Insight
What to do when your agency goes quiet mid-project
You're not sure whether this is normal. You're not sure whether it's fixable. You're not sure whether to push harder, pause payments, or start drafting a termination letter.
First thing worth saying: this is a recognisable situation. It feels unique from the inside. From the outside, the pattern is textbook, and the playbook for dealing with it is well-understood. The goal of this article is to help you diagnose your own situation calmly and work out which lever to pull first.
No blame, no drama, no "agencies are the enemy" energy. Most good agencies want this resolved as much as you do.
Why agency communication breaks down
Four causes explain most of what you're seeing, and usually it's more than one of them at the same time.
Scope creep without formal change control. Small requests accumulate. "Can we just add a filter to the search?" "Can this section be dynamic rather than static?" Each one feels minor. By week sixteen, the project is thirty per cent bigger than the brief and the agency has quietly absorbed the cost, or not, and the margin has evaporated. Once margin goes, so does responsiveness.
Resourcing changes at the agency. The team that pitched isn't the team delivering. A senior has left, a junior has been promoted into a role they're not quite ready for, or the project has been handed down the priority list because a bigger client walked through the door. Agency Management Institute's research suggests half of senior agency staff turn over every 18 months in some markets. Your project is almost certainly being run by someone who wasn't in the original pitch.
Over-commitment. The agency sold too much work, couldn't staff it, and your project is now someone's third priority behind two others that shout louder. This is endemic in the market cycle, especially when agencies are coming off a lean quarter and said yes to everything.
Brief gaps. The original brief left things ambiguous. Both sides filled the gaps differently. The gap doesn't surface as a problem until integration, UAT, or launch, at which point the cost of fixing it is ten times what it would have been at kick-off.
Notice what's not on that list: the agency being secretly bad, or deliberately dishonest. That's rare. The common causes are structural, which is useful, because structural problems have structural fixes.
Five diagnostic questions
Before deciding what to do, get honest about what you're actually looking at. These questions will take ten minutes and will save you weeks.
- When was the last genuinely useful (not polite, not generic) status update?
- Are agency responses getting shorter and taking longer?
- Has the person leading your project changed since kick-off? Quietly or formally?
- Are deadlines slipping one at a time, or in larger chunks each time?
- When you ask a direct question, do you get a direct answer, or a reassurance dressed as one?
If three or more of those flag, you're not imagining it. You're in a recoverable-but-deteriorating situation. The rest of this article is for you.
What you actually have available to you
You have more leverage than you probably feel like you have. The question is which lever to use first.
Your contract. Read it again, properly. Milestones, payment schedule, change control process, acceptance criteria, termination clauses, IP ownership. The contract is usually written in the agency's favour on the first draft, but mid-market contracts still typically include a termination-for-cause clause and a handover provision. Know what's in yours before you start making demands.
Escalation. Most agencies have a hierarchy you can use. Account manager, account director, studio manager, managing director. Escalation doesn't need to be aggressive. A neutral email to the next person up saying "the project isn't progressing the way we expected and I'd like to have a conversation about it" is usually enough to get attention. If it isn't, that tells you something important about the agency.
Independent technical review. Someone external, not selling you the next engagement, who can read the codebase, the project documentation, and the invoicing and tell you honestly where things stand. More on this below.
Paused payments. Risky, but sometimes necessary. Only pause inside what the contract allows. Talk to your lawyer before you do it. A paused payment is a communication device, not a punishment, and it works best when you name it explicitly: "We're holding this milestone payment until we have visibility on X, Y, Z."
Termination. Always an option. Usually more expensive than repairing the engagement once you factor in time lost, new vendor onboarding, and the write-off on work already done. Sometimes it's the right call anyway. The honest test: would a reasonable new vendor, looking at what's been built, want to continue it or start over?
How to tell if the project is recoverable
Not every project can be saved, and a recoverable project isn't always worth saving. The signals below are reliable.
