We have recently started using Sitecore A/B Testing and I am getting lots of questions about how the scoring works. I have been through the relevant Sitecore DMS documents but I still am not 100% sure if I understand how the scoring works.
My basic understanding is that the scores are based on Value Per Visit and my assumption is that the the value relates to the whole visit and not just the specific components we may be trying to optimize with the A/B Test.
For example, if option A has a goal associated with it worth 5 points, anyone presented with this option would get 5 points PLUS any other goal values they trigger during that visit to the site. That might add 5, 10, 50 or more to the visit score and then the option A score would be "total visits score/total visits".
Can anyone confirm if my assumptions are correct or explain where I may be off base? Can a user presented with option B change the score for option A?
By default, the Engagement Value is calculated on a per-visit basis. So your assumption is basically correct - and it does make it hard to test how a particular component variation does against another.
That being said; there are tools to help you.
We're currently implementing SBOS Accellerators into our solution. We have the same issue you are describing, and need a more fine-grained approach to testing.
Basically SBOS accellerators will allow you to track individual personalisation performance, not "just" looking at the overall Engagement Value.
Lars Petersen blogs about it here: http://www.larsdk.dk/2014/01/must-have-marketplace-modules-for-sitecore-digital-marketing-system/
Marketplace link for the module here: http://marketplace.sitecore.net/en/Modules/SBOS_Accelerators.aspx
We found a few issues in testing the module, but none were really severe. I know these issues are being fixed if they haven't already.
Related
For example, scanning the wine to tell you its name by Vivino.
But, seriously, scanning its barcode is faster and easier with high accuracy. Why do they design to use AI to scan the product by computer vision (object identification)? Is it necessary, mate?
This is not a question that is appropriate for this site.
Your question is likely to be a statement of opinion.
The answer is, not which is better, but the issue of segregation.
The product scanner does not replace the barcode scanner. At least soon.
For example, there are products that do not have source marking and are difficult to apply in-store marking.
In order to save labor, reduce time, standardize, and mechanize these products, product scanners that use AI will be good parts.
In addition, even for products that already have barcodes, it is possible to acquire the information of the target actual product and display it additionally.
At first, every product has a part that seems useless and exaggerated.
However, according to market evaluations, some of them will disappear and some will be sophisticated and essential.
That's always happening and your question seems to be one of them.
I need help/recommendation for creating an order like input form with MASTER inputs like (order no, date, customer name, address) along with DETAIL info in multiple rows like (Item no, Description, Qty, Rate, Amount).
I am using Admin-on-rest as front-end interface and Loopback for my backend api.
Being new to react/redux and still learning core concepts, I am getting a hard time in finding a good example/starting point to build this functionality. So far, I think this example based on redux-form can help in creating a custom component, but I am yet not competent enough to build this myself.
Any reference to a similar example or some simple code to get me started will be very helpful.
Ref. Image
I think you need this.
How to richly style AOR Edit page
The last answer on the page is a somewhat detailed guide on creating a custom edit component in AOR. Feel free to ask more questions here about how this will be done.
Looking at your design you will need to also think how this data will be updated at the API level. AOR itself will make a single request when you hit save. So how will your API handle updates to multiple models etc.
Could some one help in linking my multivariate testing with goals. I have successfully created A/B Testing and everything working fine apart from the value is always Zero. Is sitecore create value automatically or do we need to set goals for the pages to work? As far as i know we need to set goals for the pages.
I have Followed below sitecore documentation which does not talked anything about how to set values for the tests.
http://sdn.sitecore.net/upload/sitecore6/65/marketing_operations_cookbook_sc65-usletter.pdf
Even if we set the goals for a particular page how sitecore going to recognize whether these goals are accomplished by someone coming to that page directly or from the multivariate testing? I am bit confused.
You're mixing up two concepts here. Sitecore Engagement Value tracking, and goal conversion. From your question I gather; what you're trying to accomplish is to determine which variation of say a banner or a promotion, generates the most clicks?
You can achieve this, but your content editors are going to have to manage how they work with this. In very simple terms, it would be accomplished in this manner:
Set up the M/V test, have each of the variations link to different target pages
On each of the target pages, go to your "Analytics" ribbon, and define "Goals" for the page
Assign a different goal to each target page in this manner
Assign each goal an identical value
With these steps in place, and assuming you have no other tests running, this will produce the result you are looking for.
