Service
- How frequently is the data updated?
Data is updated once an hour for the vast majority of tables. For some of the more advanced analysis options, such as cohort analysis, the tables are updated once a day.
- What kind of uptime can I expect from your data-capture servers?
Our system architecture is designed to provide 100% uptime even if multiple server failures occur simultaneously. Kontagent has multiple, geographically-distributed data centers, each with multiple servers to ensure the best possible uptime. Using a two-tiered, DNS-driven server selection methodology, the app server-side software will automatically seek out the nearest server that is up.
- If your servers go down, how will that affect my application?
All messaging to our servers is done asynchronously, so your application will continue to function as usual. The asynchronous requests will time out and will be aborted. This can increase the number of simulataneously open connections that your server has to support.
- Will you keep our data private?
Yes. Each application is stored independently in its own sand box.
- How much will your service cost to use?
Please see our pricing page for more information.
- Do you support platforms other than Facebook?
Currently the only platform that we explicitly support is Facebook. While we welcome people to use the web API for monitoring platforms other than Facebook, please be aware that at the current time our focus is on Facebook and the support we can provide for use outside the Facebook platform is limited.
- How long do you store our data for?
At this point in time, we have no plans to remove any of the data. If we do have plans to do so in the future, you will be notified and we will allow you to download your data before we remove it from our servers
- What is your data privacy policy?
You can find out full data privacy policy here.
- How do I use Kontagent on my iPhone/iPad?
The iPhone/iPad does not support Flash so the graphs will not show up. You can still view your data in table form by clicking on the "Export" buttons next to each graph placeholder.
Instrumentation
- How long does the instrumentation process take?
The basic instrumentation process can take less than an hour. However, there are many additional features of the system that can be used to fully instrument the application and provide additional insight about your application. The time taken to instrument the application with these monitoring capabilities will vary dependent upon the number of instances in your code that you send invites, notifications, and e-mails, that you post to feeds, and the number of different templates you use for messaging. The instrumentation time also depends on the language you are using. For example, instrumenting a Ruby on Rails app is faster than instrumenting a PHP app.
- Which languages do you provide libraries for?
Currently we provide a library for PHP, with other languages in development. As an alternative, or if you are not using one of the languages where a client library is available yet, you can instrument your application using the web REST API method described here.
- How do I know if the analytics are working? When should I expect to see data?
Check the dashboard to see if any data is displayed for your application. Data is loaded every hour so there may be a lag time of a little more than an hour from when you first send us data to when it is displayed on the dashboard.
- How can I test whether I have instrumented my application correctly?
We provide a test server that will give you instant feedback when you provide data. To learn more about the test server, click here.
- When I was using the test server, I saw data, but after switch over to production, I don't see data on the dashboard! What's wrong?
The most common cause of this error is that the data is not being sent
to the right URL. Please make sure the server_url is set to api.geo.kontagent.net (note the "geo").
- Does Kontagent support FB batch API calls?
No, we currently do not support this. If you require this, please email us at support@kontagent.com to let us know. If there is enough demand, we will find a way to provide support for this.
Privacy
- Do you need my Facebook secret key?
No, your Facebook secret is not required for the Kontagent service. Any reference to a secret key in our documentation refers to the Kontagent secret key which is used to authenticate data sent to Kontagent.
- Am I breaking Facebook's TOS by using your analytics?
No, the data that you are gathering by using our service does not violate Facebook's TOS.
- Will you ever sell or share my data?
No, we will never sell any data that can be attributed to a particular application or application developer. We may however, aggregate data across multiple application and make that data available to other developers. Note that it would be impossible for other developers to reverse engineer the data to determine which applications the data was from.
Troubleshooting
- Nothing's showing up on my dashboard, what should I do?
- Check to make sure the API key in the dashboard matches the API key for the data you're sending Kontagent.
- Go to kt_settings.ini and USE_TEST_SERVER to be true and rerun php -f kt_setup.php. Alternatively, you can go to kt_config.php directly and change $backend_host to point to test-server.kontagent.com. Send an invite or a notification and see if it shows up on http://www.kontagent.com/dashboard/tools/test_server/.
- I stopped getting data, what should I do?
Go
to kt_settings.ini and USE_TEST_SERVER to be true and rerun php -f
kt_setup.php. Alternatively, you can go to kt_config.php directly and
change $backend_host to point to test-server.kontagent.com. Send an invite or a notification and see if it shows up on http://www.kontagent.com/dashboard/tools/test_server/.
- My virality is zero, what's going on?
Check the viral installs on your dashboard, it's probably zero. You can doing the following on your app to see if your URL contains the appropriate kt related parameters in its query string.
- Have two Facebook test accounts handy.
- Set up your test accounts so that one (we'll refer to it as Accnt A) has already authorized your app, while another (Accnt B) hasn't.
- Log in as Accnt A and send an invite or notification to Accnt B.
- Log in as Accnt B. Accept the invite or click on the notification link. Now it should direct you to the Facebook authorization page. Pay close attention to the value assigned to the next parameter. If post-authorized redirect URL is not specified, Facebook will forward to next after authorization. If post-authorized redirect is specified, it will forward to your post-authorized URL with the next value attached to it. Check to see if you have kt_ut or kt_sut in the URL. If not, go back to your original notification link to see if you have kt_ut and other kt_* parameters attached to it.
- After the application is authorized, see if you have a ut or a sut parameter in the final url's query string. If so, kontagent will track your install conversion. If you have specified a post-authorized URL, you will not find ut or sut in your query string; however, you should find a next parameter with your original URL properly escaped. You should be able to find kt_sut or kt_ut among the escaped characters. If so, Kontagent will be able to handle that.
- If you are using our REST API, please ensure that Invite Sent (ins) and corresponding Invite Response (inr) messages have matching unique tags, as well as matching st1, st2, and st3.
Viral Optimizer
- My install conversion for a message created with the Viral Optimizer is greater than the response conversion for certain days. How can this be?
The install conversion is referenced to the day that the install occurred, while the response is reference to the day that the invite or notification was sent, not when it was responded to. For example, consider the case where a single invite was sent to a single recipient on 1/1/2009 and that recipient responded and installed the application on 1/2/2009, and no other invites were sent or responded to for this Viral Optimizer event. The invite conversion would be 100% for 1/1 and 0% for 1/2, while the install conversion for would be 0% for 1/1 and 100% for 1/2.
|
