Good & bad metrics/statistics to watch at MWC

"There are three kinds of lies: Lies, damned lies, and statistics" (attributed to Mark Twain, Benjamin Disraeli, or someone anonymous lost to the mists of time).

Few periods in the telecoms year have more of all three types of untruth than during MWC and the run-up to it. Here, I want to focus on the third class - stats, metrics and other quantified - but misleading - data points.

I've been writing about the need for the telecoms industry to switch to "good metrics, rather than easy metrics" for several years. I've written a full report and many posts (also here and here) on this theme, and also advised regulators and policymakers on this domain.

The next two weeks will be full of "The Easy, The Bad and The Ugly", to coin a phrase.

I'm not even going to delve into the snowstorm of disinformation that comes from commissioned surveys, either of consumers, enterprises or industry insiders. But an axiom to use as a baseline is: "90% of respondents didn't understand the question, and the other 10% lied".

It's the other data-points I want to focus. Market values and volumes, adoption rates, user numbers, transaction data, investment rates, spectrum bandwidths and more.

While there are certainly some good high-quality numbers generated by national regulators and many analyst firms and crowdsourced data providers, these often don't survive the attention of lobby firms, industry groups and PR professionals.

• "Context? Never heard of it"
• "Stacked references to other dodgy data we've published in the past? Who checks?"
• "We got an economic consultancy to generate forecasts of this complex deep-tech sector"
• "Methodology and definitions? Don't bother me with trivia, just publish the chart!"
• "Hmm, that's not quite on-message. Do you have any other data points we could use?"

In general, be wary of:

• Any forecasts with spurious accuracy. If someone predicts the Widget market will be worth $32.214bn in 2033, with a CAGR of 61.47%, they clearly have the mathematical aptitude of a broken toaster. It's a done by a bucketshop's intern who knows how to "fill down" on Excel.
• Any chart that shows cumulative data, rather than year-on-year shifts. They're often used to obscure trends which are uncomfortable or not PR-friendly, such as deployment of new 5G SA networks for 2022 vs. 2023 vs. 2024. Particularly dodgy are cumulative charts of financial values, such as capex, without recognising depreciation, or discounting for future cashflows.
• Data points which seem large, but have no comparisons. Is a million somethings-per-months a lot? Or do alternative applications have a billion or a trillion?
• Charts with no axis, or which don't start at zero
• Big numbers with no breakdown into (obvious and different) sub-components. Are they obscuring shifts in the underlying elements that would be awkward to discuss? A good example here is mobile data traffic figures which don't break out MBB and FWA.
• Anything which claims to be "the fastest growing XYZ market ever", without making a distinction between customers buying-in to a completely new sector (eg buying an AR headset), versus ones just upgrading an existing installed base (eg going from 4G to 5G)

Also, distrust any report put out by an industry lobby group / trade body, and produced by a consultancy, unless there's clarity over the project's upfront requirements and the knowledge of the team involved. Was it genuine research, or was it a marcomms or policy-influencing exercise in "content generation" with predetermined answers?

Anything which "stacks" outputs from the same organisation's previous studies unquestioningly is especially suspect. Some groups have the budget to churn out report after report, and they all cite each other, unquestioningly - and won't even acknowledge any rebuttals or counterpoints that have been raised.

This gambit is increasingly obvious to spot, especially now with a simple prompt on an LLM. If someone cites a report, and hasn't even bothered to ask GenAI to "summarise the most convincing criticisms" about it, you should question their competence and independence.

I normally trust analyst firms more than pure consultancies for this, as they tend to live and die on the basis of their quantitative methodologies. I'm especially wary of economics consultancies publishing on detailed technical areas such as spectrum or network APIs, or specialised spectrum consultancies weighing in on economics or apps' demand for data traffic.

Now, let's quickly dig into some particular themes for caution, that are most likely to fall victim to MWC's most mendacious metric-mongers:

Network coverage

Whether we're talking about "homes passed" by FTTH, or "5G coverage of population", this is an area where you need caution. I've written about this at length. A few things to be aware of:

• Anything that references 5G coverage, but doesn't specify whether it's just anything that triggers a 5G logo on the phone screen, or actual "real 5G" that has enough capacity and ideally a standalone core to actually do any of the stuff that 5G promised. At least some governments are now pushing for proper 5G SA coverage, after realising they'd been hoodwinked by "4G with a new hat" deployments which didn't live up to expectations. Helpfully, GSMA now has a new term "complete 5G" to add further confusion to this.
• Does the stat just refer to population coverage? "We cover 80% of the population". Hugely misleading, as many countries have 80%+ urbanisation (see this link). What about land-area coverage? What about non-populated areas, if you want 5G to be used for agtech or industrial sites? And above all, what about indoor coverage? What % of floorspace gets decent #5G (or #WiFi for that matter)
• Homes passed by fibre isn't much more useful either. Having fibre 50m away down the main street doesn't magically give an end-user a gigabit connection. And it can't be installed for free either - that last few metres might be a nightmare of landlord permissions and wayleaves and physical obstacles. Better numbers are homes connected or maybe "ready for service". Coverage / connection stats on business premises would be nice too.

