Let me tell you what kills a weekly KPI cadence faster than anything else. It is not the discipline of running the meeting. It is not the format of the scorecard. It is the data underneath.
70,000 projects ran through my agency’s CRM over the 13 years I built and sold it. Every single one of those jobs had a naming convention, a numbered folder structure, an enforced set of fields. We built a £2.2M creative agency in Belfast and exited in 2022, and the rigour around that data is what made every other operating system work on top of it.
Sh*t in, sh*t out. If the numbers feeding your scorecard are wrong, late, or inconsistent, you are making decisions with high confidence on bad information. That is worse than no information, because at least with no information you know you are guessing.
This article is the data hygiene that has to sit underneath any weekly reporting cadence you install, plus the five failures that quietly erode it.
What bad data hygiene actually looks like in week one
Here is what bad data hygiene looks like in week one. You have just installed a weekly KPI cadence. Revenue. Margin. Capacity. Pipeline. Churn risk. The meeting is in the calendar, the owner is assigned. Week one happens, and half the numbers are late, incomplete, or wrong.
Revenue looks right. Cash looks right. Margin is broken because three people logged time into different project codes. Utilisation is empty because nobody submitted their timesheet last week. The pipeline has two deals sitting in “qualified” that have not been touched in six weeks.
The scorecard gets filled in. The team knows the numbers are soft. The conversation in the meeting becomes another debate about methodology instead of decisions. Same problem as inconsistent definitions, except now it is the data underneath that is wrong, not just the labels on top.
Why this matters beyond your operating rhythm
There are two reasons to fix this. One is for you, today. The other is for a buyer, eventually.
A buyer does not just ask what your margins are. They ask how you know. They want to see the system that produced the numbers. If the answer is “we pull it from three different tools and we reconcile it manually”, that is a control risk. Control risk reduces the multiple.
For you, today, every week that passes without visibility is a week where scope creep goes unchecked, utilisation drops unnoticed, and cash collection slips. The monthly P&L is a post-mortem. Weekly KPIs are the steering wheel. And if the steering wheel is connected to nothing, you can turn the wheel all you want.
I worked with an agency owner who had been running his business for years with zero time tracking. He could not tell whether a logo project quoted at two hours had actually taken two hours or six. If it took six, was that scope creep from the client, internal inefficiency, or a training gap? He had no idea. All he knew was the revenue and the cost. The true cost was invisible.
We installed time tracking as the first step. Not to bill hourly, not to micromanage. Just to see the financial reality behind the work. Within three months he could identify which services were profitable and which were losing money on every single job. The picture was not pretty. But it was finally honest.
That is what data hygiene gives you. Not more admin for its own sake. Visibility.
The minimum data standard
Your agency must have a minimum data standard. For every client, every project, every deal, and every invoice, there is a set of fields that must be filled in. If they are not, reporting will break.
This is not about building a perfect system. It is about defining the floor.
Think of it as a single source of truth. One system manages the entire customer journey from lead through to completion. Every other tool feeds into it. If your pipeline lives in one tool, your delivery in another, your invoicing in a third, and none of them talk, you do not have a system. You have three disconnected lists.
Client record
Every client must have:
- A standardised client name (not three different spellings across three tools)
- A primary service line
- A delivery owner
- A price model (retainer, project, or hybrid)
- A start date and a billing schedule
If any of these are missing, you cannot report margin by client cleanly.
Project record
Every project must have:
- A standardised project name (and a numbering convention; more on this below)
- A linked client ID
- A service category
- Start and end dates
- A budget in hours or pounds
- A delivery owner
Deal record
Every deal must have:
- A stage definition that matches your written criteria, not “qualified because it felt like a good conversation”
- A close date
- A deal owner
- An expected monthly or project value
Invoice record
Every invoice must have:
- A linked client ID
- The period or project it covers
- The service line
- The payment terms
I can hear the objection: this is a lot of admin. It is not. It is a one-time setup. Define the fields once, build them into your templates once, then enforcing them is a 5-minute check, not a rebuild every time.
At my agency, we baked this into our CRM from the very first day. When a team member created a new order, the system required certain fields before you could proceed. You could not skip them. That meant the data was always complete. Not because people were disciplined, because mostly they were not. Because the system enforced it.