Recoverable signals:
- Clear, even if uncomfortable, conversations are still possible with the agency lead.
- The work done so far has durable value, even if it's not all polished.
- The underlying brief is still valid. The thing being built is still the thing the business needs.
- A specific, nameable fix would unblock progress: new resourcing, tighter scope, clearer governance, better project management cadence.
Not-recoverable signals:
- The built product has diverged fundamentally from the brief, and both sides are now arguing about what the brief said.
- Technical debt has accumulated to the point where extension is unsafe.
- The relationship is broken at the principal level, not just the project-manager level.
- Continued spend has worse expected ROI than a restart, even accounting for the sunk cost.
Sunk cost is the trap. Money already spent is gone. The only question is which path forward returns the most value from where you stand today. Daniel Kahneman wrote the book on this, and the sunk cost fallacy is the one most businesses fall into at exactly this point in a project.
Why an independent review is usually the first move
Before you escalate, terminate, or pause payments, it's almost always worth getting a third-party read. The reason is simple: your judgment is currently being made from inside the relationship, and so is the agency's. You need someone with no stake in either outcome.
A proper independent review looks at three things:
- The code. What's been built, how well, how maintainable it is, how far from done.
- The documentation. Tickets, design files, decisions log, test coverage. What's tracked, what's lost.
- The commercials. What's been invoiced, what's been paid, what's outstanding, whether the numbers match the work visible in the code and the documentation.
Who can do this well: a senior digital lead who's been on both sides of the table. Not another agency, because the conflict of interest is obvious. Not a generalist project consultant, because they can't read the code well enough to separate "nearly done" from "needs significant rework". Someone who can do both.
The review produces a short report: what's actually been built, what's missing, what's salvageable, what the realistic remaining effort looks like, and what to negotiate with the existing agency before deciding whether to continue.
What happens after the review
Three paths are typical. In rough order of frequency:
Re-plan and continue. The review gives both sides something concrete to discuss. Scope gets tightened. Governance gets fixed. A new weekly cadence gets introduced. The agency often welcomes this, because it gives them a way to reset the engagement without losing face. Most projects land here.
Finish and walk. You agree on a defined end point. The agency delivers to that point. You take what's usable, launch a stripped-down version, and run the next phase with a different partner. Less satisfying, often the right commercial call.
Cut and restart. Rare, and usually correct when it happens. This is the right move when the built product is fundamentally wrong, the brief has shifted, or the relationship is unsalvageable. Expensive in the short term, cheaper over twelve months than dragging a broken engagement across the line.
Preventing this next time
Most of the ways an agency engagement goes wrong were set up at the brief. The fixes are equally upstream.
- Tighter brief, including what's out of scope. Out-of-scope lists prevent more disputes than in-scope lists.
- Milestone-based payments with acceptance criteria. "Pay on delivery of X, verified by Y" beats "pay on month three" every time.
- Regular independent quality gates. A third-party technical review at 30%, 60%, and 90% of the build catches drift while it's still cheap to fix.
- Stakeholder discipline on change requests. One named person approves changes. Everyone else routes through them. This is boring, and it works.
- Choosing the right agency in the first place. Most of what shows up mid-project was predictable at shortlist. The vendor and agency selection service exists because of how consistently this is the root cause.
Further reading worth your time: Agency Management Institute publishes useful content on how good agencies run themselves, which helps you spot when yours isn't. For a view from the delivery side, Mule Design's archives and A List Apart still hold up years on.
Where to from here
If you want to calibrate your own read before doing anything, the five diagnostic questions above are a ten-minute exercise. Walking through them with your internal team, or with a senior peer from another business, will usually make the next step obvious.
If the situation is serious enough that you want someone external to look at it properly, the project rescue engagement is scoped for this specifically. An independent read in a week, with a clear recommendation at the end.
The goal is not to fire the agency. It's to protect the investment you've already made, from either direction. Usually that means fixing the engagement. Sometimes it means something else. You can't tell which until someone looks clearly.