The headline
I led the build of Vodafone’s first SMB managed-services line for Microsoft 365 from a blank page. Atlas (internal codename for Microsoft 365 Business with Vodafone Services) covered the full customer lifecycle, from pre-sale licence advisory through onboarding and in-life support.
The business case I built with my manager Steve cleared Vodafone’s NPD gates, board ELT, and P&L sign-off on a 2.5-year payback. By the time I handed it off in November 2023, Atlas had hit the payback, was running across 5 markets, and sat behind ~20% quarter-on-quarter service revenue growth in the wider SaaS Managed Services portfolio.
Context
SMB was Vodafone Business’s growth strategy. We were already selling Microsoft 365 to a large SMB base across markets, but we had no mechanism for stickiness. Customers bought the licence and then went quiet. There was no way to extend lifetime value, no adjacent product to attach, and no service wrapper to keep them engaged at renewal.
Atlas was for SMBs who wanted to focus on their actual business and stop being bogged down by Microsoft 365 administration. The job we were hired to do was straightforward: outsource M365 IT operations and support, so the customer’s CEO-who-is-also-IT-lead could go back to running the company.
The competitive set was mixed. Microsoft sold direct to SMBs. SaaS-native players like IONOS and GoDaddy were targeting the same SMB segment with bundled productivity-plus-domain-plus-hosting offers. Licence resellers like Crayon, Bytes, and SoftwareONE were starting to layer services on top. A long tail of local IT managed service providers operated in every market, with especially deep coverage in Germany and Italy. And the biggest competitor of all was the customer doing the work themselves. Absorbing M365 admin in-house wasn’t a great outcome for any SMB, but it was the default, and breaking it was harder than beating any one named vendor.
I positioned Atlas as a SaaS-native managed service, not a traditional telco add-on. The benchmark set was Stripe, Sage, IONOS, and GoDaddy. Not Telstra Purple, T-Systems, or any other telco peer. That positioning choice shaped the digital experience, the pricing simplicity, and the GTM messaging.
To pressure-test my own assumptions about what an SMB managed service should and shouldn’t do, I bought IONOS’s managed M365 service as an SMB customer and lived with it for the full 12-month subscription. The lived experience informed multiple Atlas design decisions: what to include, what to skip, where IONOS’s CX fell short, and where it set a standard we had to match. Reading competitive analysis is one thing. Running the competitor’s service through your own onboarding, support, and renewal is another.
Atlas attached to four buyer states: new Microsoft 365 customers, existing Microsoft 365 customers, new Vodafone customers, and existing Vodafone customers. That gave it a wide addressable base across the SMB segment, including customers coming from Google Workspace or on-premise Exchange who needed to migrate into Microsoft 365 in the first place.
Steve hired me to fix that.
Problem
What SMBs were actually struggling with
SMBs don’t have IT teams. In most of the customers we interviewed, the CEO or a director was also the IT lead. They didn’t have the time or the skills to administer M365 properly: activating licences, troubleshooting Exchange, standing up SharePoint or Teams sites, managing Intune policies.
We validated this two ways. Inside Vodafone, we ran customer panels, customer interviews, support-ticket analysis, churn analysis, and workload-level adoption data. Externally, we commissioned a structured 48-SME study across the UK, Germany, Italy, and Spain (12 SMEs per market, sized 20 to 200+ employees) with Reply agencies, blending qualitative interviews with quantitative agree/disagree statements.
The signal was strong:
- 100% of UK and Italian SMEs and 92% of German SMEs agreed a managed cloud-based service would let them optimise their software use as they grow.
- 100% of UK and German SMEs agreed that on-demand experts and a help desk would ensure the service is used to its full potential.
- 92% of UK SMEs and 67% of German SMEs said they’d consider leaving their current IT supplier if a new one offered round-the-clock support with multi-channel access.
Discovery surfaced three behavioural personas we designed against: Anja the Anxious Economiser (office administrator, youngest decision-makers, ROI-obsessed), Frank the Cautious Follower (CEO/director, mid-sized firm, follower not innovator, craved personal supplier relationships), and Guy the Established Pragmatist (frequent user, long-established firm, lowest internal IT skills, trusted Microsoft and HP). Atlas’s MVP scope was shaped to land hardest with Anja and Frank.
