What the EU’s Mobile Roaming success taught me about innovation in bureaucracies

11 min readJun 14, 2017

Europe’s success in eliminating mobile roaming charges may be the first “data4policy” case study where the data was website traffic, and illustrates the rewards of allowing innovation to flourish at the edges of large organisations.

Expect to hear a lot from the EC about their 12-year campaign against mobile roaming charges, which disappear tomorrow across the EU.

They’ve frequently described this as “one of the EU’s biggest success stories”.

Really? I had a hand in this story, but I’m not sure it belongs alongside helping keep Europe peaceful for over half a century, creating the Four Freedoms for 500 million people and encouraging democratic rule of law in 10 ex-Soviet countries.

a major success story … if success is measured by popularity

The EU’s Roaming Saga is a major success story only if success is measured in terms of popularity. In fact, the end of this story is terrible news for the EC’s Department for Communications (“DG COMM”), who’ve been wringing it dry for over a decade.

But it’s also a perfect illustration of how about innovation happens (or doesn’t) within large bureaucracies. This story will, at first, look like an attack on centralisation within bureaucracies, but the truth is more nuanced.

But first, a little history

For me, the story started when I was on holiday in Northern Bretagne in July, 2005. I was listening, for the sheer novelty value, to BBC Five Live, which reached across the Channel into my car as I waited for my wife to return with baguettes and croissants. I was on holiday, in a relaxed and happy mood.

And then Martin Selmayr came on the radio

Until Martin Selmayr, spokesman for European Commissioner Viviane Reding, came on the radio and explained that the European Commission was going to “build an interactive website where every European could check real prices and see for themselves the reality of roaming charges across Europe”.

Great. As webteam leader for DG Information Society & Media (DG INFSO), helmed by Reding, building that website was for us.

This is not a Heroic Web-Building Tale

The world does not need another Heroic Web-Building Tale, but to explain my point I need to explain why it was vital that the site rocked. So bear with me.

Data journalism, without data visualisation

The EC was proposing a Regulation, not announcing one. With around 10% of mobile operators’ profits coming from roaming, they had a battle ahead.

Hence the 6-language website we launched the following October allowed users to explore how roaming charges varied for people from each EU country visiting six other countries, while roaming on up to six different mobile operators, both in terms of being called and calling home, pre- and post-paid.

We added SMS & data charges later and updated it for years, but even in its first iteration we had to present 10,000s of data points in a user-friendly way.

There were no cheap data visualisation resources back in 2005, so we used tables, filled by the relevant “policy Unit” — the very people who fought the project tooth-and-nail throughout that summer.

Guess who was caught in the middle?

Data4Policy: crashing your webserver helps

That evening I saw a screenshot of the site on the BBC 1 evening news, and that month it got as much traffic as the entire cross-EC Information Society Portal which hosted it. That site traffic crashed EUROPA, the EC’s million-document webserver. We had to write a note. It went to Reding’s Cabinet.

This sort of thing never happened. The idea of citizens crashing EUROPA through weight of numbers should have been laughable: EC policy announcements were until then supported online by posting a press pack to a site only journalists knew existed. Allowing Europeans to explore the story and the data themselves was a new idea, so full credit to Reding and Selmayr (and others I don’t know of) for pushing it, and us, to make it happen.

allowing citizens to explore the story and the data themselves was completely new

Because that internet traffic (apparently) mattered. In building a case for her Regulation, Reding faced stiff opposition from the Commission’s Vice President for Industry, Günter Vodafone (OK, Verheugen), one of whose predecessors announced his move to Spanish telco Telefonica while still at the Commission. The industry didn’t like Reding’s proposal, and neither did he.

Now this next bit is apocryphal — I can’t prove it — but I was told at the time that Reding won that argument in part by showing Commission President Barroso that website traffic note.

Up until then, all Reding had were her instincts that regulating roaming charges would be a popular move. The web traffic data proved her right, showing that this story reached outside the Brussels Bubble and showed the Commission doing something for the mythical ‘man in the street’.

