She finished the ninety-day clean-up in March. The CRM was tidied, the shared drive was reorganised, the glossary was written down, the first AI tools started returning answers the team trusted. By September she could feel the discipline slipping. Files were drifting back to people’s desktops. A new joiner had started using the old word for what the team now called something else. The forecasting tool’s outputs had started looking off again, and nobody was sure when that began.
She wanted a structured way to catch the drift before it undid the work. Not a dashboard, not a scorecard. Something she could run in thirty minutes with two senior colleagues, every quarter, that told her honestly whether the readiness she had paid for in spring was holding.
The answer is four questions. Each one cuts through a different layer of data and knowledge decay. None of them needs specialist tooling, and all of them are answerable by the owner and one or two senior people in a fixed quarterly slot.
Why does data and knowledge readiness need measurement more than other operational areas?
Because the decay is silent. A cash crunch shows up in the bank account, hiring missteps create immediate friction, churn surfaces in the pipeline. Data decay does none of these things. It accumulates quietly while everything else looks fine, and by the time it is visible it is also expensive. Salesforce found fewer than one in five organisations rate their data as ready for AI, and the trend is getting worse, not better.
The proportional cost at SME scale is real. A doubletrack analysis of 8.36 million US businesses puts the annual cost of poor data quality at roughly $4,912 per employee per year, which translates to a £49,000 to £98,000 hit on a ten to twenty-person team. Forty-two per cent of companies scrapped a majority of their AI initiatives in 2025, up from 17 per cent the year before, with poor data quality named as the primary reason. Acceldata reports the share of data professionals citing poor data quality as their top challenge rose from 41 to 57 per cent between 2022 and 2024. The decay is accelerating.
Question one, can the team find the latest version in under sixty seconds?
Pick three to five documents the business genuinely depends on this quarter. A current pricing list, a contract template, a statement of works, a recent client engagement report. Ask a team member who did not organise the system to find each one, and time them. If they average over sixty seconds the discipline is slipping, and if they cannot find the latest version at all, the firm has a data availability problem already costing time.
Findability is a decision-making issue, not a filing issue. Paperwise reports that employees in knowledge-intensive work often spend up to 30 per cent of their time searching for information, which on a fifteen-person team is close to two full-time equivalents. The risk runs beyond lost time and into using the wrong version. Compliance documents, pricing sheets and scope statements that exist in multiple versions produce client commitments made against wrong terms, invoices priced incorrectly and scope creep that nobody catches until billing.
The fix is rarely a new platform. It is a small intervention by whoever owns the question this quarter. Simplifying a folder structure, retraining the team on the naming convention, archiving the old versions properly. If retrieval has drifted from thirty seconds to ninety, that is a signal worth a fortnight. If it is holding, leave it alone.
Question two, are glossary disagreements actually being resolved?
In the past ninety days, have disagreements on definitions of core business terms surfaced. Customer state, deal stage, project status, invoice status. Were they resolved, or did they go quiet without being fixed. How many are still open. If the leadership answer is “not many” or “they resolve themselves”, the glossary governance is probably drifting and a backlog is quietly forming. If the answer is “yes, and we have a process”, the culture is holding.
The reason this matters is that definition misalignment silently corrupts every metric that depends on the definition. If sales reads a deal at “proposal” and operations reads it as “in scoping”, the forecast is wrong before anyone notices. OvalEdge documents the case-study pattern, when departments define the same metric differently, dashboards conflict and audit exposure rises, and when alignment is enforced through an actively-reconciled glossary, reporting accuracy improves and KPI disputes drop. The same dynamic plays out at SME scale, just with smaller numbers and fewer dashboards.
The owner-scale version is a list of five to ten core definitions, one named owner who maintains the glossary, and a quarterly answer to the question above. If new conflicts keep surfacing, the answer is rarely a complex governance model, it is a clearer one. Someone owns the glossary, someone maintains it, and someone has authority to resolve conflicts when they emerge.
Question three, what proportion of AI output does the team have to correct before using it?
Pick one AI tool the team genuinely uses. Across the people using it, track for a month how often the output went out as-is and how often it needed editing before use. Below 20 per cent correction is a tool earning its keep and adoption should expand. Above 50 per cent and it is creating rework disguised as automation, and the right move is to pause, fix the data feeding it, or replace it.
Workday research found that approximately 40 per cent of the time saved through AI tools is offset by the extra work created fixing AI-generated output, and the rework typically falls to the most experienced people in the team. Fourdots puts the annual global cost of AI hallucinations at $67.4 billion, with financial-task hallucination rates running 15 to 25 per cent without proper guardrails. The cost compounds, the most engaged people end up doing the correction, and trust in the tool degrades.
The fix is rarely a better AI tool. It is usually better data and governance feeding the existing one. If a deal-forecasting model is fed historical deals with inconsistent stages, the forecast will be inconsistent. The quarterly question is whether the correction rate is trending up or down across the year, which tells the owner whether the underlying data quality is improving or degrading.
Question four, what new data or knowledge debt has accumulated in the last ninety days?
What looks like new debt. What is not quite broken yet but is trending toward broken. What new silos are forming. What definitions are drifting again. What integrations are creating unexpected data patterns. Each quarter the leadership team triages. Some items will be trivial and can be ignored, some will need immediate action, and the bulk will sit in the middle, worth monitoring and worth a named owner.
This question matters because debt compounds like financial debt. The longer it sits unprioritised, the more costly it becomes, and ERP Today’s World Economic Forum framing makes the human cost explicit, every hour a senior person spends deciphering ungoverned legacy data is an hour not spent on higher-value work. A well-run quarterly cycle catches new debt early, when it is still cheap. A poorly-run one waits until a tool starts disappointing and then runs a forensic investigation that takes weeks.
The discipline is the schedule. A thirty-minute meeting, same date every quarter, owner plus two senior colleagues plus whoever runs the systems holding critical data. The four questions are shared in advance. Each answer feeds three to five action items with named owners and deadlines, reviewed at the next quarter. That structure turns “everyone should care about data” into “this person is tracking this metric, this quarter, for this team”, which is the difference between sustained improvement and silent regression. The Gainsight quarterly business review framework is a reasonable starting point.
This is the measurement that pays. It will not sell you a platform and will not look impressive on a slide. What it does do is run on the schedule owner-operated firms can actually keep, and catch drift before it costs money. Want a hand designing the quarterly rhythm for your firm? Book a conversation.