What Vodafone was losing
Underutilisation came in two shapes.
The first: customers bought the right licence tier but only used a fraction of it. Premium customers running on Exchange and Teams, with Intune, Defender, conditional access, and SharePoint all dormant.
The second: customers bought the wrong tier altogether. Premium when Basic or Standard would have done. Vodafone’s sales weren’t pushing the wrong SKU. The Microsoft licence map is complex enough that not every SMB buyer understood what they were paying for.
Both shapes hurt renewal. If the customer doesn’t see the value, they churn.
When I arrived, no one else inside Vodafone was solving this. Blank page.
Microsoft 365 Business SMB licensing primer (Atlas timeframe, 2021-2023)
300-user max, three tiers.
- Basic. Cloud Office (web/mobile), Exchange, Teams, SharePoint, OneDrive (1TB). No desktop Office.
- Standard. Adds desktop Office (Word, Excel, PowerPoint, Outlook, OneNote), Loop workspaces.
- Premium. Adds device and security management: Intune (MDM, Autopilot), Defender for Business, Defender for Office 365 P1, Entra ID P1 with conditional access, BitLocker, Windows 11 Business.
Copilot and the Defender/Purview suites came later (2024-26) and weren’t part of the licence map Atlas was designed against.
Approach and decisions
First 90 days
I started by mapping stakeholders, dug into the existing business case until I understood the numbers, and read every piece of discovery the marketing team had already paid for. The external research gave me a head start. The live customer panels and interviews were where I anchored my own conviction.
Out of discovery came four CX design principles that ran through every Atlas decision: act as a trusted business advisor, present as one Vodafone (not separate products), champion all business customers, and respect customers’ time. These framed what made the MVP cut and what got deferred.
Three decisions that shaped Atlas
Decision 1
Start with a bundled SKU at MVP, then add standalone and add-on flexibility at v2.
Alternative considered: launch v1 with all three commercial constructs (bundled, standalone, add-on) on day one.
v1 went bundled-only. Atlas plus the M365 licence sold as a single SKU on the Vodafone marketplace, targeting new customers. One transaction, one price per seat. The customer couldn’t buy the managed service without a Vodafone-purchased licence. The simplification was deliberate: a single SKU cut GTM complexity for sales, simplified pricing for market teams, and made stickiness automatic at point of sale. The trade-off I accepted: customers who already had M365 from another channel couldn’t buy just the service, which capped v1’s addressable market to net-new customers.
At v2, we extended the commercial model. Atlas became available as a standalone managed-service licence and as an add-on layered onto a customer’s existing Microsoft licence. That opened the existing M365 base to the same managed-service motion without forcing them to re-purchase their licences through Vodafone. The v2 expansion was driven by clear demand from existing customers we’d seen during v1.
Pricing varied by market because Germany’s price didn’t work in Italy or Spain.
Decision 2
Hybrid build with Avanade over full outsource or full self-build.
Alternative considered: full outsource to Avanade (Microsoft/Accenture JV) was too expensive. Full Vodafone self-build was too slow and we didn't have the skills.
I chose hybrid: Avanade built the managed-service capability and platform, staffed delivery into Cairo for the German MVP, then faded as Vodafone built its own bench. Avanade stayed on as the consultancy escalation point before tickets went to Microsoft L3. Vodafone owned billing and integration with our own systems. This let us start fast, learn from Avanade, and bring the service in-house over time without the full outsourcing cost stack.
Decision 3
Germany first over Italy, even after Italy had said yes.
Alternative considered: hold Italy as the planned MVP market after Germany lost interest.
At the 11th hour Germany came back wanting in, on a timeline I couldn’t refuse. Germany is Vodafone’s biggest market, so I pushed Italy back and put Germany first. Italy turned out to be less ready than they’d signalled. Germany turned out to be more ready than any market. The commercial call was right. The cost was real: the language and localisation rework that came with the pivot was expensive, and the lesson is in the Reflection section below.
