Guide
The repetitive admin audit: finding the hours your business loses
A repetitive admin audit is a structured walk through how work actually flows in your business, recording every point where somebody does something a computer should be doing. The output is a list of jobs with hours and costs attached, ranked by what is worth fixing. You can run it yourself in an afternoon, and this guide explains exactly how.
By James 10 min read
Why individually small jobs matter
No single piece of repetitive admin is worth fixing. That is precisely why it survives. Four minutes to retype a customer’s details is not worth a project, and everybody knows it, so nobody raises it.
Four minutes, four times per job, twelve jobs a week is 165 hours a year. Across three such jobs it is most of a part-time salary, and it is being paid by people whose time you are buying for something else.
The audit exists to turn "we are drowning in admin" into a number. Once there is a number, the decision about whether to fix it stops being a matter of temperament and becomes arithmetic.
How to run the audit
-
Step 1
Follow one job end to end
Pick a typical job and trace it from first enquiry to payment received. Write down every system it touches and every person who handles it. Do not describe how it is supposed to work, describe what happened to the last one.
-
Step 2
Mark every re-entry
Every point where information already in the business gets typed in again. These are the highest-value finds because they cost time and generate errors simultaneously. Highlight them separately from everything else.
-
Step 3
Put numbers against each one
Minutes per occurrence, occurrences per week, and who does it. Time a couple with a stopwatch rather than estimating: people underestimate familiar tasks by a wide margin, usually by half.
-
Step 4
Add the error cost
For each re-entry, ask what happens when it goes wrong and how often it does. A mistyped address, a wrong price, a missed invoice. This is usually larger than the time cost and is nearly always left out.
-
Step 5
Rank by saving over effort
Annual cost against rough difficulty. The top of that list is what to fix first, and it is very rarely the thing that annoys people most. It is usually something quieter and more frequent.
Free, before any money changes hands
Want this looked at properly, for nothing?
Ninety minutes mapping how the work actually flows, then a written plan and a fixed quote you keep either way. A working session, not a sales meeting.
- No obligation, no follow-up sequence
- You keep the plan and the quote
or call 01623 354250
Where to look first
These account for the majority of what we find, across almost every sector.
- Quotes retyped as invoices, then retyped again into the accounts package
- Enquiries arriving by phone, web form and social message, then consolidated by hand
- Job sheets written on paper and typed up at the end of the day
- Customer details entered separately into three systems that do not talk
- Invoice chasing done from a list somebody has to remember to check
- Weekly or monthly reports compiled by pulling numbers out of four places
- Appointment reminders sent individually
- Timesheets and expenses reconstructed at the end of the week from memory
- The same document produced repeatedly with three fields changed
- Anyone exporting data from one system purely to import it into another
Costing it honestly
Use a fully-loaded hourly cost, not a salary divided by hours. Employer national insurance, pension, holiday, equipment and the overhead of having a desk typically add 25 to 35 per cent to gross pay. A £26,000 administrator costs somewhere around £20 to £22 an hour once that is counted.
Where the person doing the retyping is the owner, cost it at what the owner’s time is actually worth on chargeable work rather than at an administrative rate. That number is usually startling and it is the honest one.
Then be conservative with the saving. Automating a job that takes ten minutes rarely takes it to zero; it usually takes it to two. Claim the eight and not the ten, because a business case built on optimistic numbers falls apart at the first review.
Or we run it with you, free
The free process review is this audit, run as a working session at your premises or over a call. It takes about 90 minutes, you get the written plan and a fixed quote, and the plan is yours to act on with us or without us.
What to do with the findings
Not everything on the list needs software. Sort the results into three groups before anybody quotes for anything.
Change how you work. A surprising number of repetitive jobs exist because of a decision nobody remembers making. Somebody prints a report because a manager wanted it in 2019 and nobody has asked since. These cost nothing to fix and should be done first.
Configure what you already own. Most businesses use a fraction of the software they pay for. A connection between two packages you already have, or a feature nobody switched on, will often remove a job entirely for the cost of an afternoon.
Build something. Only what is left after the first two, and only where the annual cost genuinely justifies the build cost plus running costs. That list is normally shorter than people expect, and it is far more likely to succeed because everything cheap has already been done.
For that third group, see what custom software costs and when it is worth it.
The mistakes that spoil an audit
Asking managers instead of the people doing the work. The person who does a job knows where the time goes; the person who designed the process knows how it is supposed to go. Those are different documents.
Describing the ideal path. Every process has exceptions, and the exceptions are usually where the cost lives. Ask what happens when a customer changes their mind, when a part is out of stock, when the job overruns.
Only counting time. Errors, delays and the enquiries that were never followed up are frequently worth more than the hours, and they are invisible unless you deliberately ask.
