Dear Telkom,

This was very nearly a public break-up between us.

It wasn’t me, you see; it was you. It took 12 working days and immense perseverance from me to get you to fix my dead phone line and practically useless DSL connection.

This is despite those two strands of copper netting you in excess of R40 000 in line rental fees over the past five or so years (that’s what I know about for this line, a tiny portion of which predates me).

In the end, my nagging paid off and the third technician to be scheduled on this fault solved the problem.

He was a proper veteran … been at it for 33 years he told me, in between lamenting the quality and application of his contractor colleagues.

I may not have been so lucky. Many aren’t. A casual glance at your @HelloTelkom (service/support) account on Twitter shows customers without a connection for months on end.

And so, we remain joined at the blue box outside the complex gate for a while. Next time, I’m not convinced I’ll be willing to invest so much time and effort to resolve your problem. Besides, I have this nagging feeling that fibre — something younger, faster, fitter and better than you in every way — will end up coming between us…

Yours (for now), Hilton

This experience over much of the past month (without going into much more gory detail) has laid bare Telkom’s operational, service and network challenges almost entirely. For a start, years of underinvestment in its access network has finally started to catch up to Telkom.

Its core network (backbone) has benefitted from deliberate significant investment in fibre backhaul and the next-generation network over the past five-plus years. It needed to upgrade the core not only to ensure that it was capable of dealing with the significant increases in bandwith utilisation by customers (direct, and indirect — via links it has resold to other operators), but also get it ready for the roll-out of fibre connectivity directly to homes and businesses.

In its 2016 financial year (April 2015 to March 2016), traffic volumes on its fixed network soared by nearly 40%, while the number of broadband subscribers increased by 2%. This means that on a per subscriber basis, broadband traffic volumes were up by 36% (from 338GB to 459GB). Its core and backhaul network is (largely) coping.

But the so-called “last mile” is a mess. And it’s arguably worse in residential complexes, where much of that infrastructure is shared and was installed a decade or more ago.

The inherent problem with DSL is that when it works, it works … and when it doesn’t, it doesn’t. There is no in between (and you’d expect this, given that connectivity is provided by a pair of copper wires).

Telkom’s Achilles’ heel is this last-mile access network, which is suffering from underinvestment (and the related inconsistent resolution of faults, but we’ll get to that).

The group’s capital expenditure on network “rehabilitation/sustainment” totalled R1,1bn in the 2015 and 2016 financial years, about the same as it invested in fibre-to-the-home (FTTH) rollout. But the pendulum has swung. By the first half of 2017, the FTTH capex spend was five times that invested on “rehabilitation/sustainment”!

This perfectly illustrates Telkom’s conundrum: for every R1m that it ought to spend on its legacy copper network, how much more value could it derive from investing that same amount in its next-generation fibre network? It’s obviously not quite as simplistic, but this is the crux of Telkom’s dilemma: how much should it be investing on a network that will effectively be obsolete in five years’ time?

Of course, customers — particularly those one million DSL subscribers — are stuck in the middle. Faults are creeping in a lot more frequently, if anecdotal evidence is to be believed (Telkom no longer discloses this data).

And this is where the chaos starts. Telkom, which faces the customer, is no longer in control of its network. This has become the problem of its separated wholesale division, Openserve. Theoretically, this is good for customers and for competition.

But, operationally, Telkom is now reliant on an amorphous service provider to resolve faults on its network. Not that only Openserve itself bears this responsibility! As part of Telkom’s much heralded workforce reduction, many of its field service technicians (those in the Telkom bakkies) are now independent contractors.

Huawei (and its contractors) is also involved to some extent as an outsourced service provider. Given this transition, there are bound to still be some misaligned incentives. Some technicians are employees, others contractors and these two groups are incentivised differently (contractors, for example, are paid for each fault resolved, not attended to).

For customers, this becomes potluck: have a good contractor arrive and chances are your fault will be fixed speedily. Otherwise, you wait.

Field service is hard. And Telkom admits that its speed in addressing faults has “not been what we want it to be”. Part of this lies in the contractor vs staffer (and Openserve) problem, the other in the systems and processes that manage and schedule these skills. Telkom says it will be “rolling out the next phase” of “new workforce management solutions … in the new year”.

Yes, it’s needed time to get the business into shape to compete. The major part of the overhaul of its business — and operations — support systems (tying together dozens of disparate systems so that it has a single view of its customers) is practically complete. For one, it finally has an automated fault reporting/tracking system that works (after years of trying).

But, the migration of customers from DSL to fibre is not progressing as quickly as Telkom might have hoped. As at 30 September, it reported just shy of 19 000 FTTH connections (versus a million active DSL lines).

It’s now time for Telkom to really double-down on fibre. It simply cannot afford to sustain the costs and frustrations of those customers on its copper network for another two years. It might not have much of a business left, given the aggressive competition its facing in the FTTH market.