A place to ask questions, connect with others, and stay in the know
We are in the process of implementing the Genesys Cloud and while working through it, I had a few questions that I was hoping someone that is currently using maybe be able to help me. (I should mention we are a BOP and we forecast/schedule for multiple clients, that are all slightly different)
Thanks in advance for any help.
I'll let @Daniel Chapdelaine on the PM side address most of your forecasting questions.
For #6, there are several import options for forecast: 1. resultant forecast (you create a final forecast using whatever data and methodologies and import the final one into Genesys Cloud); 2. import up to 6 weeks of source data for the weighted historical index method (WHI); 3. import up to 6 days per source day in WHI; 4. utilize historical data import (general use here is bootstrapping until you have enough native data).
For #7: Automatic Best Method Selection support forecasts up to 6 weeks (starting up to 26 weeks in the future) and a long-term interval up to 104 weeks.
General comment on manually choosing dates to include/exclude, seasonality (daily, monthly, yearly), smoothing, outlier detection, etc.: ABM method offers built-in capabilities and should be the default choice when wanting to create a forecast. https://help.mypurecloud.com/articles/automatic-best-method-forecast-method-overview/ . Some items like smoothing are really feel good measures that typically result in significantly worse forecast accuracy - note I am differentiating basic smoothing to say natural dampening affect of outliers that occurs naturally/natively with ABM.
For #4, I would recommend utilizing the min/max contiguous work time constraint. This governs the min/max time between activities marked as counting toward work time and those that are not.
------------------------------Jay LangsfordVP, R&D------------------------------
#Forecasting#Scheduling------------------------------Jamie O'BrienDialAmerica Marketing, Inc.------------------------------
Hello Jay, Thank you for answering my questions. For #7, you mentioned that the ABM would be the best method to use for long-term forecast, and based on what I have been able to see thus far, I think I would agree with that. However, being a BOP we often have to supply potential clients with long range forecasts, using their information. As that would not be on our switch, ABM wouldn't be available to use. For those, would you recommend using the Import Short-Term Interval Forecast Data 6 weeks at a time?Thank you again for your time in answering these questions, its been a HUGE help!
------------------------------Jay LangsfordVP, R&DOriginal Message:Sent: 09-29-2023 10:29From: Jamie O'BrienSubject: Forecast/Scheduling Questions
In this situation, importing a 6-week forecast would be the best method. In the future, we plan to allow multiple file loads through the HDI, which might help with these BPO situations, allowing users to quickly bring in new history as they like when they want and use ABM.
I am the product manager of forecasting, and I will answer your questions.
As for questions 6 and 7, Jay did a good job of answering them. Jamie, these are great questions and are all features that are on our roadmap (except 5) being actively worked on, and we are targeting to release some of these features hopefully in the second half of 2024. I hope this answers your questions and if you have any more please feel free to reach out to me.
I'm hoping to tag onto this conversation for question #7 brought up on forecasting. I can see the ability to create a forecast up to 104 weeks out, but I am not seeing that there is the ability to actually generate a schedule based off of that forecast past roughly 6 months from today's date. Is it possible to forecast as well as generate and publish a schedule for an entire year?
the aim is to bring scheduling in line with the forecasting time horizon. This will eventually bring together the forecasting, scheduling and capacity planning elements.
At the moment we don't have dates for when the extended scheduling will be available, we are working on a number of other scheduling-based features with a full roadmap into late 2024.
as we work our way through these we'll be reviewing the roadmap and prioritising new features to begin working on in H2 2024.As you can see from the continuous scheduling idea this is an area that has a lot of interest and we are creating the foundation, with forecasting, to support this.
Thank you Daniel!
for #4 there is an idea in the Ideas Lab for adding the ability to set the time between Work Plan Activities:
Defined time periods between activities in work plans
Please take a look and add your comments and vote, this will help us prioritise future developments and enhancements.
I just wanted to note as we were unaware of this when my company made the switch from PureConnect to Genny Cloud:
Unless you have 1-2+ years worth of data for the AI to recognize your trends/seasonality, the safest bet is to go with Weighted Historical Index forecasting.
Before we 'turned off' PureConnect, I ran some forecasts for a few weeks out to cover us as we would not have any data to refer to for the first few weeks on the new phone system.
Sure you can import data, but my experience with the formatting was not pleasant - so we gave up on it. The formatting of PureConnect's export would take an insane number of hours to reformat to Genny Cloud's import template.
We plan to use Weighted Historical Import for a while so we know which data is being collected.
For example: Mondays, we usually see 2500 interactions come through. For major holidays (Labor Day for example) - we had maybe 250 calls come through. Ideally, we would want to exclude that day and we can do so using Weighted Historical forecasting vs ABM at the moment.
I would say "Unless you have 1-2+ years worth of data for the AI to recognize your trends/seasonality, the safest bet is to go with Weighted Historical Index forecasting.", is not what Genesys would recommend. We have and continue to improve cold-start scenarios as relates to Automatic Best Method. While it is true that more data helps build better models and achieve higher accuracy, even with a few weeks of data we look at trends, level shifts, and seasonality (day of week week of month, etc.).I realize it is just an example, but you could use ABM and simply set a max offered count for the day for the forecast Labor Day. There is no reason to throw out ABM and all of its benefits for a use case that can be handled another way simply. That isn't to say we are not looking at further enhancements around tagging past periods for special handling.
Thank you Jacqueline.
Check out the Genesys Knowledge Network - your all-in-one access point for Genesys resources
Every year, Genesys® orchestrates more than 70 billion remarkable customer experiences for organizations in more than 100 countries. Through the power of our cloud, digital and AI technologies, organizations can realize Experience as a Service℠, our vision for empathetic customer experiences at scale. With Genesys, organizations have the power to deliver proactive, predictive, and hyper personalized experiences to deepen their customer connection across every marketing, sales, and service moment on any channel, while also improving employee productivity and engagement. By transforming back-office technology to a modern revenue velocity engine Genesys enables true intimacy at scale to foster customer trust and loyalty.