But the point to all of this is - one needs to fully understand what "Engagement Value" means in the Sitecore CEP, and what it can do for you. It's by long and far more than simply determining the highest conversion rate on any one component.
There are tools out there more tailored to the exact scenario you are looking for.
See my answer here: Clarification on Sitecore A/B Testing Results
And the SBOS Accellerators kit: http://marketplace.sitecore.net/en/Modules/SBOS_Accelerators.aspx
It's A/B Testing or Multivariate Testing ultimate target is to achieve a conversion So Create a Goal with a value lets say 10 otherwise values will be always zero. for any combination of component if the goal is achieved then for that combination value will be 10 it's called conversion. if conversion not happened for any combination values will never increase. after a long duration of test results will show the best possible combination.
Note : Total value never cross the max value of that goal i e 10.
Example for a particular combination say 5 times goal reached is Max 10.
for 1 conversion /5 visits, value is 2. 2 conversion /5 visits value is 4.
My team is considering setting up a MediaWiki wiki to share knowledge and manage information within the team.
Two features the team have expressed an interest in having as part of any new system are:
Access restriction for certain pages/items. Occasionally there may be a need to restrict some members of the team from viewing specific pages, perhaps relating to performance management or interview schedules/results. Does MediaWiki cater for this? I know you can lock pages to make them uneditable, but I'm not so sure you can hide them from being viewed altogether?
Document Expiry. It would be great to have an alert telling you that a page has been untouched for say, 12 months, and as a result may be out of date - flagging it for review or deletion to keep wiki content current. Does MediaWiki have this ability, or is there an extension that enables it?
Cheers for any help you can provide.
Dave.
You do have some access restrictions as outlined in Manual:User_rights, there are also various extensions you can add that may give you more flexibility in this regard.. As far as your Document expiry concern goes, it is possible to create a bot that will patrol your wiki and anything that has not been updated in a set amount of time can have a template pre-pended to it. There are also several bots out there that I am aware of such as MiszaBot (I-III) on Wikipedia that will take any post over a certain age and archive it for you. I hope this helps you figure out if this is the way for you to go.
I have a django model for an appointment in a calendar that I am attempting to write a very comprehensive test driver for. The recurring appointment occurs at some point in time and can either run on infinitely or recur for a fixed number of times. The appointment mirrors the functionality available for a Google Calendar appointment (can recur monthly/annually/weekly, every two weeks, every 3 years.)
I'm trying to come up with a unit test that will exhaustively test the basics of this implementation. I am looking for the edge cases that will define the most basic tests.
I have a lot of basic ones, but am looking for suggestions to help identify the most important cases:
1) Create a single appointment
2) Create an appointment that recurs weekly
3) ... recurs monthly
4) recurs every 2 weeks
5) recurs every 2 months
6) recurs annually
Test with final days of months, leap years, and whether it will go crazy when the year has an extra second (this one hit a driver in the zune player).
Test it behaves well when crossing years.
That said, consider whether you are re-testing something that is part of the framework. Testing date logic can get ugly real fast, so you want to draw a line on what is part of your application and what is part of the framework.
Don't forget to test annual recurrence for Feb 29 on a leap year ;)
Before you start rattling off scenarios, you really need to come up with a test plan based on your understanding of the Requirements.
Consider your user base and any other possible/future user bases (as a lower priority). What will they mostly be using it for and how much are those use cases worth to them in their business?
Ideally, create a model of the app and start from there.
Come up with a Risk Analysis of what you plan on doing. Then plan to do functional, security, localization testing, etc.
Then you can start thinking about scenarios based on how "risky" they are (from the Risk Analysis). Focus on writing and executing the "riskier" ones first.
Get Business input (signoff if possible) on your analysis of risk and how the intend to use it.
Just throwing random scenarios out there is not a good test practice and deserves all the ridicule you can get from developers. Testing should be a more engineered, planned exercise. They can hire anyone off the street to run scenarios that come to the top of their head.
That being said, I agree that the previously mentioned scenarios are tried and true. Good ideas. Also throw in Daylight Savings testing. Use different Email clients. Try publishing free/busy date. Get the developers to explain how they are publishing this information. Is it through a web service? Do they expect only Exchange users to use this? Anyone in different countries where dates are formatted differently? You can then find weaknesses and find more bugs.
Happy Testing.