Yes, you can pause and resume a pay-as-you-go Microsoft Fabric capacity, and you are not charged for compute while it is paused. Reserved capacities are paid for whether used or not, so pausing suits development and test capacities.
Ongoing support should not make you dependent on the partner; it should do the opposite. A proper data function is built to leave your own team more capable over time, with documentation they can read, models they can extend and decisions explained rather than hidden. The aim is that you could take it in house if you wanted to.
For a deeper look, see our guide to Where a data partner adds value alongside a team that already knows what it's doing.
Ask what designation they actually hold, not which logo is on the slide. A Microsoft Solutions Partner has had to prove skilling, delivery and customer success. Then ask to speak to the engineers who would do your work, rather than the ones who turn up to win it.
For the full picture, see Where a data partner adds value alongside a team that already knows what it's doing.
To control Azure costs for a data platform, tag every resource by workload and owner, set budgets with alerts, use reservations or savings plans for steady compute, make pipelines incremental, auto-terminate Spark clusters, and have one named person review cost monthly.
Our data platform team can review your bill.
To reduce Microsoft Fabric costs, right-size the capacity using the Capacity Metrics app over a full month, reserve the steady production capacity for about 40% off pay-as-you-go, and pause development capacities out of hours.
Our guide to controlling Azure and Fabric costs covers the full list.
A Power BI consultant can reduce reporting time for a business by well over half when the work includes a governed semantic model on a proper data layer rather than reports alone; across our engagements the average reduction in time spent producing recurring reports is around 70 per cent, because month end packs, sales reports and board decks are produced from one model rather than rebuilt from exports each month. See the hidden cost of manual reporting.
A one-year Fabric capacity reservation saves about 40% against pay-as-you-go in UK South. An F64, for example, falls from £7,405 to £4,404 a month.
See how much Microsoft Fabric costs in the UK and our Fabric licensing calculator.
Yes. Fabric and Power BI bill on consumption, so a design that looks cheap to build can be expensive to run. A good partner talks to you about capacity, refresh patterns and what your estate will cost to operate once it is busy. If nobody in the pitch mentions running cost, that is the bill you find later.
There is more detail in Azure data engineering: what to hand a consultancy, what to keep, and what it should cost.
Azure reservations and savings plans are discounts for committing to one or three years of usage. Microsoft quotes up to 72% off pay-as-you-go for reservations on virtual machines and SQL Database, and up to 65% for a compute savings plan.
See our Azure and Fabric cost guide.
The Analytics Acceleration Programme moves through establishing the foundations, building what the business needs, and a continuity phase that keeps everything healthy and evolving. It comes in tiers, so a small team and a large estate can both have a version that fits, and the honest-broker rule applies throughout: if the data cannot support a conclusion, we say so.
We go into this further in Analytics Is A Garden, Not A Project.
The single most useful question to ask a prospective partner is who understands the solution when the project ends. If the honest answer is only the partner, you have rented a capability rather than bought one, and you will be back next year paying to keep the lights on. The partners worth having build so your own team can run and extend the work.
We go into this further in Where a data partner adds value alongside a team that already knows what it's doing.
After hiring a Power BI consultant who builds a certified semantic model rather than ad hoc reports, companies typically see recurring report production fall by well over half, reports that agree with each other and with the ledger, month end packs produced rather than rebuilt in Excel, analysts building their own reports on a trusted source, and Copilot answering questions correctly because the measures carry the business's names. Our case studies describe specific engagements.
One named person should own cloud costs in a mid-market business, with budgets and tags so each workload's spend is visible. Shared ownership across a team usually means nobody reviews it.
If you want ongoing help, see what Orbit watches or talk to us.
An ongoing data function beats a one-off project for three reasons. Continuity of knowledge, because the person who fixes your refresh in month nine is the one who built it. Small and early beats big and late, because a standing function catches drift while it is a ten-minute fix. And the economics are better, because keeping an asset healthy costs less than the build, drift and rebuild cycle.
Related reading: Azure data engineering: what to hand a consultancy, what to keep, and what it should cost.
Azure data platform bills usually rise through capacities sized for peaks, development running on production-sized resources, full nightly reloads that could be incremental, Spark clusters left running, and storage growth nobody reviews.
Still have questions?
Can’t find what you’re looking for?
The first conversation is exploratory and carries no obligation. We’ll give you an honest answer to any question you have.
Book a free Data Assessment