MVP shape
“Wagile” build, somewhere between waterfall and agile. We got to MVP slower than I’d have liked. The v1:
- A “swivel chair” support model. Service engineers manually switched between two systems rather than waiting on full integration. Ugly but shippable in weeks.
- A single MVP market: Germany.
- A live test of the advisory questionnaire site, where sales and customers could identify the right licence tier.
- A Cairo service desk with a small team of German-speaking support engineers.
One scope cut worth naming. The early design had a Bronze, Silver, Gold tier structure for the managed service, mirroring the Microsoft Basic, Standard, Premium SKUs. We cut to a single managed service before launch. Differentiating across three tiers added GTM and operational complexity that didn’t earn its keep at launch volumes. Easier to land one well-defined service than three half-defined ones.
Cairo support centre
Cairo was cheaper than the UK or other European countries. Vodafone already operated other support services there. And we could find German-speaking Egyptian support staff for the German MVP at the volume we needed. The alternative, outsourcing all support to Avanade in the UK, was a cost stack we couldn’t justify.
The advisory questionnaire site
The questionnaire shipped as a v2 addition, not day one. Getting the dynamic logic right took time (each answer driving the next question). We loosely based it on Microsoft’s own questionnaire framework, built it with Avanade’s input, and shipped it with Vodafone’s developers. It served two audiences. Sales used it in customer conversations as an enablement tool. Customers could self-serve on the public Vodafone site.
Microsoft constraint that shaped the journey. Under Microsoft’s New Commerce Experience, you can’t downgrade SKUs or seat counts mid-term. So Atlas’s right-sizing was a renewal-cycle lever, not a continuous one. Get the advice right at point of purchase or renewal, or wait 12 months.
Exec buy-in
Steve and I built the business case together. It was detailed: ramp rates, break-even seat counts, market-level cost stacks, multi-year P&L. We took it through Vodafone’s NPD gates, board ELT, and P&L sign-off. Approved on a ~2.5-year payback.
What shipped
Service architecture
Atlas sat between the SMB customer and the customer’s Microsoft 365 tenant. Five components made up the wrap: the advisory tool that shaped the licence-tier decision before purchase, the migration capability that moved the customer onto M365 from Google Workspace or on-premise Exchange, the Cairo service desk for in-life support, the billing and Vodafone-systems integration, and the GDAP layer that let Cairo perform admin actions on the customer’s tenant without holding admin credentials.
Customer journey
Two paths, depending on the buyer state.
New customers (sales-led)
A salesperson directs the customer to the advisory site, walks them through the licence and service-tier recommendation, captures the order, and confirms licences. The deal hands off to operations.
Existing customers (self-served)
The customer logs into the Vodafone SaaS marketplace, picks the licence-plus-service bundle, and checks out. The service desk gets the notification automatically and starts onboarding without sales involvement.
For customers coming from Google Workspace, Google Mail, or on-premise Exchange, the service included a migration step into the new Microsoft 365 tenant: mailboxes, contacts, calendars, and shared workspaces. That turned Atlas into a win-from-Google motion alongside the SMB stickiness motion.
Either path lands the customer in the same place. Cairo runs the onboarding call, sets the customer up, and requests GDAP (Granular Delegated Admin Privileges) access into the customer’s M365 tenant with RBAC roles. That’s how Cairo performs support actions on the customer’s behalf without holding tenant admin credentials.
Day-to-day support runs across phone, email, chat, and text. If Cairo can’t resolve, the ticket escalates to Microsoft L3.
Sales enablement
Battle cards, playbooks, training, and customer decks. Decks included the Microsoft licence breakdown and Atlas support contact details. My group team produced the English content. The German market translated it locally, after Vodafone’s own attempts missed enough local nuance to be a problem.
