QuestionPoint Qwidget Roll up Pilot

the QuestionPoint chat widget (qwidget)The Qwidget is one of the customer interfaces that we can use at AskUsNow! (as seen on this test page). Because of the qwidget coding, customers entering AskUsNow! through the qwidget would never be "rolled up" to the back up librarians or the rest of the cooperative, making this particular interface NOT accessible 24/7. Because customers entering this interface would experience an "offline" time even though our service is indeed 24/7 we didn't deploy this interface.

Montgomery County Public Library currently is the only AskUsNow! institution to use the Qwidget. They are able to use the qwidget as a local IM-type service, above and beyond their normal AskUsNow! hours, because they purchased their own queue (available through QuestionPoint, separate from AskUsNow!'s budget at $1000/yearly) MCPL was able to funnel MCPL customers in through the qwidget interface during certain hours without the rest of the AskUsNow! monitoring librarians ever seeing those customers in queue.

The qwidget interface was introduced in March 2008 with the intent that it would NOT roll up to the rest of the cooperative until qwidget usage was better understood; QuestionPoint didn't want to bombard the cooperative unexpectedly. Now, however, a pilot project has started to bring SOME institutions already using the Qwidget into the cooperative so that customers entering that interface would get the same 24/7 service they would if they came in through the regular page-push interface. They are doing this slow to make sure the cooperative and software isn't overwhelmed.

The Qwidget Roll-Up Pilot

Quoted directly from QuestionPoint:

"The QuestionPoint Qwidget was introduced in March 2008, and its popularity with users is apparent in the steady growth of Qwidget traffic. Your Qwidget is available (open to users) when libraries associated with your Primary Queue are monitoring. Qwidget patrons are seen only by your Primary Queue librarians. However as of January 2010 we can now configure individual Qwidgets to roll up to the Cooperative: the “rolled up” Qwidget patrons will be seen by the entire Cooperative, and the Qwidget itself will remain open so long as anyone in the Cooperative is online (basically, 24/7).

There are two potential impacts to Qwidgets rolling up to the Cooperative:

  1. Server response time: Once the pilot starts, the software will have to perform an extra check to see if a given Qwidget patron is eligible to be seen by the Cooperative.
  2. Cooperative traffic: Since the Qwidgets are so popular with users, we anticipate that traffic in the Cooperative will increase when Qwidgets are rolled up.

Possible impacts include: (1) longer wait times, (2) more patron traffic during “after hours” (evenings and weekends) when Cooperative librarians are not online, causing more of a load on Backup Staff, and (3) increases in Quality complaints (due to librarians handling more sessions simultaneously and longer patron waits).

In order to measure the impact of Qwidget rollup, we will begin a pilot on February 15. Only libraries/groups with an Answering Percentage of over 75% (based on your Primary Queue) will be eligible to participate in the pilot. The pilot will last up to 6 months, but will end immediately if server response time is impacted negatively, or if Cooperative traffic increases beyond the capacity of the service (based on wait times, backup staff impact and quality complaints). If the Answering Percentage of any pilot participant falls below 75%, the pilot will end for that library/group. I will be contacting eligible pilot participants this week with more details. Pilot libraries will not all roll up on February 15, instead we will stagger the pilot participants for a gradual roll out. Once the pilot concludes, we will determine if Qwidget Roll Up is something we can offer on an ongoing basis and, if so, whether an Answering Percentage of 75% is the appropriate level of staffing required.

Libraries participating in the pilot will be monitored closely to make sure their Answering Percentage stays above 75% and that wait times for their queues remain manageable (i.e., average wait not to exceed 4 minutes in any given hour).  Pilot libraries that do not meet these requirements will be dropped from the pilot.  In addition, the pilot may end for everyone if:  (1) server response time is negatively impacted; (2) Back Up staff are stretched too thin during evenings and weekends; (3) Quality complaints rise; (4) overall Cooperative Answering Percentage drops lower than it is now (it is now 62%); or if there are  other unforeseen consequences which negatively impact the Cooperative."

The Pilot has already started with the following institutions (as of 3/8):

  1. University of Arkansas

  2. Central Rappahannock Public Library

  3. California Community College group (or “CA Com Coll”)

  4. NY AskUs247 – both public and academic libraries in Central and Western New York State

  5. University of Hawaii

  6. Washington Academic

  7. Spokane Falls Community College

  8. University of Washington

How does this affect chatting?

It doesn't. If you are monitoring globally, you will now start to see customers entering the service through the Qwidget (it will say "Qwidget" before the customer's question so you know what interface they are using) but if a customer appears in your Chat Monitor, you should pick them up as usual. This applies regardless of whether the patron has come in through a Qwidget or through the usual chat form.  If a patron is in your Chat Monitor, pick him or her up!

The only thing to be aware of is that you can't say things like "did the page show up on the right?" when the customer is in the qwidget interface because they won't have the page push! (They will get links that will open a new window when clicked).

Please contact Julie with any questions about this.

Comments

Update: week five

Julie's picture

From Susan McGlamery yesterday:

We have just completed week 5 of the Qwidget Roll Up pilot, and I wanted to give you an update:

We are monitoring several elements on a weekly basis:

  • Server response time
  • Answering Percentage (AP) of the pilot libraries, on a weekly basis
  • Cooperative AP, academic and public
  • Cooperative wait times

Just to recap, all groups with an AP of over 75% were asked if they wanted to participate in the Qwidget roll up pilot.  Participation is voluntary, however if a group agrees to be included in the pilot, then that group must also agree to keep their AP over 75% each week.  If a group has an AP of 75% or below for 3 weeks in a row, then they are dropped from the pilot.

There were 20 groups (based on queues) eligible for the pilot. Of those, 12 accepted the terms of the pilot.  Of the 12, one was dropped due to a low AP for 3 weeks in a row.

The current pilot participants (listed by queue) are:

Arlington Heights Memorial Library (IL)
AskUs247 public (NY)
AskUS247 academic (NY)
Central Rappahannock PL (VA)
QandANJ public group (NJ)
Spokane Falls Community College (WA)
Timberland PL (WA)
University of Arkansas
University of Hawaii
University of Washington
Washington Academic

Each of the above groups (queues) have kept their AP above 75%. In some cases, a group’s AP has increased during the pilot.  The AP of the pilot group as a whole, calculated from the start of the pilot, is 87% (for last week, it was 82%).  Server response time is not impacted significantly as a result of the pilot.  The average server response time during the week before the pilot was 0.058.  The server response time during the latest week of the pilot (week 5, the week beginning March 22) was 0.081, which is not a significant difference.   

Most pilot participants have seen an increase in their Qwidget traffic during the pilot, but not all have.  In some cases this may be due to spring break (for the academics).  Last week (week 5 of the pilot), 648 Qwidget sessions rolled up to the Cooperative. Of those, the Cooperative picked up 456, the Back Up staff picked up 122, and 71 Qwidget patrons were “abandoned” (they logged out before a librarian could pick them up).

The wait time in the Cooperative has stayed relatively stable during the pilot: last week, average wait time in the Public Coop is 84 seconds, average wait time in the Academic Coop is 72 seconds.  Before the pilot (month of Feb 2010), average wait time for the Public Coop was 85 seconds and for the Academic Coop was 71 seconds.

We will continue to monitor the Qwidget roll up pilot through April, and provide an assessment of the pilot in May.

Please let me know if you have any questions regarding the Qwidget Roll Up Pilot, and thanks for all your help with the Qwidget roll up pilot!

Regards,

Susan McGlamery
QuestionPoint senior product manager
OCLC