The web traffic data proved her right

The EU’s Constitution had been rejected by France and the Netherlands that Spring. Barroso needed a good story. He sided with Reding.

If true, this was the first (and probably still the only) time EU legislation has actually been influenced by website traffic data, and is probably the first example of data-driven EU policymaking.

True or not, DG COMM has been milking it ever since ;)

Innovation starts at the Edge

Which is ironic, because if DG COMM and the Commission’s central IT department (“DIGIT”) had had their way, we would never have been able to build that website.

By 2005, most EC Departments (“DGs”) were publishing websites using Documentum-based software developed by DG COMM & DIGIT. In direct contradiction to Commission “EUROPA 2nd generation” policy (decentralised, cross-silo, interactive websites), Documentum provided zero interactivity and centralised control over EUROPA. And they were keen that everyone used it.

As set out later, they had good reason to centralise IT in this way. But we could never have built that site with Documentum.

Of course, some DGs used more dynamic technologies, often built for them by subcontractors. However, even by end August the Cabinet-DG shitstorm was still in full swing. Subcontractors could never have defined the roaming site to get a contract signed in time to build the site and get it live by early October.

an integrated team of communicators and developers

At INFSO, however, we had our own developers, and freedom to act. That flexibility was crucial, and stemmed from the DG’s genesis as a ‘task force’, with its own team of web developers, in the 1990s. Somehow INFSO’s Comms Unit had managed to hang on to them, creating an integrated developers-alongside-communicators team that most 20th century organisations (private or public) would not achieve for 5–10 more years.

That integrated, in-house team was critical to the roaming site’s success. If INFSO had followed the “Central Policy” (and God knows I had to write a few analyses defending our position), the roaming site would not have happened as it did, and would not have gotten that traffic.

Decentralisation, in other words, played a part in the roaming saga.

Innovation rarely escapes the Edge

Another example shows how over-centralisation can prevent innovations being mainstreamed across large organisations.

One of the reasons we had those developers was that in 2002 we launched an EC pilot project: the EC’s first cross-silo, thematic portal, covering the work of 14 different Departments. This required a network of around 100 editors and content providers across the EC. No chance of doing that with Documentum.

That editorial network primarily used Newsroom, a cross-silo web content management system we developed to internally share and externally publish time-dependent content and library items. It also distributed that content via multiple enewsletters and (another first) RSS and user-customised XML feeds. Combined with the EC’s first Web2.0 site, which we had also launched in 2002, this meant we had released multi-application ‘INFSO accounts’ in 2003, possibly before Google had released ‘Google accounts’.

By 2005 the Newsroom had already been adopted by other DGs, probably the first time an online tool had been built by a “policy DG” and used elsewhere. But it was a rocky road: when a manager at another DG discovered his team using INFSO software, for example, he forced it off their website, harming their users, until they had built their own. Sigh.

And it was only in 2015 that COMM & DIGIT ‘corporatised’ it, making it available to the entire EC, after trying to build a Newsroom in Documentum.

EUROPA would have been better, earlier, if it had not taken 12 years to mainstream this innovation

Today Newsroom is used by almost 2000 EC publishers, publishes into over 60 environments, and manages 600,000 subscribers of over 200 newsletters. Call it a success story if you like, but I can’t help think that EUROPA would have been better, earlier, if it had not taken 12 years to mainstream this innovation across the Commission.

Centralised economies of scale v. Decentralised innovation

This post is not an attack on centralised IT services.

It is an observation that:

if INFSO had followed the path COMM & DIGIT wanted back then, the roaming site could never have been published with all that data, and COMM could have missed out on one of their best stories. Or maybe not — I’ll probably never know for sure.
the Newsroom could have been mainstreamed a full decade earlier, benefiting the Commission greatly.

I suggest that large organisations need an innovation process specifically designed to counteract the downsides of centralisation.

Centralised services: lowest common denominator

To design that innovation process, you first need to understand why IT is centralised in the first place — to:

create economies of scale in hardware, software, skills, etc.
impose uniformity in look and feel, message, etc.

