What Dynamics 365 partners should do when project work dries up

From access to accountability: a four-quarter reset for Microsoft Business Applications partners facing slower demand, margin pressure and AI-driven delivery compression.

Key takeaway

Dynamics 365 and Power Platform demand is not disappearing, but generic implementation margin is being repriced.
Partner margin moves towards workflow accountability, governance, vertical process memory, data quality, adoption and managed value.
The next four quarters should be used to specialise, productise, prove value, and cut weak-fit work.

I recommend reading Dana Willmer's Partner Economics article first: From Access to Accountability: The Year the Channel's Margin Was Repriced.

The line that should stop every Microsoft partner leader is simple: "the channel is being relocated".

That is a better framing than panic.

The Microsoft Dynamics 365 and Power Platform market is not disappearing. But the old partner margin model is being repriced. Access to product knowledge, configuration skill, implementation labour, generic advisory and time-and-materials delivery are less scarce than they used to be. Microsoft is absorbing more capability into the platform, and Microsoft Foundry makes that platform shift more explicit: AI apps and agents can now be built, grounded, governed, routed, observed and connected to business systems through a Microsoft-native enterprise AI layer. Customers are building more themselves. AI is compressing the effort needed to produce analysis, documentation, code, tests, training material and support responses. Buyers are also more sceptical about large transformation programmes that cannot prove value quickly.

If you run a Dynamics 365 or Power Platform practice and you are already seeing a downturn, treat it as a business model signal, not merely a sales problem.

The Partner Economics forecast argues that the durable layer is moving to "workflow ownership, vertical depth, outcome-linked economics". For a Microsoft Business Applications partner, that is the whole strategic question for the next four quarters.

Do you sell access to Microsoft capability, or do you own a customer workflow consequence?

The first-principles question#

The first-principles question is not:

How do we sell more Dynamics 365 and Power Platform projects?

The better question is:

What business uncertainty does the customer still need us to remove?

A customer does not wake up wanting Dataverse tables, a model-driven app, a Power Automate flow, a Copilot Studio agent, a Microsoft Foundry agent pattern, a Dynamics 365 Sales configuration, a Customer Service queue, a Business Central extension or a release pipeline.

They want fewer process failures, better decisions, lower operating cost, faster response, less compliance risk, clearer accountability and a system their people actually use.

For years, partners could translate that uncertainty into implementation work. Discovery, configuration, customisation, migration, integration, reporting, testing, training and post-go-live support became the economic engine. Many practices then priced that engine as labour: days, hours, retainers, blended teams and change requests.

That model is now exposed because the inputs have changed.

Old scarcity versus new strategic pressure for Microsoft Business Applications partners.

Old scarce input	What is happening now	Strategic implication
Product access	Customers already own more Microsoft cloud capability.	Resale and licence-led influence are weaker differentiators.
Product knowledge	Microsoft Learn, community content and AI assistants reduce knowledge scarcity.	Knowing the feature is not enough.
Configuration labour	Low-code, templates, copilots and AI-assisted development compress effort.	Billable task volume falls.
Generic advice	AI makes general guidance cheap and fast.	Advice must be contextual, accountable and tied to a decision.
Delivery capacity	Offshore, nearshore and AI-assisted delivery expand supply.	Generic capacity competes on price.
Customer trust	Still scarce.	Trust becomes more valuable when systems act on behalf of users.
Workflow accountability	Still scarce.	This is where margin can move.

The practice should not ask how to defend old scarcity. It should ask where new scarcity sits.

In Microsoft Business Applications, new scarcity sits in five places:

Workflow accountability. Someone must be responsible when the process, app, data, automation or agent produces an operational consequence.
Data and context quality. Copilot and agents are only useful when the business data, security model and process context are reliable.
Governance and risk. Power Platform sprawl, connector risk, environment design, DLP, identity, model choice, agent permissions, tool access, observability and auditability are board-level concerns once automation scales.
Adoption and behaviour change. Go-live is not value. Usage, habit change and process compliance are closer to value.
Vertical process memory. The partner that understands the real exceptions, rules, handoffs and failure modes of a specific industry becomes harder to replace.

