By Simon Woodhead
A number of you have asked what has been happening on the Simwood network throughout the COVID-19 pandemic. A lot of this is simple interest and benchmarking their own business, and in other cases it is because with our breadth of wholesale customers and the diversity of [generally] businesses they represent, we’re a good proxy for the activity of UK PLC.
Others have commented on ‘internet’ traffic and are probably better placed to do so. We have seen similar trends of the peak not shifting from the evening, an increase in day-time traffic, marginally so when working from home increased, more pronounced when the schools closed. ‘Nothing to see here’ is the main message – networks just dealt with the extra load at certain times of day as they’re built to handle a peak which didn’t move or change.
Voice calls is where I think we can add some colour because our sample size there is more statistically significant.
Anecdotally, I’ve been saying things began to slow after the lockdown. Each day the economy flagging as people seemed to be enjoying the paid holiday! Then reality seemed to set in and we entered a phase akin to early January every year – the party is over, let’s crack on! And now we’re entering the new normal where working from home is routine, and collectively we’re getting better at it! Underlying all of those things we’ve seen quite a pronounced rotation between customers reflective of their particular niche – some completely furloughed along with their hospitality customers, others doing many times the volume as they ride a working-from-home wave.
But what do the numbers show? Below we’ve taken each day of the year (by week and day to eliminate days falling at different times) and contrasted 2020 with 2019. Rather than educate our peers on our absolute volumes, or indeed distorting figures with seasonal affects, we’ve instead plotted the percentage change in the figures from one year to the next. We think this gives us the best means of comparing like with like, i.e. how has our growth changed. Of course, our growth rate differs drastically to the industry and economy, but we suggest that other things being equal, the change in it doesn’t. We tried plotting absolutes, and they show the same picture, but smoothed by our high growth rate, so lets run with growth.
If we first look at total calls across the network, and overlay some events, you can see a very clear trend. The dotted line is the day-by-day growth rate, the solid line is a moving average and thus lags. The black dots represent the following newsworthy events in the UK:
- A request to curtail non-essential travel
- The economic measures to assist with COVID-19 impact
- UK schools closed
- UK hospitality closed
- UK lockdown
In each case, these events have generally occurred in the 5pm Number 10 briefing so their impact can be seen from the following day.
That’s a pretty clear trend there I think! We were going gangbusters right up until the request to curtail non-essential travel and the subsequent two days until schools were closed. Those three days saw a halving of our growth rate from +50% to +25% (or thereabouts). You can see this quite clearly in the (remember: lagged) moving average. The outlier is the day the lockdown was announced where we had a record day in absolute terms, +64% on 2019. The lockdown was anticipated in the news then, and arguably overdue (some of us had already closed offices the week before), so was this flurry people preparing and clearing the decks?
The other trends I anecdotally described above are also evident here – a sharp slowing, before levelling off. In our case that is levelling off slightly down on 2019, but given we were pushing +50% before, what does this mean for the economy?
In absolute terms peak to trough is a reduction of 45% in traffic, but both were exceptional. If we contrast totals for week 13 and compare to week 10, the decline is more like 36%. Maybe those economic reports of -35% activity weren’t too far off?
We’ll dive a bit deeper shortly, but there’s another trend in the aggregate figures we think is noteworthy. We monitor the average call duration because across a universe of calls it is indicative of quality. Now, quality being equal perhaps it indicates something else.
Isn’t that interesting? We’d need deeper analysis (i.e. at the per customer level) to understand this fully. It could simply be the rotation we talked about between customer profiles, but I’d really like to think this is demonstrating the prevalence of humanity and compassion we’ve seen since this crisis began. Whilst I imagine the usual douche-bags are selling COVID-19 cures door-to-door, and posting sales figures on LinkedIn, the rest of us seem to be being nicer to each other and caring more. That is one trend I really hope will stick.
So call volumes are -35% but call duration is +35%; not much to see here after all perhaps?