These are important, worthy goals — EUROPA was an absolute mess (as opposed to today’s relative mess), with each DG doing its own thing, no coherence in design or technology strategy, and zero cooperation. Centralisation can — if done right — save money, create critical mass and bring order to chaos.

Centralisation can save money, create critical mass and bring order to chaos

However, by definition it results in:

“lowest common denominator” services, to be used by many departments — after all, that’s where the economies of scale are
rules on what is and is not allowed - innovation-stifling, by definition.

The problem: these two results combine to create rules against innovation.

The Mother of Invention lives at the Edge

Centralised services cannot be expected to understand the diverse specialised needs of the many diverse DGs along the organisation’s edge.

Roaming and Newsroom are two examples of how Necessity is the Mother of Invention — i.e., new needs come from “the Edge”.

Within the EU Commission, the Edge is where the 20+ different DGs who actually do Europe are, whether they’re subsidising agriculture, funding research, regulating the telco industry or supporting gender equality.

[And sorry, but here I’ll quote myself explaining Communities of Practice to DIGIT & COMM, just to stress that this is not a criticism of these DGs, but simply inherent to the architecture of centralisation:

“I remember looking at the blank faces and thinking “these guys have absolutely no idea of what our users actually need”. It took me a while to realise that this wasn’t their fault.”
- The UK’s GDS meltdown couldn’t happen in Brussels, right? (Feb. 2015)]

Centralised services are not being ‘anti-innovation’ — they’re simply doing their job

Centralised services are not being ‘anti-innovation’ — they’re simply doing their job: creating economies of scale and bringing order to chaos. But in the process they create conditions which prevent their clients (the DGs) meeting the specific needs of their clients — European citizens.

Balancing centralisation and innovation

So it was a perfectly natural for the Commission, facing the monster-that-was-EUROPA, to centralise control and bring the Beast to order.

The benefits of that approach are, hopefully, beginning to emerge from COMM’s digital transformation programme.

The policy DGs’ specialised needs, however, haven’t disappeared.

This problem is not unique to the Commission. After researching best practices for a client last year, I ended up recommending balancing centralised control with a formal innovation pipeline.

Innovation pipelines: from Edge to Centre, and Back

There are many forms of innovation pipeline. In this particular case, it exists to ‘funnel’ innovations developed from the Edge into the Centre, from where they are Mainstreamed out to the rest of the organisation.

It looks a little like this (notes follow the slideshare):

Probably shouldn’t have bought that Surface Pro

In this environment, innovation is allowed at the Edge under controlled conditions via an innovation proposal system where:

Innovators (“Pathfinders”) at the Edge propose ideas reflecting their specialised needs, possibly after developing betas or MVPs to test their assumptions
the Centre helps polish and refine the proposal, in the process:

steering them to existing solutions they didn’t know about
clustering similar ideas together and building a Movement around them
imposing innovation management best practices (ringfenced resources, top-level support, fixed evaluation deadline, clear success definitions and metrics, etc.)

3. the Centre evaluates each finalised innovation proposal objectively, using criteria such as likely impact, feasibility and effort, and:

actively supports the best ones
kills the bad ones
doesn’t get in the way of good ones

4. when each innovation project reaches its pre-agreed deadline, it is evaluated using the success definitions and metrics agreed in phase 2

5. the failures are shut down and lessons are learnt

6. the successes are celebrated and Mainstreamed: migrated to the Centre, and so provided to the rest of the organisation

A supporting culture

The Newsroom followed this path, but it took 12 years because it faced a decade of headwinds as it traveled down that funnel. That headwind was cultural resistance.

Establishing a formal innovation process will not, by itself, change those headwinds into the tailwinds needed to encourage innovation. An innovation funnel will only function well within an environment where:

resources and experiments are allowed at the edge
pilots are encouraged, and failure is a possible outcome
the modus operandi is partnership, not rivalry
innovations — and innovators — are celebrated, not feared

In the end, processes are nothing without a supporting culture.