Why cloud growth can still feel like downturn#

The uncomfortable part is that Microsoft Business Applications can grow while traditional partner practices shrink.

Microsoft's FY26 Q3 results, released on 29 April 2026, showed Microsoft Cloud revenue of US$54.5 billion, up 29%. Dynamics 365 revenue increased 22%, and Azure and other cloud services revenue increased 40%. Microsoft also said its AI business had surpassed a US$37 billion annual revenue run rate.

So the platform is not weak.

The issue is that spend is reallocating.

The Partner Economics forecast frames this as repricing rather than contraction. Its argument is that budgets are not simply vanishing. They are moving away from generic seats and generic labour towards AI capability and the layer that converts capability into business results. Gartner's 2026 IT spending forecasts, as reported in April 2026, point in the same direction: worldwide IT spending was revised up to about US$6.31 trillion, with AI infrastructure and software driving much of the increase.

That distinction matters.

A partner can experience a downturn when its offer sits on the wrong side of the reallocation.

The systems map: the utilisation trap#

A Dynamics 365 and Power Platform practice is a system, not a set of isolated departments.

When demand slows, the instinct is to protect utilisation. That response is understandable, but it can create a reinforcing loop that weakens the practice.

How utilisation-first decisions can weaken a partner practice.

Utilisation-first loop	Result
Fewer large projects	Revenue confidence falls.
Revenue confidence falls	Leaders chase any available work.
Leaders chase any available work	The firm discounts and accepts weak-fit clients.
Weak-fit work increases	Senior staff get pulled into rescue mode.
Rescue mode rises	Delivery margin and reference quality fall.
Margin falls	Less money is available for IP, skilling, content and productisation.
Less reinvestment	The practice looks more generic next quarter.
Generic positioning	Win rates fall again.

This is how capable firms become commodity firms.

The alternative is a different loop.

How outcome-first positioning improves partner economics.

Outcome-first loop	Result
Narrower market problem	The practice becomes easier to understand.
Easier to understand	Content, referrals and Microsoft alignment improve.
Better-fit opportunities	Discovery is faster and proposals are sharper.
Reusable delivery patterns	Gross margin improves.
Better margin	More reinvestment goes into accelerators, training and proof.
Stronger proof	Customers trust the firm with more consequential work.
More consequential work	Recurring revenue and switching costs improve.

The second loop is harder because it requires saying no while the market feels uncertain. But it is the loop I would want to be in.

Where Microsoft Foundry fits#

Microsoft Foundry should not be dropped into this conversation as another product logo. It changes the architecture of a Dynamics 365 and Power Platform partner practice.

The practical architecture now looks more like this:

Where Microsoft Foundry fits in a Dynamics 365 and Power Platform practice.

Layer	Microsoft context	Partner question
System of work	Dynamics 365, Power Platform, Dataverse, Business Central, Customer Service, Field Service, Sales	Which workflow outcome are we willing to own?
Business-facing agent creation	Copilot Studio	Which agent can a business user safely use, extend and adopt?
Enterprise AI engineering and governance	Microsoft Foundry, Foundry Models, Foundry Agent Service, Foundry IQ, Foundry Control Plane	Which models, tools, grounding, evaluations and controls make the workflow dependable?
Data and organisational context	Dataverse, Microsoft Graph, Fabric, Microsoft IQ	What trusted context does the agent need, and who is allowed to use it?
User experience	Microsoft 365 Copilot, Teams, Dynamics 365 apps, custom apps, portals	Where should the work happen for the user?
Operating assurance	Entra, Purview, Defender, DLP, monitoring, evaluation, human review	How do we prove the automation is safe, useful and improving?

This matters because the Partner Economics article argues that customers will increasingly "route each task to the cheapest model that can handle it". Microsoft Foundry is part of that same direction of travel. It gives customers a broad model catalogue across providers, plus model routing, grounding, tools, agent services, observability and governance.

For a Business Applications partner, that is both threat and opportunity.

