The way most Perth business owners go about choosing an MSP is largely the wrong way around. They shortlist providers from a Google search, sit through three pitch meetings, look at the headline prices, pick the one that felt right, and sign a three-year contract. A year in, when service delivery does not match the sales conversation, they discover the contract has no clean way out.
This is the comparison every business owner doing serious due diligence should run before committing. Not just “what does it cost,” but “what kind of MSP is this, what does their service delivery actually look like under stress, and what are the contractual realities I am signing up for.”
We sit on both sides of this conversation. As an MSP, we lose deals to competitors regularly. We also pick up clients from competitors regularly when their previous provider stops delivering. The pattern is depressingly consistent: the businesses with the most painful switches chose their original provider on price and personality rather than substance.
Here is what serious due diligence on a Perth or Australian MSP actually looks like in 2026.
Most people shopping for an MSP do not realise the market splits into four meaningfully different categories. Mismatching your business to the wrong category is the single biggest source of regret.
Consolidator MSPs. Private-equity-backed, growth-by-acquisition strategy. They have bought multiple Perth and Australian providers over the last three to five years and rolled them into a single brand. Standardised processes, centralised NOC and SOC, predictable but impersonal. Usually 200 to 1000+ staff Australia-wide. Pricing tends to be premium. Service quality varies meaningfully by which legacy team you end up with after acquisition, and which way that team is heading as the parent integrates them.
Niche specialists. Focused on a specific vertical, commonly legal, healthcare, mining, NFP, or construction. Deep domain knowledge within their sector. Premium pricing inside their niche. Limited adaptability to anything adjacent. If you are in their core sector, the fit is strong. If you are one step outside, it gets awkward fast.
Modern AI-native MSPs. Smaller (20 to 100 staff), usually founded or rebuilt post-2020, technical depth above average, faster adoption of modern tooling and security. Higher engineer-to-client ratio. More direct access to senior technical people. Lower turnover risk because they are not on PE rollup paths and the founders are still running the business. Pricing varies widely.
Traditional family-run MSPs. Long-standing relationships, often serving the same clients for 10 to 20 years. Stable, relationship-driven, but variable on technical depth. Some are excellent and continue to invest. Some are coasting on legacy goodwill. Heavily owner-dependent. If the founding partner exits, the business often degrades within 24 months.
Each has trade-offs. The mismatch we see most often is a 30-staff business signing with a consolidator because the sales pitch was polished, or a 200-staff business signing with a family-run shop because the relationship felt comfortable. Neither survives the first year of pressure.
For Perth businesses 10 to 50 staff: AI-native or traditional family-run, depending on whether you value innovation pace or continuity. The AI-native option is increasingly the right call as compliance, security, and AI governance pressure builds.
For 50 to 200 staff: AI-native is usually the sweet spot. Technical depth without consolidator overhead, modern stack, direct access to senior engineers.
For 200+ staff: Either a consolidator with a strong local team, or a high-end AI-native that can demonstrate the scale. Niche specialist if your sector is one of theirs.
For specialised verticals (defence supply chain, regulated finance, healthcare with significant patient data): niche specialist if a strong one exists in your space. Otherwise an AI-native with demonstrated relevant experience.
This is the section that did not exist in the 2023 version of this conversation. It is now one of the most important questions to ask. Most buyers do not know what to ask about it. Most MSPs do not have a good answer.
AI capability inside an MSP is not “do you have Copilot licences.” Every MSP has Copilot licences. The question is whether AI has been built into the way the business actually delivers service, or whether it sits on the marketing page as a buzzword.
What real AI capability looks like operationally:
The questions that separate real AI capability from marketing fluff:
“Walk me through one operational workflow where AI has materially changed how your team works in the last six months.” Specific answer with details means real capability. Vague answer or pivoting to client AI services means it is not there yet internally.
“What percentage of tickets are currently touched by AI before a human engineer sees them?” The mature operators are at 60 to 90 per cent on certain ticket categories. The aspirational ones are at zero per cent and planning to start.
“Show me an AI agent you have built and explain what it does.” If they can show you one, the capability is real. If they cannot, they are using off-the-shelf AI features at best.
“What is your AI governance position when deploying tools into client environments?” The answer should reference frameworks (ISO 42001, NIST AI RMF, Australia’s Guidance for AI Adoption), not just “we follow best practice.”
The MSP buying decision in 2026 is increasingly the AI buying decision. The MSPs that are not investing in this are the ones that will be unable to compete on cost or service quality within three years. Choosing one of them is choosing a provider whose service-delivery economics are going to deteriorate while their competitors improve.
Standard “10 questions to ask your MSP” lists cover uptime, response time, and certifications. These are floor-level. Every MSP that survived past year three has reasonable answers. The questions that separate excellent from average:
The contract is where the sales pitch meets reality. The things to read carefully:
Initial term length. Twelve months is reasonable. Twenty-four is borderline. Thirty-six should require a real discount and a real exit clause. The 36-month locks are almost always priced at the same monthly rate as the 12-month, which means you are giving up flexibility for nothing.
