Power Automate is great at some things and bad at others, and most consultancies will not tell you which is which because they only sell one tool. We sell whichever tool is right. For roughly half the projects pitched as Power Automate builds, we would recommend custom code because licensing maths or workflow complexity does not fit. Honest about where Microsoft's tool shines, where it falls over, and what the per-user-per-month creep looks like at scale.
Power Automate is the right answer for a specific shape of automation problem. The shape is more constrained than the marketing suggests, and the cost of using it outside its sweet spot is high. Three conditions usually need to hold.
Outlook is your email, SharePoint your document store, Teams your communication. The standard Office 365 licence is already paid for, which means standard Power Automate connectors are bundled at no incremental cost. Automating "when an email arrives in shared inbox X, save the attachment to SharePoint folder Y, log a row in Excel Z, notify the team channel" is a 30-minute build that runs forever for free.
A trigger, a few steps, maybe an approval, an outcome. Linear or near-linear logic. No more than three or four conditional branches. No loops over arrays. No complex data transformations. No error handling more sophisticated than "retry once, then notify someone." Most workflows that look simple actually are, and Power Automate handles them well.
Hundreds of runs per month, not hundreds of thousands. Latency per step (30 to 120 seconds for some connectors) is tolerable. Internal, async, "completes within a few minutes" is fine.
When all three hold, Power Automate is the right answer. The build is fast, the cost near-zero, the IT team can maintain it without external help. Most expense-claim approvals, leave requests, document handoff, basic notification routing, and SharePoint-driven workflows fit this shape.
Almost every workflow we are asked to automate already exists in some hybrid of Excel, Outlook rules and Teams messages:
It breaks predictably. The person who maintains the Excel goes on holiday and nothing gets logged for two weeks. Outlook rules apply only to one mailbox so the team cannot share visibility. The Word template has three slightly different versions floating around. The SharePoint list has stale columns nobody dares delete.
Power Automate replaces this shape well. The flow becomes the source of truth for "what happened when," the SharePoint list or Dataverse table the durable record, notifications go to the channel automatically, the Word document is generated by the flow. For the simple case, this is a 1 to 3 week build and a meaningful upgrade.
The shape that does not work well: anything with rich business logic. Those workflows end up with 40-step flows in the designer, sub-flows calling sub-flows, error handling tacked on, an audit trail impossible to read. At that point, a 200-line script in a proper backend service is faster to build, faster to run, easier to maintain, and an order of magnitude cheaper.
These are illustrative shapes drawn from UK Microsoft-anchored business patterns. The fixed-quote bands are the real ranges we quote for builds of this shape; the workflows described are the patterns that fit each context.
Profile: deep in Microsoft 365. SharePoint document libraries, Teams channels per client, Outlook shared inboxes. Typical pain: when a client emails the shared inbox, nobody is sure who picked it up. Verdict: perfect fit. The shape that fits: email arrives in shared inbox, create SharePoint list item, post to Teams channel, assign owner via approval, log resolution. Typical band: £3,500 to £6,000 fixed quote, two to four weeks. Standard connectors, no premium licence, internal IT maintains.
Profile: Dynamics 365 sales and ops, Sage 200 finance, custom WMS. Typical pain: when a sales order is created in Dynamics, the warehouse needs pick instructions within 15 minutes, currently a sales admin copies across. Verdict: workable but expensive on Power Automate. Sage 200 and custom WMS are premium connectors, requiring Premium at £12/user/month for everyone on the flow (around £7,000 to £10,000 build plus £4,000 to £6,000/year licensing). The shape that fits better: an Azure Functions service that polls Dynamics and pushes to WMS. Typical band: £5,500 to £8,000 fixed quote, £60 to £120/month hosting.
Profile: matter files on SharePoint, case management on Proclaim, e-sign on Adobe Sign. Typical pain: when a matter is closed in Proclaim, the SharePoint library should be archived, audit logged, closing letter generated. Verdict: hybrid fit. SharePoint and Adobe Sign parts are perfect for Power Automate; Proclaim needs an HTTP call with auth. The shape that fits: Power Automate for SharePoint + Adobe Sign + Word template, with a small custom auth-handler for the Proclaim trigger. Typical band: £5,500 to £8,500 fixed quote, three to five weeks.
