Nick QuirkEvery agency over three years old has one. A delivery system nobody designed, assembled from tools that were each a sensible choice at the time, held together by spreadsheets, Zaps, naming conventions and one senior PM's memory. The Franken-System.
It works. That's the trap. It works well enough that ripping it out feels riskier than living with it — right up until the person who understood the connective tissue leaves, or a client asks a simple question and it takes four people 90 minutes to answer.
Agency tool sprawl is not a software problem. It's an accumulation problem. Here's how to audit what you've actually built, decide what to kill, and consolidate without torching a quarter of delivery capacity.
What agency tool sprawl actually looks like
A real stack from a 14-person SEO agency I worked with, counted honestly:
- Project management tool for tasks (but not time)
- Separate time tracker, synced nightly by an automation that silently broke in March
- Google Sheet for retainer hour budgets — the real source of truth
- Second Sheet for capacity planning, updated Mondays by the ops lead
- Slack channel per client, plus a general channel where half the decisions actually happened
- Looker Studio dashboards, manually refreshed before each client call
- Shared Drive folder structure with three competing naming conventions from three different eras
- Technical audit findings in one tool, dev tickets in the client's Jira, status in the PM tool
- Invoicing in accounting software, reconciled against the hours Sheet by hand each month
Nine systems, four of them spreadsheets, and at least six manual bridges between them. Nobody chose this. It grew.
The three tells
Duplicate entry. The same fact — a task is done, a client approved something, a deliverable slipped — gets typed into two or more places. Anywhere you type something twice, you have a seam.
Single-operator processes. One person "does the numbers" each month. Nobody else knows how. That's not a workflow, it's a hostage situation, and it becomes obvious the moment they take annual leave. We've written about this failure mode in more depth in surviving employee turnover in a marketing agency.
Reconciliation meetings. If you have a recurring meeting whose main purpose is making two systems agree, the meeting is the workaround.
Why the Franken-System forms
Each piece was a rational local decision. The Sheet existed because the PM tool couldn't do retainer burn. The second tracker existed because the first one's reports weren't invoice-ready. The Slack channel existed because a client hated logging in.
The cost of each addition was small — 20 minutes a week, one more tab. The cost of the system is the product of all of them plus the joins, and nobody is measuring that number. That's the asymmetry that lets sprawl grow indefinitely.
There's a second driver: workarounds are how teams absorb scope creep without confronting it. If clients routinely push requests through side channels, you'll build side-channel machinery to cope. Fixing the stack without fixing the boundaries just gives you a cleaner system that degrades again in six months — which is why client boundary setting belongs in the same conversation.
Run a 30-day workaround audit
Don't start with tool evaluation. Start with evidence. For 30 days, every person logs each workaround they perform, in one shared doc, with four fields:
- What you did — "copied hours from tracker into the retainer Sheet"
- How long it took — in minutes, honestly
- Why it exists — "tracker export doesn't group by retainer"
- Who else can do it — names, or "nobody"
The agency above logged 41 distinct workarounds in a month. Total time: 27 hours. At a blended internal cost of £45/hour that's roughly £1,200 a month, or £14,600 a year — before counting the errors, the two under-billed retainers they found, and the invisible tax of context switching.
The log matters more than the total, though. It tells you which workarounds are frequent and fragile versus rare and cheap. You want that distinction before anyone opens a pricing page.
Triage: kill, keep, consolidate
Sort every logged workaround into three buckets using blunt rules.
Kill (about a third of them)
Workarounds that serve a process nobody needs anymore. The weekly status Sheet that three people update and zero clients read. The tagging convention invented for a client that churned in 2023. The dashboard refreshed for a monthly call that became quarterly.
Test: stop doing it for two weeks and see who complains. In practice, most don't get missed. This bucket is free capacity — take it first, because it builds momentum before you touch anything structural.
Keep (a smaller third)
Some workarounds are genuinely cheaper than the system that would replace them. A five-minute monthly export is fine. A client who insists on receiving reports as PDFs by email is not a tooling problem, it's a client preference, and honouring it costs you ten minutes.
Rule of thumb: keep it if it runs monthly or less, takes under 15 minutes, and at least two people can perform it. Document it in one paragraph and move on.
Consolidate (the expensive third)
Anything weekly or daily, anything over 15 minutes, anything only one person can do. These are the load-bearing workarounds, and they're almost always clustered around the same three seams: tasks ↔ time, time ↔ retainer budget, and delivery ↔ client reporting.
Those three seams are where agency-specific needs diverge from generic PM software — which is exactly why so many agencies end up with a Sheet bolted onto a tool that was built for product teams. If you're evaluating replacements, look at how candidates handle those seams natively rather than at feature counts; a structured comparison of agency PM tools is more useful here than a demo reel.
Consolidate in the right order
The instinct is to pick one platform and migrate everything in a weekend. Don't. Sequence it by risk.
First: the billing seam. Hours, retainer budgets and invoicing. It has the clearest financial payback, the audit trail is objective, and getting it right stabilises cash flow visibility — a related issue covered in fixing feast-or-famine cash flow. Run the old Sheet and the new system in parallel for one full billing cycle. Reconcile. If they disagree by more than 2%, find out why before you cut over.
Second: delivery and task tracking. Migrate live projects only. Do not migrate archives — export them to a read-only folder and leave them. Agencies lose weeks to importing three years of closed tasks nobody will open again.
Third: reporting. Once the data underneath is clean, reporting usually shrinks from a build job to a configuration job.
Never migrate everything at once. Pick two clients as the pilot — one straightforward retainer, one messy project-based account. Two weeks. Fix what breaks. Then roll out.