Auto-renewal terms. The contracts that auto-renew with 90 days notice required have you locked in. Push for 30 days maximum. If the response is “we cannot change that,” it tells you something about how they think about the relationship.
Out-of-scope billing. What gets billed extra? Project work, after-hours, weekends, public holidays, vendor coordination, hardware procurement margins? Vague exclusions become surprise bills. Get a written schedule of what is in scope and what is out.
Data and asset ownership. Who owns the documentation, scripts, automations, and tooling configurations built during the engagement? You should. Make sure it is explicit. The MSPs that resist this clause are signalling something about how they handle exits.
Performance SLAs. Most SLAs measure response time (acknowledging a ticket) not resolution time (fixing the problem). Response-only SLAs are weak. Resolution-based SLAs with tiered severity are stronger. The strongest add automatic credits if SLAs are missed repeatedly.
Price escalation. CPI-linked is reasonable. CPI+2% or CPI+3% is the provider hedging at your expense. CPI+5% is gouging. Push back.
Termination for convenience. Can you exit without cause with 30 to 90 days notice and a defined wind-down? If not, you are signing up for a relationship that has no honourable exit. Walk.
“Give me three references” produces three carefully-curated success stories. Useless. The references you want:
References whose engagement is at least two years old. New clients do not yet know what they have signed up for. The honeymoon period for an MSP relationship runs six to twelve months. Year two is where the real picture emerges.
References from businesses similar to yours in size and complexity. A 200-staff reference does not tell you how the MSP serves 30-staff businesses. A reference from a different sector does not tell you how they handle yours.
Ask references the questions about disappointment, not delight. “What surprised you negatively in year two?” “When did service delivery dip and how did they handle it?” “Have you considered switching and what stopped you?” The answers reveal more than any glowing testimonial.
Ask the MSP for a reference of a client they have lost, and what happened. Most will not provide. The ones who can show maturity in handling that conversation are signalling something useful about their self-awareness.
Patterns we see in sales processes that consistently lead to regret 12 months later:
Vague answers about who delivers the work. “Our team” is not an answer. Get names, get bios, get continuity assurances.
No technical engineer in the pitch. If only commercial people are presenting, technical due diligence has been deferred to after the contract is signed. By then it is too late.
Over-confident claims about uptime, security, and response. “99.99% uptime guaranteed” is marketing. Real engineers do not talk like that because they know what happens when telco infrastructure fails or a critical patch breaks production.
No references provided proactively. Reluctant references usually mean no references that would stand up to honest questioning.
Pricing that is significantly below the market. Either they are losing money to win the deal (unsustainable, you will see service degrade by month six), or they are cutting corners somewhere that will eventually cost you more than the saving.
Pressure to sign quickly. “Special pricing if you commit this week.” Walk. Any legitimate MSP will hold pricing for 30 days minimum.
No defined onboarding methodology. If they cannot show you what the first 90 days look like in detail, they do not have a methodology. You will be onboarded ad-hoc by whoever is available, which is the single biggest predictor of a rocky engagement start.
Most MSPs will not propose this. The good ones will negotiate one in. The structure:
A 90-day pilot period with defined success criteria agreed upfront. Specific metrics: response times achieved, tickets resolved, projects delivered, security posture improvements demonstrated.
An exit clause if criteria are not met, with refund of paid fees or transition to a short-term arrangement while you find an alternative.
A quarterly business review at the end of the trial period that decides whether to convert to a long-term contract.
The pushback you will get: “we cannot structure that, the up-front investment is too high.” Sometimes genuinely true for very large engagements. For typical SMB engagements, the resistance reveals more about the MSP’s confidence than the engagement economics.
A 30-staff Perth business signs a 36-month contract with a consolidator because the sales pitch was polished. The service delivery from year one is impersonal and slow. The contract makes exit expensive. They renew because switching is hard, then quietly degrade until a breach, an outage, or a key person leaving forces the issue. The business ends up where it should have started: looking for an AI-native or mid-market provider with technical depth and a healthier engagement structure.
The way to avoid this pattern is to slow down at the front end. The MSP decision is a 5 to 10-year relationship in most cases. Spending three or four weeks doing real due diligence before signing saves years of recovery time later.
Build a shortlist of three to five providers across at least two of the four categories. Do not shortlist only consolidators or only AI-natives. Spread the exposure.
Run each through the questions above. The answers will eliminate one or two providers before you get to pricing.
Compare contracts side by side, not just pricing. The differences in clauses tell you more about how each MSP thinks about the relationship than any sales presentation will.
Negotiate the trial structure. If two providers will not accept any kind of trial or shorter-term commitment, and one will, that signal is informative.
Check references properly, not ceremonially. The reference call should be 30 to 45 minutes of real questions, not a five-minute confirmation.
Then commit. The 5 to 10-year relationship deserves the three to four weeks of due diligence at the front.
We help Perth businesses evaluate MSP shortlists independently, including ones where we are not one of the candidates. Two-week engagement, written assessment of each provider against the framework above, contract clause review, reference question lists, and a comparative recommendation. No obligation to use Epic IT as your MSP.