The threat is obvious: if Foundry, Copilot Studio, templates, connectors and AI-assisted development absorb more of the generic build work, the partner cannot defend margin by saying, "we know how to configure the tool".

The opportunity is better: most customers will not know which workflow should become an agent, which model should be used, what data can be trusted, which action needs human approval, how to test failure modes, how to govern tool access, or how to report value after deployment. Microsoft says Foundry can connect agents across more than 1,400 business-system connections including Dynamics 365, and can govern agents, tools and knowledge sources through a control plane integrated with Entra, Purview and Defender.

That is where the partner margin moves.

Copilot Studio remains highly relevant, particularly where the customer wants to create agents, extend Microsoft 365 Copilot, declare instructions, connect tools and knowledge, and publish across channels. Foundry matters when the conversation becomes more architectural: multi-model selection, cost-aware routing, grounding, evaluation, observability, pro-code agent services, MCP tools, and enterprise controls.

So the offer is not "we implement Copilot Studio" or "we build in Foundry". The offer is:

We help you redesign, govern and operate the workflow where Dynamics 365, Power Platform, Copilot Studio and Microsoft Foundry meet.

That is a much stronger position than being a generic implementer of whichever product name is in the release note this month.

The five structural shifts Microsoft partners need to accept#

1. The product roadmap is moving towards agents, orchestration and AI-assisted work#

The Dynamics 365 2026 release wave 1 plan covers April to September 2026 and includes AI and agentic capabilities across Sales, Customer Service, Contact Center, Field Service, Finance, Supply Chain Management, Customer Insights and Business Central.

The Power Platform 2026 release wave 1 plan is just as direct. It highlights Copilot Studio, multi-agent orchestration, evaluations, Dataverse APIs, MCP servers, Work IQ, agent security, AI-powered governance and visibility into Copilot credit consumption.

Microsoft Foundry reinforces the same direction. Its public positioning is not only model access. It is an enterprise AI platform for building, grounding and governing AI apps and agents at scale, with model choice, agent services, knowledge and tools, observability, trust controls and integration into Copilot Studio.

This changes the customer conversation.

Customers will still need implementation. But the higher-value questions are shifting:

Which workflows should be redesigned before we automate them?
Which agents are safe enough to deploy?
Which agent belongs in Copilot Studio, which belongs in Foundry, and which should not be built yet?
Which model is good enough for the task, and when does cost-aware routing matter?
Which tasks should remain human reviewed?
Which data can be trusted?
Which tools and business-system actions should an agent be allowed to call?
Which actions should be logged, escalated or blocked?
What does good agent performance look like?
Who owns the outcome when automation goes wrong?
How do we measure business value after go-live?

Those are excellent questions for a strong partner. They are poor questions for a generic build shop.

2. Microsoft partner measurement is moving towards customer growth and usage#

Microsoft's Partner Capability Score measures partners across performance, skilling and customer success. The scoring framework includes net customer adds, certifications, usage growth and deployments. Business Applications is one of the six solution areas in the framework.

Whether a partner cares about the badge or not, the signal is clear.

The ecosystem rewards partners that add customers, build real capability and drive deployed usage.

A practice that celebrates project completion but does not track usage, active adoption and expansion is optimising for an old game.

3. Marketplace and private-offer motions are becoming part of the buying system#

Microsoft has been consolidating its business app and AI buying experiences into Microsoft Marketplace. In September 2025, Microsoft announced that Azure Marketplace and Microsoft AppSource were unified into Microsoft Marketplace, a single destination for cloud solutions, AI apps and agents.

For a partner leader, the practical implication is this: your offer needs to become more product-like, even when the delivery includes services.

That does not mean every partner must become a pure ISV.

It does mean your offer needs a clear name, target buyer, trigger event, scope, price logic, delivery pattern, proof point and renewal path.

"Talk to us about Power Platform" is not an offer.

"Six-week Power Platform Governance and Agent Readiness Reset for organisations with uncontrolled maker growth" is closer.