I'm expecting to see some satellite coverage metrics start to be used here too. "We've got 100% coverage", without the second half of the sentence "if you stand outside and want to send a text once every 10 minutes".

Data traffic

There's usually a lot of emphasis on the growth of (mobile) data traffic. This year we can expect some embarassment about declining growth rates, which are now down in single-digits in parts of the Nordics, UK and elsewhere.

This is awkward for those in the industry who both love and hate "exponential" growth, as it gives them a stick to demand some sort of traffic-tax on web/cloud companies, usually with a lie about "generating" traffic, while simultaneously demanding more spectrum in future.

Some projections still expect 20%+ annually, usually with a vain hope that things like AR/VR will drive traffic. Or this year, probably AI, even though most of it generates text output and no, doesn't involve lots of uplink video that needs "edge inferencing".

If you're at a PR event or conference session when somebody spouts data points, make sure to fact-check them, and also ask what % of mobile data growth is coming from FWA fixed-wireless, rather than actual true mobile broadband. Ask when the forecasts were updated to reflect recent flattening-off. Laugh out loud at anyone using the word "exponential" and suggest they should be fired for incompetence.

Remember, FWA wasn't even in the original 5G use-cases or infamous ITU "triangle". It's an optional extra for MNOs. It creates 20-30x more traffic per user, so if data volumes are problematic and causing so much extra cost, why are they pushing it?

For a real laugh, ask how much traffic is from the much-promised URLLC critical services (answer = 0%) or is used indoors in places where there's decent Wi-Fi and poor in-building 5G coverage / capacity.

And if you want to really scare people, ask what happens if we ever see data traffic deflation if we get better AI-led compression, or when Wi-Fi autoconnection becomes simpler and normalised.

Messaging and RCS

We'll undoubtedly get the usual blather about RCS, now in its 17th year of failure and zombificiation. We can expect some data on the number of users that theoretically have access to it, especially given Apple's recent (and very half-hearted) support of it.

Let's see if anyone gives DAU (daily average users), especially if they can work out how many people use it for anything more than "continuity SMS" or inbound 2FA messages, because it's based on the same app. Bonus points for comparisons to WeChat, WhatsApp etc. Quoting MAUs (monthly) rather than DAUs is a massive red flag.

There will probably be some stats on the number of messages sent, both without comparisons to other apps and also...

...not recognising...

...that the relevant...

... number is actually conversations or interactions not individual messages, Harder to count, sure, but that's their problem, not yours.

Then there's RBM for business messaging or advertising. Let's see if there are any decent multi-channel comparisons, that compare user interactions vs. dedicated in-app interactions, SMS, phone, WhatsApp business, email, web chat sessions and even voice and agentic AI-based sessions.

There's probably still more remote interactions done by fax than RCS if you include B2B, although to be fair, I suspect the RCS zombie is now finally ahead of telex, carrier pigeons and semaphore for CX.

Network APIs

Never mind the McKinsey nonsense about $100 billion (or sometimes $300 billion) of API-led revenue uplift for telecoms. Nobody believes that, but repeating phantom numbers with lots of zeros is like a religious mantra for the industry, and usually needs the high priests of consultancy voodoo to intone them regularly. Hopefully nobody sacrifices a chicken in desperation to make it come true.

The real numerical questionmarks are around developers. Never mind how many MNOs have signed up to [insert aggregator or initiative name], what makes a difference is how many devs are engaged, or better still, numbers of shipping applications or commercial API calls are being made?

(I'll save some blushes here, but I signed up for one telco's dev programme a while back. I joined a webinar about QoD APIs. It had a grand total of 5 participants, including two from the MNO, and myself).

Ask for a comparison of CAMARA vs. non-CAMARA API calls for added fun - some operators have billions of proprietary accesses to their network or back-office functions.

Full points to any operator that shows a chart of how many t-shirts or pizzas they've handed out at hackathons, though.

GDP uplift

There's always some cringeworthy tosh at MWC about the contribution of 5G, or spectrum, or edge computing or something similar to additional GDP for a given nation or region. Sometimes its "consumer surplus", and sometimes they'll add in everything else they can think of, such as the cost of digging up the road, or employing people to climb mobile towers.

I realise that macroeconomic calculations (and especially forecasts) are hard, but what's absolutely inexcusable is when the modellers don't take account of the real world or the critical technical enablers required - or assumed.