The five failures that erode data hygiene
These are the mistakes I made or watched others make. Each one costs months of compounding bad data before anyone notices.
Failure 1: Too many tools with no governance
If you have five tools, no naming convention, no shared join key, no data standard, every team member creates projects how they want. After six months, the data is so inconsistent that nobody trusts it. Everything is back in your head as the founder.
The fix is not more tools. It is fewer tools with stricter rules. We had a stretch where job numbers on project folders were inconsistent. Some team members misnumbered or filed work in the wrong folder. The chaos was real. We wrote a small automation that, when a project was created in the project management system, the corresponding folder and subfolders were created automatically with the correct naming. Folder errors dropped to zero overnight, forever.
Failure 2: Migrating without parallel running
I watched an agency rip out their old system and install a new one on a Monday morning. By Wednesday afternoon, nobody could find anything.
The clients who do this well run both systems in parallel for 8 to 12 weeks. During the overlap they find bugs in the new system, identify gaps in the old that could be built in, and get the team comfortable. By the time the cutover happens, every conversation is “this works” instead of “where is that file?”
If you are changing tools, run both for at least two months. The friction is worth it. The team’s buy-in even more so.
Failure 3: Building when you should buy
This is the most expensive failure I made.
We built our own CRM from scratch. It cost over £100,000 in developer cost and pulled my attention away from tasks that would have scaled the business considerably more across the two years it took to build and maintain. Looking back, knowing what I know today, I would have bought an off-the-shelf CRM and project management system.
It is not the time to build that matters. It is the time to maintain. The upkeep, the security patches, the team training. That is a continuous tax. Unless tool development is your core business, do not pay it. Buy an off-the-shelf tool and enforce your standards within it. The tool does not create reliability. Governance does. This is part of why the agency ops stack matters more than which specific tool you pick.
Failure 4: No consequences for bad data
If someone can skip fields, submit it, or log work to the wrong project code with no follow-up, the standard erodes in weeks.
The hygiene ritual is the enforcement mechanism. When the owner reviews data weekly and flags gaps, compliance improves. The only way I have seen this work cleanly is small perks tied to compliance: if your time is logged accurately every week, if your projects are on deadline, you qualify for whatever the perk is. It is binary, not subjective. And someone is the gatekeeper checking the criteria every week. That turns data hygiene from a chore into a habit with reward.
Failure 5: Confusing time tracking with surveillance
Time tracking is not about checking whether your team are working seven and a half or eight hours. It is about understanding where the capacity in your business goes.
Which clients consume the most delivery time relative to revenue? Which service lines are profitable, which are not? The data answers questions that gut feelings cannot. Frame it that way to your team and the resistance disappears.
The simple test
After 60 days of running hygiene rituals, ask one question. Can a non-founder answer these four questions in 30 minutes?
- Top 10 clients by revenue and margin.
- Utilisation last week.
- Pipeline coverage for the next 30 to 60 days.
- Accounts receivable ageing, or top overdue invoices.
If yes, the hygiene is working. If not, either the standards or the enforcement need tightening.
What to do this week
Two actions.
One: open your CRM, your project management tool, and your finance system. Pick one client. Can you trace that client across all three using a consistent name or ID? If not, that is your first fix. Create a naming convention and a client ID that links everything together.
Two: write your minimum data standard. One page. What fields must exist for a client, a project, a deal, and an invoice? Print it. Pin it. Make it the reference document the team uses when they create anything in any system.
Those two actions take less than an hour. They are the foundation that makes everything else work. Your KPI cadence, your margin reporting, your capacity planning, your retainer renewals, all of it depends on the data being true.
If you want to see how your reporting and data hygiene affect what your agency is worth in a buyer’s eyes, the free Agency Valuation Calculator walks you through that scoring in about 10 minutes. Data hygiene is one of the levers buyers actively discount when it is missing, so a clean read on where you sit is useful whether you are six months or six years away from any conversation.
For structured support installing data hygiene and the operating cadence on top of it, the Strategic Growth Programme covers exactly this work over 12 months. Book a discovery call if you want to talk through the state of your data.
Hear this in practice on Exit Ready:
- Aedin O’Neill on process discipline as protection against AI mistakes - what happens when junior team members lean on AI without the data fundamentals underneath.
- David Kieran on automating Zoma operations through Notion - one tool, one source of truth, enforced through automation.