A good systems integrator is one of the most valuable partners you'll have. The right firm brings people who've configured the platform a dozen times and can move faster than any team you could assemble in-house. I've worked alongside SIs that earned every dollar, and a few that were paid a great deal more than they were worth.
The brand on the statement of work mattered much less than how the relationship was set up, starting with something nobody says plainly at kickoff: an SI gets paid to build. Efficiency, simplicity and "actually, you don't need that" aren't what its model rewards.
Most people on the ground are conscientious and want a successful go-live as much as you do. The incentives work on them anyway, week after week. Over a two-year program the gap between what serves the engagement and what serves your outcome shows up as a hundred small decisions, each one defensible on its own.
1 The deal pays the SI to build more than you need
| What the SI is paid for | What you actually need |
|---|---|
| Effort and duration, whether the contract is time and materials or milestone based | The smallest configuration that runs your business |
| Building what's asked for, at the pace the engagement can sustain | Someone saying "you don't need that" and meaning it |
| Protecting the delivery date the revenue plan is built on | A date that moves when the evidence says it should |
| Keeping named experts billable across the portfolio, rather than on your program | The people who pitched you, in the seats, by name |
| Scope questions resolved as change orders | Scope questions resolved against what was already agreed |
Nobody on the SI side has to behave badly for this to happen. The structure of the deal does it, and programs drift when everyone pretends otherwise.
So you need someone in the room whose only incentive is your outcome. Sometimes that's you. Often it should be someone with no stake in how many hours the program burns.
Hire the integrator for what it can build, and keep someone independent to judge whether it should be built at all.
2 Score the integrator on what predicts delivery
The pitch is usually excellent and the named experts are impressive. Then the contract signs, those experts rotate to the next sale, and you get a team you've never met. Score the things that predict delivery instead, and weight them.
Weight, out of 100 points
- 25
Relevant, recent, comparable experience
Your platform, your scale, your industry, inside three years. Old success on a different product counts as a reference at most.
- 25
The real delivery team, named in writing
Ask to meet the people who will staff the program. If the firm can't or won't name them, that's your answer.
- 20
Willingness to be measured
Objective acceptance criteria and holdbacks, agreed without flinching. Reluctance here is the most honest signal you'll get.
- 20
References you call and press
Don't take the curated list at face value. Ask what slipped, what the change-order experience was, and whether they'd hire the firm again knowing what they know now.
- 10
Cultural fit under disagreement
You'll be in hard rooms with these people for two years. How they handle disagreement during the sale is a preview.
The A-team bait and switch. It's so common it's practically a business model. You defend against it on paper: named personnel in the statement of work, with a substitution clause that requires your written consent and equivalent seniority.
3 The SOW is where you keep or give away your leverage
Most statements of work describe hours, roles and phases and stop short of committing to a result. That's comfortable for the integrator and expensive for you.
Statement of WorkBuyer markup
1.The Integrator will deliver workshops held, documents produced and support providedoutcomes, each with an objective definition of done the Client can verify.
Most SOWs describe activity: hours, roles, phases.
2.A deliverable is complete when the Integrator judges it completeit meets acceptance criteria written before the work starts, rather than negotiated at handover.
"Complete" as a judgment call becomes an argument.
3.Scope is written broadly, and changes are handled as they arisewritten tightly, paired with disciplined change control.
A vague scope is an open invitation to bill the gaps.
4.Payment is due on the milestone datesdemonstrated results, with holdbacks released on evidence.
Milestone dates arrive whether or not the work does.
5.Key personnel may be substituted at the Integrator's discretionare named in this SOW and may be substituted only with the Client's written consent.
Most SOWs leave key people unnamed, or name them with free substitution.
6.The Client will have access to the plan file and the underlying numbers.
So access doesn't depend on the relationship.
Of the six, the holdback in clause 4 does the most work once things get tense.
Money still in your hands is the only part of the contract that reliably commands attention when a program is under stress.
What to have in writing before the SOW signs
- The delivery team named, with substitution only by your written consent
- Every deliverable written as an outcome with a definition of done
- Acceptance criteria agreed before the work starts
- Payment on demonstrated results, with holdbacks released on evidence
- Your access to the plan file and the underlying numbers
4 Governance should make candor routine
Governance stops working the moment it becomes a status meeting where everyone reports green. Done right, it's the machinery that surfaces problems while they're still cheap to fix.
- 1
Joint steering with real decision makers
People who can say yes in the room and have it stick, rather than delegates who take it back to someone.
Test: can this room release budget today?
- 2
Reporting you can interrogate
The underlying numbers and the plan, rather than a summary of them. If you can't open the file, you're being briefed instead of informed.
Test: ask for the source data once and watch.
- 3
An escalation path agreed in advance
Written down before anyone needs it, because agreeing a route in the middle of a crisis is how the crisis wins.
Test: name the next level up without looking.
You need a standing forum where uncomfortable conversations are expected. If hard issues only surface once something has already gone wrong, your cadence is decorative.
5 Raise issues early and on the record
Hold the firm to the scope and the quality bar you agreed, and have the difficult conversations early, in writing. An issue raised in week six is a conversation. The same issue raised at go-live is a crisis with your name on it.
Watch the change orders above everything. Each one on its own is usually reasonable. Collectively, "every issue is a change order" is how a fixed budget becomes a moving one without anyone deciding it should.
Twelve defensible change orders, one a month
6 Any three of these together mean you're already behind
Once you've seen these a few times they're hard to miss. None of them is fatal on its own.
Vague status that never answers the question you asked
Do this. Ask for the same detail twice in writing. The second refusal is the data point.
Reluctance to share the underlying numbers or the plan file
Do this. Put access to source artifacts in the SOW so it doesn't depend on the relationship.
Quiet churn, with experienced people replaced by juniors
Do this. Track the roster monthly against the named-personnel schedule. Churn is visible long before anyone admits it.
Every issue arrives priced as a change order
Do this. Keep a running log of change orders by root cause. A pattern of "requirement was unclear" is a scope-writing problem you can fix once.
Optimism that never survives contact with a date
Do this. Score the process and ignore the reassurance. A milestone that slips twice will slip a third time.
You and your integrator don't want exactly the same thing, and the contract is where that gets settled. Set it up well and most of the difficult conversations never have to happen.