Look back at the 5G predictions of GDP impact from 6-7 years ago, and then ask how many of the applications and use-cases cited are dependent on indoor coverage (patchy at best), standalone cores (only now being turned on), or the totally-fabricated concept of millisecond latency (URLLC is basically a fantasy at this stage). All predictable - unless you're an economist with zero understanding of how a wireless network works, or where.

Most such figures don't do decent comparisons with alternative options - whether that's using 4G instead of 5G, or a developer using clever software techniques to optimise a video-call or industrial system, rather than network-derived QoS.

Is the option being suggested uplifting GDP from a baseline of zero, as if that new application doesn't exist at all? Or from another technological approach which might be simpler & cheaper?

And is there an estimate of how much of that use-case's "GDP uplift" is attributable to the wireless connection, rather than the device, the application, the IT/cloud infrastructure or the other 100 elements? Or do they try to claim credit for 100%?

Green CO2 savings

A similar set of issues arise in "green network" metrics and stats.

There are many claims that use of mobile (or 5G or indeed fixed networks) leads to huge savings in carbon footprint - fewer flights because of videoconferences, wireless traffic controls reducing congestion, connected solar panels and batteries, and so on.

It's fair to say that a fibre network uses less energy than a copper one. And that newer 5G radios are more efficient than older 3G / 4G ones, nowadays sometimes in absolute terms, not the spurious and meaningless "per bit" metric. It helps if you can switch off 2G / 3G as well.

But again, this is a massive exercise in double, or 10x-counting. If I do a videoconference with a client, there's at least 2 access networks, interconnects / long-haul fibre, two end-devices, two screens, two cameras, lots of chips, a cloud platform, a video app or browsers, mics etc.

Being generous, one of the access networks (probably fixed+WiFi rather than mobile anyway) maybe accounts for 5% of the total savings. And only a handful of trips are replaced with video, in any case - most video sessions are incremental, or replace a lower-energy voice call.

On the topic of energy use, watch out for dodgy claims that more data traffic = more energy, or that watching a movie / running a GenAI query is killing the planet. There's a lot of detail and nuance that gets ignored with those claims - most of the energy goes into deploying networks irrespective of traffic levels, or manufacturing devices.

We may see some lobbying types talk about pushing for "sustainable levels of data traffic". This is disingenuous, and arguable a straight-up lie. It's just another attempt to implement some sort of data-traffic tax, and put inspection / policy software in the data path. Telcowash, basically.

If they really wanted to save power, they'd suggest people turn down the brightness of the screen, offload macro-RAN mobile traffic to WiFi wherever possible, and dial down "gigabit" targets to 500-700Mbps that's easier to deploy with wireless or satellite.

Or they'd just shrug and incentivise tech companies and datacentre owners building and deploying their own clean energy sources. Let's have telcos & Internet companies running their own nuclear SMRs or solar farms.

Other statistical minefields

This article is already long enough. I could also mention fallacies relating to:

• "Investment Gap": This is a favourite fabrication by some Brussels lobbyists and policymakers. It refers to the supposed levels of investment needed to hit the EU's arbitrary and self-imposed targets for networks. Its often linked to data traffic volumes, despite almost all the CAPEX being used to deploy coverage, not capacity. It would still be needed even with zero traffic. Another big chunk is for (mandated) generational upgrades to 5G or FTTX. And as I've noted in a recent post, there's €100s of billions in capital sitting on the sidelines from various sources, waiting for sensible regulatory reform. Instead, policymakers are still trying to find a way to charge TechCo fees by any means.
• Spectrum: Especially hilarious are demands for xxxxMHz for future 5G / 6G demand. I got kicked off an MWC webinar on this one year, because I asked too many awkward questions. Often linked to GDP uplift predictions. They're uniformly awful. I've recently taken CTIA to task for its dodgy definitions of "mid-band" in its documents and congressional testimony in the US, and it wouldn't surprise me to see similar nonsense repeated in Europe.
• Latency: another terrain of awful takes. Frankly even I don't understand some of the nuance here, so leave it to the professionals at organisations like Domos, LibreQoS and others. Lots of complexity around "latency under load", "end to end" (which ends?) and more. Be skeptical!

Overall, you get the picture - metrics and statistics cited in press releases, keynote speeches and published papers by industry groups are often highly suspect. Make sure to understand definitions, methodologies, assumptions - and ask detailed and awkward questions to stress-test them.

Good data is out there, but it's often not free, and not issued by marcomms or policy folk.

And if you're a journalist or commentator following MWC, you might find you get a better story by questioning the stats & their source (and underlying objective for publishing them), rather than just reprinting the chart. Drop me a note if you'd like a rapid rebuttal or a suggestion for some follow-up questions!