4. AI adoption is ahead of operating-model redesign#

Microsoft's 2026 Work Trend Index found that Microsoft 365 Copilot and agent use is growing, but organisational readiness is the constraint. Microsoft surveyed 20,000 AI users across 10 markets and found that organisational factors such as culture, manager support and talent practices accounted for more than twice the reported AI impact of individual behaviour alone.

This is where partners should see opportunity.

The customer problem is not only tool deployment. It is operating-model redesign.

A Dynamics 365 and Power Platform partner can help customers decide how work should be split across people, apps, automations and agents. That is much more valuable than installing another tool and leaving the customer to work out the consequences.

5. Services firms are being valued differently#

In June 2026, Accenture's own Q3 FY26 results showed full-year revenue growth guidance narrowed to 3% to 4% in local currency, with new bookings down year on year.

A large global integrator is not the same business as a mid-market Microsoft partner. But the signal still matters.

The market is scrutinising services businesses that depend on large volumes of human labour, weak differentiation and billable effort.

That scrutiny will move down-market.

The new practice thesis#

If I were still running a Dynamics 365 and Power Platform partner business, I would make one thesis explicit to the entire leadership team:

We are moving from implementation labour to accountable workflow systems on the Microsoft cloud.

That would change what we sell, how we price, how we staff, how we deliver and how we measure success.

The practice would need four layers.

The four layers of an accountable workflow systems practice.

Layer	What it means	Why it matters
Outcome offers	Named packages tied to business problems.	Easier for buyers and Microsoft sellers to understand.
Reusable IP	Templates, reference architectures, agents, assessment models, scripts and playbooks.	Improves margin and differentiation.
Managed value services	Recurring governance, optimisation, adoption, agent operations and value reporting.	Reduces dependence on project starts.
Evidence ledger	Proof of cost reduction, usage growth, error reduction, cycle-time improvement and risk reduction.	Supports premium pricing and renewals.

The goal is not to abandon implementation. The goal is to stop selling implementation as the main product.

Implementation becomes one component of a broader outcome system.

Foundry makes this more urgent. If the customer can access models, agent services, grounding, tools and control-plane capability directly from the Microsoft stack, the partner has to bring the missing system: vertical workflow judgement, data quality, adoption discipline, governance, measurement and accountability.

What I would change in the offer portfolio#

I would stop leading with broad capability statements and build a small number of productised offers around urgent customer problems.

Offer 1: Power Platform Governance and Agent Readiness Reset#

Trigger: The customer has uncontrolled maker growth, environment sprawl, unclear DLP, rising connector risk, early Copilot Studio experimentation, or confusion about where Microsoft Foundry belongs.

Outcome: A practical governance model for apps, flows, environments, connectors, Dataverse, Copilot Studio agents, Foundry-backed agents, security, ownership and support.

Deliverables: Current-state assessment, risk register, environment strategy, DLP recommendations, agent readiness model, Foundry and Copilot Studio decision guide, maker operating model and 90-day governance backlog.

Offer 2: Dynamics 365 Service Outcome Sprint#

Trigger: Customer service leaders face rising case volume, inconsistent routing, poor knowledge quality, slow resolution or contact centre pressure.

Outcome: A redesigned service workflow using Dynamics 365 Customer Service, Contact Center, Power Platform and agent-assisted triage where appropriate.

Deliverables: Process map, case taxonomy, knowledge health review, automation backlog, agent candidate map, KPI baseline and pilot plan.

Offer 3: Copilot Studio and Microsoft Foundry Workflow Agent Lab#

Trigger: Executives want agents but lack a safe way to choose, build and prove use cases across Copilot Studio, Dynamics 365, Power Platform and Microsoft Foundry.

Outcome: One governed workflow agent with human review, logging, escalation, evaluation criteria, model-selection logic and a measurable business case.

Deliverables: Use-case selection, Copilot Studio versus Foundry decision, agent design, model and routing recommendation, data and permission review, tool and MCP review, test harness, failure-mode review, adoption plan and value measurement.

Offer 4: Dataverse and Business Data Foundation for Agents#

Trigger: The customer wants Copilot, Copilot Studio agents or Foundry-backed agents, but data quality, security roles, duplicate records, ownership and business definitions are weak.