Profile: Office 365 E1 via the non-profit programme, SharePoint documents, Dynamics 365 for fundraising. Typical pain: grant application workflow involves multiple internal approvers, document version control, write-up to Companies House. Verdict: excellent fit. Non-profit licensing makes Premium essentially free, the workflow is approval-heavy which Power Automate handles natively. Typical band: £4,500 to £7,000 fixed quote, three to five weeks.
Profile: Office 365 Business Premium, all client work in Excel, no Dynamics. Typical pain: client onboarding requires populating 6 Excel workbooks and 3 standard emails, currently a manual 90 minutes per client. Verdict: poor fit despite the Microsoft appearance. The 6 workbooks each need open/write/save with the Excel connector's quirks; Power Automate's Excel handling is awkward at this depth. The shape that fits better: a custom onboarding form with a backend service that generates all 6 workbooks from a single source. Typical band: £6,500 to £9,000 fixed quote, four to six weeks, no licensing creep.
Pattern across the five: Microsoft estate is necessary but not sufficient for Power Automate to be the right answer. Workflow shape matters more than connector landscape, and licensing maths matters more than the marketing suggests.
Each option has a sweet spot. A consultancy that only sells one option is selling whichever they happen to sell, not whichever is right for you.
| Option | Typical cost | Sweet spot | Pain zone |
|---|---|---|---|
| Power Automate (standard connectors) | Bundled with Office 365 | Microsoft-only workflows, moderate volume, simple logic, internal use | Non-Microsoft integration, complex logic, scale |
| Power Automate Premium + per-flow plans | £12/user/mo + £80/flow/mo | Microsoft estate plus a small number of premium connectors | 10+ flows, 50+ users; licensing creeps to £20,000+/year |
| Zapier / Make / n8n (low-code automation) | £20 to £400/mo per workspace | Non-Microsoft estates, SaaS-to-SaaS integration, marketing and ops workflows | High-volume runs, regulated workflows, anything needing data residency |
| Custom Azure Functions / serverless | £3,000 to £15,000 build, £30 to £200/mo hosting | Real backend logic, volume, regulated workflows, integration with anything non-Microsoft | Tiny one-off workflows where Power Automate would take 30 minutes |
| Custom Node, Python or .NET service | £8,000 to £40,000 build, £30 to £500/mo hosting | Anything genuinely business-critical, high volume, complex, owned outright | Workflows that genuinely live within the Microsoft estate alone |
| Hybrid: Power Automate + custom backend | £6,000 to £18,000 build | Microsoft-estate triggers and notifications with heavy backend logic in a proper service | Single-tool simplicity is preferred (the hybrid has two systems to maintain) |
The hybrid pattern (Power Automate for Microsoft-estate triggers and notifications, custom service for heavy logic) is where roughly a quarter of engagements land. It is the right answer when neither pure tool fits cleanly.
"Use Power Automate Desktop for everything." Power Automate Desktop is RPA: it drives the UI of existing applications like a person would. Genuinely useful when there is no API. But fragile: UI changes break the automation, latency is bad, concurrency is hard. Organisations build core processes on Desktop and spend the next two years firefighting. If the target has an API, use it. Desktop is the last resort, not the default.
"Move all our integrations to Power Automate to standardise." Sounds clean. The reality: some integrations belong in Power Automate, some in custom services, some in the source systems themselves. Forcing everything into Power Automate is how organisations end up paying £40,000/year in licensing for what should have been £8,000 build plus £100/month hosting. Standardise on a design pattern (event-driven integration with error handling and audit trail), not a tool.
"Build the whole business-critical process in Power Automate because IT can maintain it." Surface logic. Reality: any flow beyond about 20 steps becomes harder to maintain than a properly structured 200-line script would have been. The visual designer is excellent at small flows and quietly terrible at large ones.
