Bucket three, operations

The same fact typed into four systems by three people is not a process. It is a tax you pay daily.

Almost every business at this size runs on re-entry. A job is completed on site, then written on a card, then typed into a spreadsheet, then read out to accounts, then keyed into the invoice, then repeated to whoever asks where it got to. Nobody designed that. It accumulated. The work on this page is the removal of it, and the argument underneath is simple: one event, captured once, should be able to reach every department that needs it without a human being used as a courier.

Start with the value chain audit The point of the whole model

What actually gets built

Four things, and all four exist to produce one record.

None of this is exotic. It is the ordinary furniture of a business that can prove what it did: an app in the hand of the person doing the work, an app in the hand of the person paying for it, a checklist that captures rather than reassures, and a certificate or job card that survives the van, the site and the filing tray. What makes it worth doing is not any one of them. It is that all four write to the same record, which is what the next section is about.

The employee app

The operational truth as the person doing the work needs it: what is assigned to me, what I must capture, what I am not allowed to skip, and what happens when I mark it done.

Captures at the point of work, on a phone, offline where the signal is bad and syncing when it returns.
Carries the photograph, the signature, the meter reading, the serial number and the time, because those are what a dispute turns on a year later.
Replaces the WhatsApp thread that nobody can search and that leaves the company when the phone does.

Where the capture then goes

The client app

The same operational truth, surfaced to the other audience. Not a different version of events, a different view of one set of facts, filtered to what the client is entitled to see.

Shows status, history, documents, certificates and what is outstanding, so that the phone call asking where things stand stops happening.
Removes the most expensive failure in service businesses, which is a client who is not being ignored but has no way of knowing that.
Deliberately narrower than the employee view. Two audiences, one record, different permissions, and never two databases telling two stories.

How this reaches the sales side

Checklists that produce a record

A paper checklist produces a tick. A tick is evidence that someone held a pen. A digital checklist produces a record: who, when, where, and what they captured at each step.

Each item can require proof rather than confirmation: a reading, a photograph, a scanned asset tag, a countersignature.
Steps can be made blocking, so a job cannot be closed with the safety item or the client sign off missing.
Personal information captured this way is governed by the Protection of Personal Information Act 4 of 2013, so what is collected, who can reach it and how long it is kept are design decisions rather than afterthoughts.

Where policy and practice diverge

Digital certificates and job cards

The documents that prove work happened, issued from the record rather than typed up afterwards from memory and a photograph of a page.

Generated from what was actually captured, so the certificate and the job it describes cannot drift apart.
Numbered, timestamped, attributable and retrievable in seconds, which is the difference between an audit and an archaeology project.
Delivered to the client as they are issued, which removes the second most common reason an invoice is queried.

The evidence trail governance needs

The heart of the argument

One event should cross every departmental boundary by itself.

This is the centre of the entire model, and everything else on this site exists to make it possible. Departments are an accounting convenience. The work is not divided that way and the customer certainly does not experience it that way. But in almost every business of this size, the boundary between operations and finance, and between finance and sales, is crossed by a human being carrying a number in their head or on a page.

Follow one job. A technician closes it on the employee app: the checklist is complete, the client has signed on the screen, the parts used are recorded against the asset. In the businesses we are describing, that fact now enters a queue. Someone collects the cards on Friday. Someone types them into a spreadsheet on Monday. Accounts raises the invoice on Wednesday from the spreadsheet, occasionally from the wrong line. The account manager learns the job is finished when the client mentions it. The management accounts learn about it the following month, and by then the decision it should have informed has already been taken.

Now follow the same job through work that has been built properly. The technician closes it, and that single event does four things at once, without a person in the middle of any of them.

It raises the invoice. Priced from the record, from the parts and hours actually captured, against the contract terms that already exist for that client, and issued with the certificate attached.
It updates the sales pipeline. The account moves state, the renewal or follow up work is created, and the person who owns the relationship sees the completion before the client does rather than after.
It feeds the financial agents. The revenue, the cost and the margin on that job reach the AI financial management agents in governance the same day, so the cash position and the exception queue are describing this week rather than last month.
It closes its own record. The certificate is issued, the job is filed, the asset history is updated, and the whole thing is retrievable in one search by anyone with the right permission.

Nobody rekeyed anything. That is the entire claim, and it is deliberately modest. Davenport and Ronanki (2018) describe exactly this category of work, robotic process automation moving structured data between back office systems, as the least glamorous and most reliably useful of the three kinds of project they identify. We agree, and we lead with it for that reason. It is not the future of anything. It is the plumbing, and the plumbing is where the money is.

Two honest limits. First, this only works if the record is trustworthy, which means the capture at the point of work has to be genuinely better for the technician than the old way, or it will be worked around within a fortnight. Second, an automation that crosses a boundary carries authority with it, so the thresholds, the approvals and the exceptions have to be governed rather than assumed. A system that raises invoices unsupervised is a control failure dressed as efficiency. That is why this bucket sits on the governance foundation and not beside it, and why we build the two in that order. The full model, and why operations is a primary activity while governance is firm infrastructure, is set out on the value chain page.

The value chain audit

Before automating anything, find out what is actually there.

A value chain is the sequence of activities a business performs to turn an input into something a customer pays for, plus the functions that keep those activities running. Porter (1985) draws the distinction the audit uses: primary activities create value directly, and support activities make the primary ones possible. Most SME improvement effort goes into the primary activities, because that is where the noise is. Most of the actual constraint sits in the support activities, quietly, where nobody is looking.

The audit is an in depth review of processes, policies and procedures across the whole chain, followed in real cases rather than as drawn on a whiteboard. It exists so that we automate a process worth keeping rather than encoding a bad one permanently, which is the most common and most expensive automation failure we are called in to undo.

Primary function analysis

The path a job takes from first enquiry to final invoice, followed through real cases. Where it waits, who it waits for, how many times it is handled, and where the same information is captured twice.

See what this usually finds

Secondary function analysis

Procurement, people, technology, finance and administration. The functions that never appear in a sales meeting and that set the speed limit for everything that does. This is where the cheapest wins usually are.

Where the people side picks up

Process, policy and procedure

What is written down, what is actually done, and the distance between the two. A policy nobody follows is not a control, it is a liability, because it is evidence that the business knew what it should have been doing.

Why the gap opens

Bottlenecks

The single points through which too much passes. Usually a person rather than a machine, and usually someone senior enough that nobody has suggested taking work away from them.

Why a board notices these

Inefficiency and rework

Duplicated capture, approvals that approve nothing, reports produced for a person who left, and manual steps that exist because a system was configured for a smaller business and never revisited.

The rekeying tax, priced

Governance gaps

Delegations that were never written, spending authority that lives on trust, and decisions that cannot be reconstructed six months later. King V (2025) is applied by explaining how a board gave effect to a principle, which is impossible if nobody recorded who decided what.

How the foundation closes them

We do not replace your attorney, your labour specialist or your verification agency, and we say so where specialist advice is what you need. What the audit does is find where things sit unowned, and put a name and a date against each one so they stop arriving as emergencies. What it produces is a written picture of the chain, the constraints in order of what they cost, and a sequence for fixing them that a business can actually afford to run while still trading.

The question this bucket exists to answer

What keeps a business trapped between R20 million and R250 million.

Not ambition. Not mindset. Not the economy, most of the time. Businesses that stall in this band are usually run by capable people working extremely hard, which is precisely why the motivational answer is so unhelpful and so common. The causes are structural, they are boring, and they can be found.

Every decision routes through the founder. Quotes above a certain value, any unhappy customer, any hire, any supplier change. It worked beautifully at R8 million because the founder genuinely was the best judge of all of it. At R60 million it means the business moves at the speed of one person's inbox, and everyone else has learned to wait rather than decide.

Processes live in one person's head. How pricing really works, which clients get which terms, how the month end actually closes. It is not written anywhere because the person who does it has always done it. That person cannot be promoted, cannot take proper leave and cannot be replaced. The business is not scalable, it is memorised.

Systems built for a smaller company. The accounting file, the job cards, the spreadsheet that became load bearing, the folder structure. All of it was a sensible answer at R5 million and none of it was ever rebuilt, so the business runs ten times the volume through tooling designed for a tenth of it and absorbs the difference with people and overtime.

Compliance handled reactively. Nothing is done until a customer asks for a certificate, a regulator writes, or an employee lodges a dispute. Each event then consumes senior attention for weeks, at whatever moment it arrives. Reactive compliance is not cheaper than the planned version, it simply moves the cost somewhere you cannot budget for.

These four compound. A founder who must approve everything has no time to write anything down. Undocumented processes cannot be moved onto better systems. Bad systems make compliance manual, so it only happens under pressure, and then the founder is pulled into that too. That loop is the ceiling, and no amount of additional revenue breaks it, because more revenue simply pushes more work through the same constriction. Whether your business is in the band where this work pays for itself is set out plainly on how we choose, and the operating record behind the opinion is on the about page.

Where this work stops

Restructuring does not fix a demand problem.

This work is worth buying when the business has work and cannot get it through cleanly. It is not worth buying when the business simply does not have enough work, and the most expensive mistake we see is a company reorganising itself when the real problem sits outside the building.

If the enquiries arrive and the website loses them, through a form that fails, a page that will not load on a phone, or a site that does not answer the question the visitor came with, that is a web problem and it belongs to AutoHive Website Development.

If nobody knows the business exists, that is a demand problem, and no amount of internal automation creates an enquiry that was never going to arrive. That belongs to AutoHive Agentic Digital Marketing.

Restructuring fixes neither. We would rather route you to the right division and keep the relationship than sell you an audit that was never going to work. If you are not sure which of the three you have, say so in the first conversation and we will tell you plainly, including when the answer is that you do not need us. Where the line falls between building the sales machine and filling it is set out on the sales page, and the engagement page covers what each way in involves.

The governance foundation

Structure, constitutions, succession and the financial agents that everything built here reports into.

Read bucket one

The value chain model

Why there are four buckets, how they map onto primary and support activities, and the diagram.

Read the model

How to start

The ways in, what each involves, roughly how long it takes and what drives the price.

Read the chapter

References

Davenport, T.H. and Ronanki, R. 2018. Artificial Intelligence for the Real World. Harvard Business Review.
Institute of Directors in Southern Africa and the King Committee. 2025. King V Report on Corporate Governance for South Africa. Effective 1 January 2026.
Porter, M.E. 1985. Competitive Advantage: Creating and Sustaining Superior Performance. New York: Free Press.
Protection of Personal Information Act 4 of 2013. Republic of South Africa.

One clear next step

Find the constraint before you buy the cure.

Tell us what one job looks like from enquiry to paid invoice, including every person who touches it and every place the same information gets typed again. That single description is enough for a first conversation, and it is usually where the answer already is. You get back a written view of where the chain is losing time and money, and the order we would fix it in. If the honest answer is that the problem is demand rather than operations, that is what it will say.

Request an operations audit

Or read what each engagement involves before you decide.