A digital logbook starts paying for itself the day you stop using it to file completed tasks and start using it to find patterns. Months of temperature readings, flush records and signed-off checks sit in one place, searchable. Read them properly and they show you exactly where your water system is quietly drifting.

Most teams never get that far. The work gets logged, the dashboard turns green, and nobody asks the sharper questions: which outlets keep failing, which checks keep slipping, and how long a fault stays open before anyone fixes it. Performance improvement lives in those questions, not in the completion percentage on the home screen.

None of this needs a data scientist or a new tool. Useful log analysis is a couple of hours each quarter and a habit of reading the tail of the data, not the headline.

The reviews worth running each quarter

Run these against a defined period. A quarter is usually long enough to show a trend without drowning you, and grouping your findings the same way each time keeps the next review comparable.

Find the repeat offenders.

  • Pull every outlet, calorifier or tank that logged an out-of-range temperature, and count how many times each one did it. One miss is noise; the same sentinel outlet failing month after month is a design or scheduling problem wearing a fancy dress.
  • Cross-check those repeat offenders against their use pattern. A low-use shower that keeps reading tepid needs a different fix from a busy outlet that fails once.

Find where the routine slips.

  • Calculate the missed-task rate by location, by operative and by month. A wing running a fifth of its flushes late is a real exposure, not an admin tidy-up.
  • Flag any tasks closed late or back-dated in a cluster. A cluster usually means the round is scheduled wrong, not that the operative is careless.

Measure how fast problems close.

  • For each out-of-range reading, record the gap between the failed check and the verified re-check. A long lag means control exists on paper but not yet on site.
  • List every exception still open at the end of the period and write down why.

Look for drift, not just breaches.

  • Track cold-water readings through the warm months and hot-return readings at far outlets through the cold months. This is where trend analysis earns its place: slow creep toward the limit is the early warning that a single pass/fail check hides.
  • Watch for readings that are suspiciously identical week to week. Real water varies, and a flat line can mean the number was typed rather than measured.

Turning a finding into a fix

A pattern is only worth finding if it changes a decision. The discipline that separates a managed scheme from a tidy archive is writing down the decision, not just the reading. L8 expects duty holders to keep records of their precautions and monitoring and to keep the assessment under review [1]; an analysis that ends in a documented change is that review in action.

Fix the cause the data points at, not the symptom in front of you. An outlet that fails temperature every August because its riser shares a duct with a warm pipe will keep failing every August no matter how many times you re-check it. Re-route or insulate it once and the breach disappears from next year’s data. A flush that is always missed on the third floor is usually telling you the round is badly sequenced, not that anyone is cutting corners.

Then close the loop. Note the change, the date and the effect you expect, and check the next period’s data to confirm it worked. That before-and-after is the most persuasive entry in your audit trail, and it is the difference between continuous improvement and box-ticking (see Continuous improvement for the wider auditing habit it feeds).

The patterns people miss

Three things hide in plain sight.

Averages flatter. A 98% completion rate reads beautifully until you notice the missing 2% is the same rarely-used shower every single week, the exact outlet most likely to throw stagnant water into the air when it is finally turned on. Analyse the tail, not the mean.

A logged task is a claim, not a guarantee. The record tells you someone pressed “done”; it cannot tell you the tap ran long enough to reach temperature, or that the reading was taken at the outlet rather than guessed in the van. This is why a handful of physical spot-checks against the log still matters. The analysis narrows where to look; it does not replace looking.

Context turns a number into a finding. A laboratory result means far more read next to the temperatures and flushing logged around that outlet in the preceding weeks than it does on its own. Pairing the two is covered in understanding lab results.

Where data analysis stops

Analysing logs tells you what your records say happened. It cannot tell you whether a reading was taken honestly, whether a flush actually ran to temperature, or whether an outlet was even in use that week. Treat the analysis as a management aid sitting on top of competent, site-specific control under your risk assessment and written scheme, never as a replacement for physical verification. The temperature targets, acceptable limits and monitoring intervals you measure against come from that assessment and from HSG274, not from a dashboard’s default thresholds [2]. The same goes for sampling: where it is used, its frequency follows the system and the risk assessment rather than a fixed calendar [3].

Where to start

You do not need a year of data or a new platform. Export the last full quarter from your existing digital logbook and run only the first review: the repeat offenders. Most teams find two or three assets carrying the bulk of their out-of-range events, and fixing the worst one usually removes more risk than another round of sampling ever would. Write down what you found, what you changed and the date you will check whether it worked. That single page is where record keeping turns into improvement.

FAQ

What should I actually measure to show our Legionella control is improving?

Track the things that move: the number of out-of-range temperature readings, the missed-task rate, and the average time to close an exception. Falling figures across consecutive periods are stronger evidence of real control than a high completion percentage, which can stay green while the same outlet quietly fails.

How often should we analyse the logbook data?

Often enough to act before a trend becomes a breach. A quarterly review suits most sites, with an extra look after any change to the system, the building’s use or the people exposed. Your risk assessment sets the underlying monitoring frequency; the analysis cadence sits on top of it [2].

Does analysing the data count towards compliance, or is it extra work?

It supports the review duty rather than adding a new one. Keeping records and keeping the assessment under review are already expected of duty holders [1]; using those records to find and fix weak points is simply doing that review well instead of merely filing it.

Sources

[1] HSE, “Legionnaires’ disease. The control of legionella bacteria in water systems - Approved Code of Practice and guidance (L8)”. https://www.hse.gov.uk/pubns/books/l8.htm [2] HSE, “Legionnaires’ disease: Technical guidance (HSG274)”. https://www.hse.gov.uk/pubns/books/hsg274.htm [3] HSE, “Testing and monitoring your water system for legionella”. https://www.hse.gov.uk/legionnaires/testing-monitoring-water-system.htm