Outcome: A trusted data and security foundation for future automation.

Deliverables: Data model review, Dataverse design improvements, security-role review, data-quality backlog, integration review and roadmap.

Offer 5: Managed Business Applications Value Office#

Trigger: The customer has gone live but cannot prove adoption, usage growth or operating value.

Outcome: Recurring management of adoption, governance, backlog prioritisation, Copilot Studio and Foundry agent performance, usage analytics and quarterly value reviews.

Deliverables: Monthly optimisation, executive value report, adoption telemetry, governance review, release-wave impact review and continuous improvement backlog.

These offers are not magic. They are concrete enough to sell, deliver, measure and improve.

How pricing needs to change#

The old pricing model was simple: estimate effort, apply rate card, add contingency and defend scope.

That will not disappear. But it should become less dominant.

I would move to a four-part commercial model.

Commercial motions for a less labour-dependent partner model.

Commercial motion	Use it for	Pricing logic
Fixed-fee diagnostic	Governance reset, readiness assessment, value discovery.	Price for decision value, not only effort.
Fixed-scope sprint	Pilot, workflow redesign, agent lab, data foundation.	Price for speed, clarity and defined outputs.
Managed value service	Governance, adoption, optimisation, agent operations.	Monthly recurring fee tied to scope and operating cadence.
Outcome-linked component	Cost reduction, error reduction, cycle-time reduction, adoption uplift.	Fee-at-risk or upside component where measurement is credible.

Be careful with outcome pricing. Do not take risk you cannot control. But start building the muscle now.

The point is not to gamble on customer outcomes. The point is to show that the practice is willing to be measured against value, not only activity.

The next four quarters#

As of 29 June 2026, I would treat the next four quarters as:

Quarter 1: July to September 2026.
Quarter 2: October to December 2026.
Quarter 3: January to March 2027.
Quarter 4: April to June 2027.

Quarter 1: Stabilise, choose and expose the truth#

The goal for Quarter 1 is to stop guessing.

Quarter 1 leadership actions#

Segment revenue and margin by work type. Separate licence margin, net-new implementation, enhancement work, support, managed services, training, adoption, governance, Power Platform, Dynamics 365, Copilot Studio, Microsoft Foundry and AI-related work.
Classify the pipeline. Label every opportunity as legacy implementation, modernisation, governance, data foundation, agent, adoption, managed service or weak-fit custom work.
Choose two market lanes. Do not pick seven. Good candidates are service operations, field service operations, regulated Power Platform governance, Business Central modernisation, customer insights and data readiness, or governed workflow agents using Copilot Studio and Microsoft Foundry.
Freeze weak-fit discounting. Protect cash, but stop creating future losses just to keep utilisation high.
Create the new offer catalogue. Build two or three named offers with a clear buyer, problem, scope, timeline, price range and proof requirement.
Re-baseline delivery cost with AI. Measure where AI can reduce internal effort in documentation, testing, code review, release notes, support triage, proposal drafting and training content. Then decide how that efficiency will improve margin, not just create more unbilled waste.
Brief the whole team. Do not hide the shift. Consultants already know the work is changing. Give them a plan.

Quarter 1 measures#

Quarter 1 stabilisation measures.

Measure	Target by 30 September 2026
Revenue map	100% of revenue classified by work type and margin.
Pipeline map	100% of active opportunities classified by new categories.
Focus lanes	Two chosen, with one-page market thesis each.
Productised offers	At least three named offers drafted.
AI internal use	Five delivery activities measured for time reduction.
Stop list	Named weak-fit work types no longer actively pursued.
Cash view	13-week cash forecast reviewed weekly.

Quarter 1 decision trigger#

If more than 60% of future revenue depends on generic project labour and the pipeline is falling, assume the model is already at risk.

Quarter 2: Productise and prove#

The goal for Quarter 2 is to convert positioning into evidence.