Digital experience design
We benchmarked the customer-facing patterns against Stripe (dashboard and analytics), Sage (quote questionnaire), Slack (chat support), CloudM (migration progress), and Bulb (migration comms) before settling the UX for quote, dashboard, support, and migration progress reporting. A narrative customer-journey storyboard, built around a fictional Italian SME persona working through the full lifecycle from awareness to renewal, was the internal source-of-truth for design decisions during the build. (Storyboard is a Vodafone internal artefact, not shown publicly.)
Operations
A live dashboard tracking customer count, support desk call volume, and the rest of the standard managed-services metrics. SOPs for the Cairo desk. Training material for the support team.
Externally visible artefacts
The live product page sits at vodafone.com, with country variants for Germany, UK, Italy, and Spain. Everything else lives behind Vodafone’s intranet and under NDA.
Outcomes
Atlas was a 0→1 strategic bet for Vodafone Business. The right success metrics for the first 24 months are payback period, growth rate, and customer health, not absolute scale. The customer base at handoff was the runway being built, not the destination already reached.
- 5 markets launched. Germany (MVP), Italy, Spain, UK, Ireland.
- Germany held ~50% of all Atlas seats at handoff. Two reads of the same number. First, validation that the 11th-hour Germany-first pivot was the right commercial call. Second, evidence that the playbook hadn’t yet fully translated to Italy, Spain, UK, and Ireland. The next 50% was harder to win than the first.
- ~20% quarter-on-quarter service revenue growth across the wider SaaS Managed Services portfolio Atlas sat inside.
- ~53% healthy customer share.
- 2.5-year payback period hit by the time I handed Atlas off in November 2023.
A note on attribution. Vodafone won Microsoft Telco Partner of the Year (2025) under Microsoft’s Elevate program. That award covers the wider Vodafone Microsoft 365 portfolio, not Atlas specifically. I worked across both, and Atlas was a contributing pillar, but the award isn’t an Atlas number.
Reflection
Three things I’d do differently
- Hold markets to commitment. The Italy → Germany flip looked clean on paper. The language and localisation rework underneath it cost us more time and money than the commercial logic credited.
- Cost stack before pricing. We didn’t have a fully bottom-up costed operating model at the start. Pricing didn’t yield enough margin, and we fixed it in v3. Next time, the cost model comes first.
- Smaller MVP. I cut MVP scope, but not hard enough. The build took longer than it should have. Cut to the smallest viable subset of services next time, even if it feels embarrassing on day one.
Three things Atlas taught me
- Big-org navigation. In enterprises like Vodafone, you need multiple yeses to move and one no to stop. 80% of the senior PM job in big orgs is navigating “no”, 20% is building “yes”. Plan your air-cover and your blockers as carefully as you plan your roadmap.
- Continuous discovery for SMB. SMB customer needs shift on a monthly cadence, not an annual one. Discovery has to be a continuous loop with a live customer panel, not a sprint at kickoff. If your discovery insights are more than a quarter old, they’re decoration.
- Partner-led delivery works with guardrails. Avanade worked because we defined scope, ownership, interaction protocol, and handoff up front. Without those four, partner-led delivery becomes partner-led chaos and the cost stack creeps. Negotiate the fade-out path at contract time, not at scale time.
What I’d build next on Atlas
Atlas’s roadmap was mine to own. Four bets shaped where the product would go after my tenure.
- Google Workspace coverage. Extend the managed service to wrap Google Workspace customers, not just Microsoft 365. Same advisory motion, same Cairo onboarding model, same in-life support. Doubles the addressable SMB productivity-suite market without rebuilding the operating model.
- Copilot adoption services. Add a services layer to drive Copilot uptake inside existing Atlas customers. Adoption is the gating issue for Copilot ROI, and a managed service is the right vehicle to close that gap.
- Microsoft Enterprise licence coverage (E1, E3, E5). Take Atlas up-market beyond the SMB Business SKUs into mid-market and enterprise. Different operating model, bigger tenants, more compliance scope, named account managers, but the underlying value-prop transfers.
- Re-introducing service tiers. Once Atlas volumes justified the GTM complexity, reintroduce a tiered structure (Bronze, Silver, Gold or equivalent). Day 1 cut to a single service was right. The long-term answer was probably tiers.