External variables
Channel credit
We should check if there's enough credit in the channel to run the survey.
There are many ways to estimate this
Over time we tend to think that the safest way is assuming a worst case scenario in which the survey contacts every respondent in the sample
For SMS
Check the number of initial SMSs (the SMSs that are included in the first message, usually language selection) and the number of retries. (i.e. if the survey has 4 initial SMSs and 3 retries, then the number is 12)
Multiply that number by the sample size, that will give you a good estimate of the costs. The other costs, like the once involved in answering the survey are usually trivial compared with these big numbers
For IVR
Check what the channel does with not connected calls (most of them doesn't charge anything, but unfortunately that's not always the same) if there's no cost for the not connected calls a good way to estimate is getting the target multiplied by the number of minutes it takes to complete the survey. (i.e. if you need to get 300 completes and it takes 10 minutes to complete the survey you'll need 3000 minutes of calls, over that number you should add 20% for the respondents that start the survey but doesn't complete it) It is true that if you consider partials as complete it might lower the number a bit, but it is safer to be a bit pessimistic.
Channel velocity
Is important to know how many contact attempts the channel is able to process per unit of time to estimate some variables needed in Surveda.
Channel velocity can be estimated by creating a survey with a narrow contact window (i.e 1 hour per day), pre-tests are ideal for this purpose.
Once you get the number of calls that can be done in one hour you can estimate channel velocity.
This will help you to set uo the channel limit in verboice and the channel capacity in surveda.
Internal variables
Channel capacity
In the channels section you can define a channel capacity, this will define the size of the batch that each survey will send to the channel per round. This value should be estimated based on channel velocity.
On standard Surveda installations it polls every minute, then it is fair to assume that the channel broker should exhaust a batch every minute. Based on this assumption the calculation for the channel capacity can be.
If an IVR channel is able to make 1k call attempts per hour and it takes 2 minutes to fail a non-answered call then the channel limit should be around 30.
(Number of calls done in 1 hour / (60 * maximum delay to complete a call)) (1000 / (60 *2)) = 8.33
If an SMS channel is able to send 1k MT (AO) then the calculation will be
1000/60 = 16.66
Therefore the channel limit should be 16
The critical aspect of the channel capacity is prevent overloading the channel so if you keep the capacity below the real one, there will not be big issues (maybe some slight delay)
For IVR, remember to check the channel’s Limit in Verboice, it should be close to the channel capacity in Surveda.
Initial Success Rate (ISR)
This value is used to calculate the initial batch size and influences in the calculation of the following batches. Success rates varies quite a lot depending on the target population and the questionnaire. It is important to know that if you use small ISR you'll get large batch sizes and vice-versa.
Reasonable ISRs go from 0.02 to 0.2 (2% to 20%)
If you have reliable information on this it is important to use it in the project, it will help the survey broker to work efficiently.
if you don't have success rate information you can use an average of the values defined above ~10%
Batch limit per minute
This variable is crucial when you're running more than one survey at a time, we should keep this under 500 to prevent getting Surveda stuck in one survey for more than 60 minutes. So if you are running surveys in parallel keep this value under 500.
Next step
https://mobilesurveys.freshdesk.com/support/solutions/articles/19000046976-checklist-create-a-survey