Quarter 2 leadership actions#

Sell two lighthouse engagements. Discount scope if needed, but do not discount learning. Each lighthouse must produce a measurable case study, reusable asset and sales proof.
Create reusable delivery assets. Build templates, workshop scripts, governance checklists, Dataverse patterns, Copilot Studio and Foundry decision trees, model-routing guidance, agent evaluation rubrics, demo data, proposal language and value-review packs.
Reshape pre-sales. Stop writing bespoke proposals from scratch. Use offer-specific proposal patterns and qualification questions.
Train sales on business problems. Sellers need to lead with service cost, cycle time, revenue leakage, compliance risk, data quality, operational control and adoption, not only product features.
Align skilling to the Microsoft scorecard. Use the Partner Capability Score logic as a management signal: customer adds, skilling, usage growth and deployments.
Prepare the marketplace path. Decide which offers should become professional services, transactable IP, private offers or co-sell motions.

Quarter 2 measures#

Quarter 2 productisation measures.

Measure	Target by 31 December 2026
Lighthouse wins	Two signed or in delivery.
Reusable assets	Minimum ten assets created and used.
Offer win rate	Tracked separately from generic work.
Proposal effort	Reduced by at least 30% for named offers.
Delivery margin	Baseline established by offer.
Customer value metrics	Defined for every new project.
Microsoft alignment	Target accounts and offers mapped to partner motions.

Quarter 2 decision trigger#

If named offers are not generating qualified conversations by the end of Quarter 2, either the market lane is wrong, the offer is unclear, or the sales team is still selling capacity.

Quarter 3: Shift the operating model#

The goal for Quarter 3 is to make the new model operational, not just presentational.

Quarter 3 leadership actions#

Move top talent into repeatable offers. Do not leave your best people trapped in one-off custom delivery.
Launch a managed value service. Every project should have a post-go-live adoption, governance or optimisation path.
Change delivery reviews. Review adoption, usage, risk, reusable asset creation and customer value, not only budget and milestone status.
Create an agent operations capability. Build the ability to monitor, evaluate, improve and govern workflow agents after deployment across Copilot Studio, Microsoft Foundry and Dynamics 365 work surfaces.
Redesign roles. You need fewer pure task-takers and more people who can lead process design, data quality, governance, adoption, architecture, agent evaluation and executive conversations.
Tie incentives to new revenue. Reward recurring revenue, reusable IP, proof points, customer value and margin improvement, not only utilisation.

Quarter 3 measures#

Quarter 3 operating model measures.

Measure	Target by 31 March 2027
Recurring revenue	Clear monthly run rate from managed services.
Post-go-live attach	Offered on every relevant project.
Reusable delivery	Assets used in more than 50% of new projects.
Agent operations	Defined service model and first customer pilot.
Utilisation mix	Less dependence on generic build work.
Case studies	Two evidence-backed stories ready for market.
Staff transition	Role map completed for every consultant.

Quarter 3 decision trigger#

If recurring value services cannot attach to implementation work, the practice is still selling go-live rather than business outcomes.

Quarter 4: Scale, cut or restructure#

The goal for Quarter 4 is to make hard decisions with evidence.

Quarter 4 leadership actions#

Double down on the strongest lane. If one segment is responding, focus harder. Do not dilute the signal.
Turn the strongest assets into IP. This may be a managed service, assessment framework, industry accelerator, Copilot Studio agent pattern, Foundry agent pattern, Dataverse model, marketplace offer or value measurement method.
Exit weak offers. If a work type has poor margin, weak strategic learning and no recurring path, stop selling it.
Decide build, partner, acquire or merge. If you lack data, Fabric, security, adoption, marketplace, industry or agent capability, make a structural decision. Do not pretend a training course will solve every capability gap.
Rework compensation. Sales and delivery incentives must reward the business you want, not the business you are leaving.
Review capital options. If the practice lacks scale, cash or capability to shift, consider a merger, acquisition, specialised partnership or deliberate retreat to a narrower niche.

Quarter 4 measures#

Quarter 4 scaling and restructuring measures.

