VK3ARR's SOTA Blog

Ego loqui ad viros super montes

Month: August, 2017

VK Activator Stats – Part 3

Because it seems I have nothing better to do, and people are interested, I decided to dig a bit deeper into the retention/churn dimension discussed in the last post and continue the analysis started here. At some point I’ll actually get around to updating my blog with tales of my own activations, rather than data on everyone elses 😀

First, let’s define a new activator as someone who has their first activation in a particular year. Let’s then define an old activator as someone who has had their last activation in a particular year. By definition, this is taken at the end of the year, so for 2017, there are no old activators yet.

new_versus_old_activators

All VK new versus old activators

Unsurprisingly, there’s a big increase in new activators early in the SOTA program’s history in VK, and once we hit 2015, this rate slows down dramatically, and is overtaken by people who have, by our definition, ‘left SOTA’.

I acknowledge that a person who last activated in 2016 and hasn’t activated in 2017 yet will be counted as an old activator, but 7 months of inactivity isn’t an unrealistic definition, and so I am comfortable with the statistics as they currently stand.

From this, we can calculate net retention rates. I’ve chosen the new year as an arbitrary cutover date of our ‘subscription’, but while this analysis could be done for any particular date, I’m pretty happy the results will not dramatically change the analysis.

To calculate retention, we take the total number of activators last year, then calculate the total number of activators at the end of the next year. This is last year’s figure, plus the new activators this year, minus the old activators. We compare the ratio of this year to last year to get our retention rate.

retention_by_year

Calculated retention rate over all VK

As can be seen, a steady decline, with a little bit of noise in the data. This one is probably the more concerning one for me given the long term trend. No analysis has been performed on why, but the original post speculates.

I was then asked by Grant VK4JAZ about splitting by association. This isn’t easy as people can change association, but based on their stated “Home Association” in the database, the stats are as follows:

new_activators_by_home_assoc

Activators by year and home association for date of first activation

old_activators_by_association

Activators by home assocation and year based on date of last activation

From these, and using the same methodology above, we can track, by VK association, the total number of active activators:

activator_count_by_region_year

Total activator count – calculated as total activators last year, plus new activators this year minus old activators last year

This suggests that, apart from VK7, all associations have seen decreases in activators and activity since each association kicked off. It remains to be seen if this could be called a plateau, or if it’s a genuine decrease. The retention rate figures suggest its a decrease.

Advertisements

VK Activator Statistics – Part 2

Some further statistics (again just for VK activators) – firstly, time since last activation:

time_since_last_activation

Time since last activation (in months)

This to me is a measure of how many ‘orphaned’ activators we have out there. Of all activators that have ever activated, over a third (36.5%) have activated something in the last 6 months. About the same percentage haven’t activated anything in over 2 years.

Some of those would have tried SOTA and not liked it, some may have been brought along for an activation and never got around to doing another. Some may have been active but fallen out of favour. I’m not aware of any SKs in the VK population at this point (touch wood).

Interestingly, there are spikes again, that seem to correspond to spikes in the First Activations graph from last post. This reinforces, to me at least, that these are likely of the “tried SOTA on the basis of an AR article, never did another one” flavour.

Next, we can examine the longevity of activators – how long between their first and last activations:

time_between_first_and_last_activations

Time between first and last activations in months

Five activators have stayed activating over a period of 60 months (5 years), and the reverse of that is 84 people (32%) have only ever been active for 1 month (likely a single activation).

Over half of activators have been active for a period of less than a year. Note that this is not the same as saying people are unengaged – an activator that only started 6 months ago but is still racking up activations will be in this cohort, but given roughly half of the cohort are in the 0 months bucket, it’s fair to assume that most in the cohort are no longer active in SOTA, and fit into the third of people from above that haven’t activated anything in over 2 years.

In the language of business (and customer engagement) this is called churn. People who buy a product or subscription then don’t renew their access to it. For a business looking to grow, churn is a problem because it means that any churned customer has to be replaced with net new business – and, on top of that, if they churned, you probably don’t have a promoter of your product out there!

The immediate question any business has to ask when confronted with churn is not, what do we do about it, but what is our expected, natural, level of churn. 90% stay with us? 50%? 5%? It’s never going to be no churn. Once you establish what your natural level is, then targets can be set, and we understand how many new customers we’d need to bring in to continue to grow.

Stepping out of the business world and back into SOTA, our current growth rate seems to be about 12-15 net new activators a year (from the last post). Our current retention rate (the inverse of churn) is at least 42% (107 active in last 12 months of 249 ever active). Note the use of the word “at least”, because that definition of retention is not strictly accurate. Our growth rate on 107 activators is therefore a bit over 10%.

So, in my opinion, if we want to continue to keep activity levels up, we need to concentrate on why people do one activation and no more, and, the question then is, how do we encourage those folks to stay active in SOTA, and, for those who have already activated, what would it take to get back into SOTA?

(Disclaimer: views are mine, not the MT’s, but I think you knew that)

VK SOTA Activator statistics

An interesting discussion on SOTA_Australia about perceived declining SOTA activity prompted this question from Paul VK3HN, and my response below

The correlation between Yahoo group messages and SOTA related activity can obviously be challenged — it could be the popularity of this group has waned but the number of activations has not. A more telling analysis would be to report on SOTAWatch alerts by VKs per month over the previous 5 years. Anyone care to do that?

Straight from the horse’s database, we have activation counts grouped by month (where callsign starts with VK or AX):

activations_by_month

Number of VK activations each month

and as a measure of how many new people are coming in, this is the monthly count of activators for whom this is their first activation in the database (again, where callsign starts with VK or AX):

activators_by_first_activation

First time activators

The first definitely shows a bell curve, complete with peaks at the New Year UTC rollover point. As people say, this is most likely correlated to sunspot cycle, but also I imagine that the early adopters aren’t getting out as much. Personally, I feel like I’ve hardly done any activating this year – most of my SOTA this year has been IT related!

The second graph is more important and that shows that the rate of new activators joining is only slightly down on the overall trend, with a few peaks probably associated with AR articles or the like. Interestingly, October last year was the only month where there wasn’t at least one new activator in VK. That’s suggests there’s still life in the old nag yet.

Equivalent chaser graphs could be generated of course, and I’ve not done detailed scrutiny on the veracity of the numbers, but a quick check seems sane.