Well, let’s dive even deeper. Simwood is a licensed carrier in 13 USA States and that is a rapidly growing side of our business. With a slightly different timescale to eliminate distracting factors the picture looks less severe.
The two data points are actually three (the third occurring over the weekend preceding the second whilst weekends are excluded on the charts):
- The State Department suspended routing visa services at all embassies and consulates worldwide
- Travel bans for foreign nationals who had visited Europe in the past 14 days
- NY, CA and other large states ordered many businesses to close and people to stay at home
It’d be easy to conclude that the US doesn’t show much of a slow-down but note the scale on this chart. We were growing +200% coming into this, peaking above +300%, before retracing to +200%. If we compare week 10 (the volume highs) to week 12 (immediately after the lockdown) it shows a -20% swing in absolute volume. This is far from insignificant but we would exercise caution in comparing it directly to the UK’s reduction for two major reasons:
- The US is a huge place and the virus is phasing across it rather than hitting the entire country simultaneously. It is more like Europe in that respect.
- We have fewer customers in the US than the UK so this is more sector focussed.
What is interesting though, is that we’ve seen little in the way of rotation between customers in the US. This means the dramatic increase in average call duration cannot be attributed to rotation. Maybe that human factor is at play after all?
Coming back to the UK, let’s look at that through a few lenses.
A similar picture to before, but now with the US (and RoW) removed, the downturn is more pronounced. Our record day remains the same but the decline after is greater. With the softening effect of the US removed and ignoring the week 11 peak, the UK in isolation shows a 38% decline week 10 to week 13 in absolute terms. Over the same period, call duration is +32%, for a net decline in total call minutes of 18.5%.
And what about mobile? You’d expect those working from home who don’t benefit from the kind of technologies our customers provide, would default to mobile. Many are mobile-first regardless of the UC they’re given, so may not change behaviour at all.
It follows the same general pattern in volumes. Note though that there’s a more pronounced uptick in call duration on mobile, although the average duration itself remains only 75% of the UK as a whole, reflecting the quality differential probably. In absolute terms (week 10 vs week 13) this traffic is -30% but with a +51% increase in average duration, duration for a net increase of minutes passed amounting to 6%.
The picture with non-mobile is remarkably similar, the y-axis being slightly stretched due to the lower peak, and indeed the presence of earlier actual traffic peaks. Absolute volumes (week 10 to week 13) are -35% but duration is 30% higher, for a net decline in minutes passed of 16%.
If call volume across a universe of customers can be an indicator of economic activity, we appear to be about 38% down for the UK, 20% down for the US, in the period since the last business-as-usual period, and the first lockdown period in respective countries.
Over the same period though, call durations have shot up, 32% in the UK and 56% in the US. Whilst we started out hoping that increase was due to the increased humanity and compassion we’d observed, its presence across two regions with very different customer profiles, and indeed the absence of any change in mean-customer profile in one of those regions, leads us to now struggle to identify another factor that could be responsible! And that is worthy of celebrating and hanging on to as we all get back on our feet!
Lastly, these figures belie another major trend: video. Since we spun up https://meet.simwood.com we have found ourselves using video far far more than before. We had used video-conferencing in the past where people couldn’t be present, now we’re using it by default and we see a lot of usage of it from rather surprising places – like a major Chicago University using it to teach classes for one example. It not only enables things to continue but offers an antidote to isolation both professionally and personally. We know colleagues have had family gatherings, and hung out with friends purely socially on Simwood Meet, and we’ve all heard the headlines about Zoom usage and issues. We bet that if there was an easy way of metering video usage, we’d see a significant increase in minutes of communication overall, and that is a good thing.
What we have no clue about is whether we’ve hit the trough and will slowly recover, or have further to fall in call volumes and (assuming the premise) economic activity. If we do start to recover, we also don’t know to what extent video will stick as an acceptable means of attendance, or whether we’ll again be choosing between travelling hours, apologising for not properly attending, or resigning and spending the valuable time with our families we’ve come to enjoy.