Intelligent Solutions

Pricing Strategy – we need your help

Simon Woodhead

Simon Woodhead

24th July 2019

By Simon Woodhead

We wrote recently how, for customers and prospective customers, comparing wholesale prices without actual traffic or retrospective CDR analysis was muppetry. Some of you have asked though, how do we price?

Firstly, what we don’t do is manually edit a spreadsheet every month when we find we’re losing money, whilst cost reductions are pocketed. Nor, at the other end of the scale, do we have per customer rate decks targeting a top 10 of fictitious volume and rates. Neither approach seems fair or sensible.

You’ll notice that our rate updates don’t just feature changes to one or two destinations, but rather the entire rate table is regenerated. Further, we don’t do deals. They’re hard to manage from both respects, and giving one flighty prospect a discount today, but not to the loyal customer who’s still going to be loyal tomorrow, seems unfair and somewhat predatory. We have one rate-deck per service level, and customers can choose which service level their account is on. That seems fair all round.

Given many customers mark our rates up directly, even though retail margins have grown exponentially over recent years (another rant!), where those prices come from and how they vary is quite significant. Any small change at the wholesale level gets amplified, making the retail rates look expensive, or, heaven forbid, you making too much margin.

It is here that I think we’re really different. We’ve always tried to make this as automated as it can be; that is the only way updating the entire A-Z every rate update works, but it is hard, and has evolved lots over the years.

We used to base rates purely on actual recent cost, but chugging through even modest volumes those years ago was hard. It was also backwards looking. We could look also at the routes in place but this is really problematic as the cheapest routes aren’t necessarily the ones which work, or where we’ll send traffic, and it could be traffic actually goes to the 6th cheapest route for example. Furthermore, the presumed code used to look up that cost may differ to actual traffic, causing arbitrage vulnerabilities. We also pegged to competitor rate decks as well, but when they were rarely updated and often had errors, we passed those on by price matching. We built a system which pulled all this together and proposed a rate change but a human had to go through manually approving them. This took all day and, I remember well, was so tedious the colleague concerned often did it looking out the window whilst clicking.

Then something odd happened. We’re going back many years but all of a sudden, international carriers began to court us, and our cost-base improved a lot. Our volume also centralised on a few routes that were reliable. This meant we could dramatically simplify how we priced by doing largely what our customers do – peg a model per service level to one of these big carriers as a proxy for cost in our cost+ models. That has served us well since, but is far from perfect, especially now when we only send them traffic as a last resort and have 65 interconnects with other operators on all manner of terms, from mutual forgiveness to involuntary extortion. Our rates are cost+ still, but where the cost is diverging from reality and that doesn’t feel right.

So, the times are a-changing again. If you’ve noticed our rate sheets over the last couple of iterations, you’ll see change. We’ve tweaked the model we apply to prices slightly (favourably for you) but, moreover, we’ve addressed the question of ‘what is our cost’ that we’re applying the model to. We’ve got the compute nowadays to perform deep statistical analysis on vast histories of data and we’ve been making it work hard. We’re also toying with machine learning and even some AI models. The goal is pricing that is more stable, but at the same time learns, and that feels exciting.

It is a journey we’re starting, we haven’t arrived yet. To get there, we need your help. Others are not as transparent as us, and even where rates aren’t hidden behind agreements, they’re not all equal. It is shocking how often smaller prospects we talk to have dramatically better rates than bigger contemporaries for reasons unknown – so it follows they don’t want you knowing that. We don’t know what we don’t know, and we can’t measure traffic we’re not seeing. Therefore, where you use other carriers, we’d love to know what they’re doing right for you pricing wise. Send us rate decks, CDRs, whatever. We’re not going to price match or do anything short-termist, but we will be refining our models to reflect what you teach us.

Who knew pricing could be so much fun!

Related posts