of the page, valuable real-estate that we use to show important details.
Bango’s spinning wheels and fancy colors make a nice view of the “23% unclassified operators”.
Let’s compare the Dashboards:
I think the pictures speak for themselves.
In the amount of time they spent on fancy Ajax gizmos, they could have made the data more accurate and made the reports more useful. We’ll go into a detailed comparison in another post.
And Lastly for today-
5. Uptime Shmuptime, if it goes down, blame it on our hosting provider (Bango) - This one amazes me. As soon as we realized the incredible number of pageviews were getting, we took a step back and changed our back-end to take advantage of scaleable cloud computing. Downtime is NOT an OPTION. If Amazon’s Computing Cloud should go completely down, we can still track all the data in our other data center and process later. This took us 4 months to re-architect. I suppose we could have just ignored it and got to market faster, but again this was not an option. Bango has been down a number of times recently which not only brought down the analytics, but it brought down our site as well since we used the Bango Links in our menus and ads. All of these issues in the last month affected Bango analytics.
The point of this post was not to attack our competition. (Although I can’t resist showing this screen shot showing 202 unique visitors with 292 unique handsets. Someone tell me how that’s possible!)
The point was to communicate the differences in the commitment to excellence between us. Rather than take the easy way, we have built a product that “keeps it simple” while at the same time provides so much more with a level of accuracy that we believe is unmatched.
To us it’s about the accuracy of the data and the usefulness of the tool. Fancy graphics, animation, expensive PR firms, inaccurate data, and missing features are not our answer to delivering a quality, usable product. Even a free one!