Measure	Target by 30 June 2027
Strongest lane	Clear from revenue, margin and pipeline data.
Weak work	Removed, repriced or redesigned.
Reusable IP	At least one offer-level asset ready to scale.
Recurring revenue	Material and growing.
Delivery margin	Better than 2026 baseline.
Sales narrative	Clear enough for clients, Microsoft and staff.
Organisation shape	Aligned to the new model.

Quarter 4 decision trigger#

If the practice is still mainly selling generic implementation capacity by June 2027, assume the market has already repriced you.

What to stop, start and continue#

What to stop, start and continue during the partner practice reset.

Stop	Start	Continue
Selling generic Dynamics 365 or Power Platform capacity.	Selling named business outcome offers.	Deep Microsoft platform expertise.
Taking low-margin customisation to keep people busy.	Creating reusable assets in every project.	Strong delivery governance.
Treating go-live as the finish line.	Selling managed adoption, governance and optimisation.	Customer relationships built on trust.
Pricing everything by labour.	Adding fixed-fee, recurring and outcome-linked pricing.	Sensible scope control.
Marketing broad capability lists.	Publishing point-of-view content around industry and process pain.	Community presence and Microsoft relationships.
Leaving attribution to chance.	Managing CPOR, PAL, DPOR and CSP attribution as commercial discipline.	Certification and skilling discipline.
Treating agents as demos.	Building governed agents tied to real workflows across Copilot Studio, Microsoft Foundry and Dynamics 365.	Human judgement in high-risk work.

The monthly leadership scorecard#

The leadership team should review this every month.

Monthly leadership scorecard for the new partner model.

Category	Measure	Why it matters
Market focus	Qualified pipeline by chosen lane	Shows whether focus is working.
Offer quality	Win rate by named offer	Shows whether buyers understand the offer.
Margin	Gross margin by offer	Shows whether productisation is real.
Customer success	Usage growth and deployment growth	Aligns with Microsoft partner direction.
Recurring base	Monthly recurring service revenue	Reduces project-start dependency.
Delivery leverage	Reusable asset usage	Shows whether each project improves the next one.
Capability	Governance, data, agent, adoption and architecture skills	Shows readiness for the new work.
Attribution	CPOR, PAL, DPOR and CSP mapping	Protects ecosystem recognition and influence.
Adoption	Active users, process completion and value review outcomes	Moves the practice from go-live to business value.
Risk	Escalations, overruns, support debt and agent failures	Shows whether delivery complexity is under control.

Scenario planning#

Base case#

Project demand remains uneven. Large transformation programmes are harder to win. Smaller modernisation, governance, AI readiness, Copilot Studio, Microsoft Foundry, data foundation and managed adoption work grows.

Move: stay lean, specialise, protect margin and shift 25% to 35% of revenue towards recurring or repeatable services within 12 months.

Optimistic case#

AI and agent demand converts faster than expected. Microsoft sellers need credible partners who can turn Copilot Studio, Microsoft Foundry and agent interest into governed Business Applications outcomes. Marketplace and private-offer motions improve deal flow.

Move: invest in IP, hire selectively and build a repeatable industry accelerator.

Pessimistic case#

The project market contracts further. Buyers delay discretionary work. Competitors discount heavily. Staff utilisation falls.

Move: preserve cash, cut weak-fit work, specialise harder, partner for missing capability and consider merger or acquisition if the practice lacks the scale to build the new model.

Take Action Now#

A downturn exposes the business model you actually have.

If the model is "we sell capable people by the day", AI, platform maturity, buyer caution and global competition will keep squeezing it.

If the model is "we help customers redesign and operate business workflows on Microsoft technology", the opportunity is still there.

The next four quarters should not be about waiting for the old market to return. They should be about building the practice the new market is already asking for.

The winners will not be the partners with the longest capability list.

They will be the partners who can say, with evidence:

We know this workflow. We know the failure points. We can get you to value faster, safer and with less waste. And we will stay accountable after go-live.

That is still worth paying for.

Mark Smith is Principal AI Strategist at Cloverbase. To discuss this article or work with me, contact me at Cloverbase.