Real bands. Fixed quotes, scoped before any build starts.
| What it is | Timeline | Fixed quote |
|---|---|---|
| Scoping (assessment of whether Power Automate is the right tool, written delivery plan, fixed quote for the build) | 1 to 2 weeks | £1,500 to £3,500 |
| Single focused Power Automate flow (Microsoft-only, standard connectors) | 2 to 4 weeks | £3,000 to £8,000 |
| Multi-flow Power Automate build (5 to 15 flows, premium connectors, approval workflows) | 4 to 8 weeks | £8,000 to £20,000 |
| Hybrid Power Automate + custom backend (the right answer for complex workflows) | 5 to 10 weeks | £10,000 to £30,000 |
| Full custom replacement of an over-engineered Power Automate estate | 8 to 16 weeks | £20,000 to £50,000 |
| Power Platform audit (review of existing flows, licensing optimisation, recommendations) | 1 to 2 weeks | £2,500 to £6,000 |
| Tech partnership (we run, maintain, improve) | Monthly, no lock-in | £450 to £1,500 / month |
The audit engagement is increasingly the first piece of work for organisations who have built up a Power Automate estate over two or three years and are paying more in licensing than the workflows justify. A two-week review surfaces the creep, identifies flows to migrate to custom code, and quantifies the annual savings. Adjacent: AI implementation, app and product builds, and the tech partnership retainer for ongoing maintenance.
I am Oliver Wakefield-Smith. More than twenty years building software for businesses that needed systems to talk to each other: satellite comms, UK broadcast, sports media, water utilities, retail, industrial engineering, public sector. Integration is a constant theme. The lesson: the tool follows the problem.
Power Automate is one tool of several. We are not a Microsoft Partner, so we have no commercial incentive to recommend it when it is not the right answer. We will tell you honestly, project by project, which tool fits, and build whichever wins the scoping argument. The team that builds is a directed team of AI agents, which is why we can deliver custom-code alternatives at fixed-quote prices that compete with low-code tools on speed.
Adjacent: where AI earns its place, the custom app build shape for workflows that have outgrown low-code, and our AI consultant London page.
Three conditions. First, you are already deep in Microsoft 365 and paying per-user licences anyway. Second, the workflow is genuinely simple: a trigger, a few steps, an approval, an outcome. Third, volume is moderate, not tens of thousands of runs per day. When all three hold, Power Automate is the cheapest, fastest way to ship. When any one is not true, the picture changes.
If your systems are largely non-Microsoft (Google Workspace, Salesforce, Xero, NetSuite, custom tools), the strength evaporates and per-flow licensing becomes painful. If the workflow involves real backend logic (five-plus conditional branches, loops, complex transformations, real error handling), the designer becomes a maintenance nightmare. At scale, premium connectors at £12 to £40 per user per month plus per-flow run charges add up. For anything time-critical, the 30 to 120 second per-step latency is too slow.
The advertised price is misleading. Standard connectors are included in most Microsoft 365 Business and Enterprise licences. Premium connectors (Salesforce, SAP, custom HTTP, on-prem gateways) require Power Automate Premium at ~£12/user/month, or a per-flow plan at ~£80/flow/month. A Premium licence for 50 staff is £600/month before per-flow add-ons. A mid-market firm we audited last year was paying £24,000/year in Power Automate licensing for a dozen workflows; the same in custom code would have been £8,000 build plus £30/month hosting.
UK day rates for a Power Automate consultant range from £600 (freelance) to £1,400 (mid-market consultancy) to £2,200+ (specialist Microsoft partner). Day-rate engagements typically run 5 to 20 days for a focused build. Fixed-quote engagements with us are £3,000 to £15,000 for a focused Power Automate workflow build (3 to 6 weeks), £15,000 to £40,000 for a broader automation project that may or may not use Power Automate depending on what the scoping concludes.
No. For roughly half the projects pitched as Power Automate builds, we would recommend custom-code alternatives because the licensing maths or workflow complexity does not fit. For around a quarter, Power Automate is exactly right. The remaining quarter are hybrid: Power Automate for Microsoft-anchored steps, custom service for heavy logic. We tell you which category your project falls into in scoping.
Yes. The Power Platform is a stack: Power Automate for workflows, Power Apps for canvas and model-driven apps, Power BI for reporting, Dataverse for data, Copilot Studio for AI assistants. Each has a sweet spot where it is the right answer, and a zone where custom code is faster, cheaper and more maintainable. We will be honest about which is which.
Tell us the workflow you want to automate and the systems involved. We will tell you honestly whether Power Automate is the right tool, the right hybrid, or whether custom code wins. No charge for the conversation.
Email oliver@digitalsignet.com