Stopping at the obvious one. The loudest complaint is rarely the most expensive job. It is worth finishing the audit even after you have found something worth fixing.
Doing it once. Businesses accumulate new admin continuously. An audit is worth repeating annually, and it goes much faster the second time.
A worked example, end to end
Take a nine-person heating firm doing roughly twelve jobs a week. Following one job from enquiry to payment turns up six re-entries: the customer’s details typed at enquiry, again on the quote, again on the job sheet, again on the invoice, again into the accounts package, and once more into the VAT spreadsheet at quarter end.
Timed rather than estimated, the first five take about four minutes each. That is twenty minutes per job, twelve jobs a week, forty-eight weeks, around 192 hours a year. At a loaded cost of £22 an hour, that is a shade over £4,200 of pure retyping before anything goes wrong.
Then the error cost. Across a year the firm can identify four invoices sent to the wrong address, two jobs quoted at the previous price, and roughly a day a month spent reconciling what the accounts package says against what the job sheets say. Conservatively another £3,000.
So the honest annual figure is somewhere near £7,000, of which about £4,200 is directly recoverable. A £6,000 build that removes five of the six re-entries pays back inside eighteen months on the recoverable figure alone, and considerably faster once the errors stop. A £20,000 build does not, and the audit is what tells you that before anybody has quoted.
What the audit tells you about the business
The findings are usually more interesting than the numbers. Repetitive admin accumulates for reasons, and the reasons are diagnostic.
Work that exists because two systems cannot talk usually means a purchasing decision made in isolation. Work that exists because one person prefers a particular format usually means nobody has asked why in several years. Work that exists because of a customer requirement is often the most valuable to keep and the most valuable to automate properly.
The most revealing category is work that exists because somebody does not trust a system. If invoices are checked manually because the software got it wrong once in 2022, the real problem is confidence rather than admin, and no amount of automation fixes it until the underlying issue is found and named.
That is why the audit is best run as a conversation with the people doing the work rather than as a form. The number is the deliverable; the explanations are what make the number actionable.
Turning the findings into a decision
An audit that produces a list and no decision is a document that gets filed. Two habits stop that happening.
Put a named owner and a date against each item, including the free ones. "Stop producing the Friday report" needs an owner in the same way a software build does, and it is far more likely to be quietly ignored precisely because it costs nothing.
Then set a review date three months out and measure the same tasks again. The second measurement is what tells you whether the change held. A surprising number of process improvements survive four weeks and then quietly revert, usually because the old method was faster for one edge case nobody mentioned during the audit.
Where a build is justified, take the audit into the quoting conversation rather than a wish list. A supplier quoting against measured hours produces a materially better proposal than one quoting against a description, because the scope is anchored to a number you both agree on. It also makes the supplier’s judgement visible: if they price everything on the list as worth building, without pushing back on a single item, that is the answer to a different question.
Frequently asked questions
If the answer is not here, ask us. You will get a straight one, from someone who does the work.
Mon–Fri, 9am–5:30pm
What is a repetitive admin audit?
A structured walk through how work actually flows in your business, recording every point where somebody does something a computer should be doing. The output is a list of jobs with hours and annual costs attached, ranked by what is worth fixing first.
How do I calculate what repetitive admin costs us?
Minutes per occurrence times occurrences per year times a fully-loaded hourly cost, gross pay plus 25 to 35 per cent for employer costs and overhead. Then add the error cost: mistyped details, wrong prices, missed invoices. That second figure is usually larger and nearly always omitted.
How long does the audit take?
About 90 minutes to walk one job end to end and put numbers against what you find. Timing two or three tasks with a stopwatch rather than estimating adds a little and is worth it, because people underestimate familiar tasks by roughly half.
Does everything we find need software to fix?
No, and sorting the findings matters. Some jobs exist because of a decision nobody remembers making and can be stopped for free. Many more can be removed by configuring or connecting software you already pay for. Only what is left is a build, and that list is shorter than most people expect.
Who should be in the room for the audit?
The people who actually do the work, not only the people who designed the process. Those two groups will describe different businesses, and the difference between the descriptions is where the cost is hiding.
How often should we repeat it?
Annually. Businesses accumulate new repetitive admin continuously: a new system, a new reporting requirement, a new customer with a different process, and the second audit is considerably faster than the first.
What do we do once we have the list?
Put a named owner and a date against every item, including the free ones. Those are the most likely to be quietly ignored precisely because they cost nothing. Then set a review three months out and measure the same tasks again, because a good number of process changes survive four weeks and quietly revert.
What does the audit tell us beyond the numbers?
Why the admin accumulated, which is usually more useful. Work existing because two systems cannot talk points at a purchasing decision made in isolation. Work existing because somebody does not trust a system points at a confidence problem that no amount of automation will fix until it is named.