How do you analyse webinar success? Some thoughts…

(Authors note: This is a first set of thoughts, noted reasonably quickly, whilst they are fresh in my mind. I am interested to know what people think and what approaches people take to analysing their own webinars.)

Today was a good day. At lunchtime, I co-hosted a webinar entitled “Managing and Analysing Data to Understand Learning Impact”. My co-host was Stephanie Stretton of Rosetta Stone, a smart, interesting and thoughtful coworker and an excellent addition to any network. The show was managed by the LPI who do seem to be able to gather an interesting and eager audience. This was not different, they contributed fully and kept me on my toes too. Michael Strawbridge was our capable and attentive ringleader, as usual.

In the 45 minutes, as they flew past us, we covered some fundamental themes of digital: ever heightened expectations of a good user experience; the key to personalisation being personal relevance and the idea of signals of value being more useful than the quest for proof. (I have posted on that last point before). 

My perception is that it went well. It was certainly enjoyable, the participants were active and meeting new people with different ideas and other views is always gratifying. As I sat on the tube to my following meeting, I started to wonder though, in the spirit of data analysis, what should I look at to test my hypothesis that it went well? What are the data signals of value?

So, here are some thoughts on the kinds of signals I think might make for a good analysis of the event. I realise that there are many webinar and VILT specialists out there and I am really keen to know what you look at to assess how your events work. I also realise that this data may be standard from platform providers (is it?), so this might not be new – working out loud and all that. 

I have divided them up into some categories that seem relevant as I type. There are too many here but this is something of a brainstorm for blogging, please bear with me.

Participant engagement (I don’t like that word but it seems to have stuck):

  • % of finishers (those who stayed to the end)
    • And drop off rate for those whose glass is half empty
  • Number of commenters
  • % of commenters per participant (comment reach)
  • Comments per participant (comment depth)
  • Comment length – single words versus longer form
  • Number of repeat commenters – the engines or domineers?
  • % of repeat commenters from total commenters (comment concentration?)
  • Number of peer to peer comments amongst participants (conversation strength?)
  • Some kind of sentiment analysis might be interesting for a large audience too or for certain topics?
  • We didn’t use audio so measures of ‘speaking’ versus commenting would not have helped but seems like a good metric to me, if you do use it
  • And finally…if you disable commenting and speaking how do you know what is going on?


  • Some kind of simple rating would be good at the close: how useful was this webinar on a 1,2,3 scale, maybe; did you get what you came for etc.
  • I do like a Net Promoter Score but a survey might be over-egging it

Commercial measures (these are invented by me so not a reflection of partner needs):

  • Cost per participant (efficiency?)
  • Lead generation:
    • Reach to new contacts
    • Social follows etc, form attendees
    • Social mentions from attendees – new and existing
  • Conversion of attendees to other events
  • Conversion of attendees from other events
  • % of membership or customer base attending (internal reach)

Marketing measures

  • Number of registrations
  • Sources of registration (email, socials, direct etc.)
  • Conversion rate: % of reached audience who registered
  • Activation: % of registrants who attended
  • Profile of participants – not sure about the data here and what agreements are in place re data usage
  • Downloads of webinar recording
  • Requests for slides/content
  • Contact from attendees

A closing point: any of these metrics will need a comparison point to understand them properly, preferably trend data. So data will need to be gathered and analysed for more than one event to get a sense of what good might look like. 

So, what do you think? What have I missed? Have I missed the point? There is too much here, for sure, so what seems most useful to you?

One comment

Leave a Reply

%d bloggers like this: