AdapData

Established 2015 Notes

Notes · I Strategy advisory

Value creation plans that involve automation

In short

A value creation plan sets out how an investor intends to raise the value of a company it owns. Where artificial intelligence and automation appear in such a plan, the credible version removes steps rather than adding software. The test is whether a named process becomes shorter, more accurate or less dependent on a scarce person, and whether that change is visible in a number the company already reports. Anything else is a pilot, and should be labelled as one.

Note

What belongs in a value creation plan?

A value creation plan is the investor’s account of how the business will be worth more at exit than at entry. It usually combines commercial initiatives, operational improvement, pricing, acquisitions and, increasingly, a technology component. The plan is written at entry, revised annually, and used to hold management and the board to a shared view of what is being attempted.

The technology component is the section most likely to be written in general terms. Where the rest of the plan names an owner, a baseline and a date, this section often names a category. The imbalance is worth correcting before the first hundred days rather than after them.

Why fewer steps rather than more software?

Software added to a process leaves the process intact and adds a licence, an integration and a training obligation. Value appears when steps disappear: when an approval is no longer required, when information is entered once rather than three times, when a report is no longer assembled by hand each Monday.

This is a more demanding standard than it sounds. Removing a step requires someone with the authority to accept that it was not necessary, and to carry the consequence if it proves to have been. That decision is not technical, and it is why most automation programmes stall at the pilot.

Where do the largest opportunities usually sit?

Rarely inside a function. The costly work in most companies sits between functions: at the handover from sales to operations, from operations to finance, from finance to the client. Each handover involves a restatement of the same information in a different form, and each restatement introduces delay and error.

Examining those seams is more productive than examining any single department, and it has a second advantage: nobody owns a handover, so nobody has already optimised it. The work is unpopular for the same reason, and needs a sponsor who can act across two functions.

What should an operating partner demand as evidence?

Four things, and the same four each time. A baseline measured before the change, taken from a number the company already produces. A defined scope, stated as a named process rather than a department. A control of some kind, whether a comparable site, team or period. And a named owner in the business who carries the result whether or not the technology performs.

Two claims deserve particular scepticism: time savings expressed in hours per employee per week, which almost never appear in the accounts, and accuracy measured on the examples used to build the system. Neither is dishonest. Neither is evidence.

In what order should the work be done?

The first work should be chosen for the quality of its measurement rather than the size of its prize. A modest, well-instrumented change teaches the organisation how to judge the next one, and leaves a baseline that later work can be compared against.

Sequencing then follows dependence. Processes that supply data to others are addressed before those that consume it. Where a company cannot report reliably on a process today, reporting is the first piece of work rather than an obstacle to be worked around. Plans that begin with the largest prize in the least measurable part of the business are the ones that disappear from the second-year review.

What should be retired, automated or merely assisted?

Three outcomes are available for any step, and choosing between them honestly is most of the discipline.

Retire the step where the output is no longer used, which is common in reporting and internal approvals. Automate it where it is repetitive, well specified and tolerant of a defined error rate under review. Assist the person where judgement is required and the value lies in preparation: assembling the material, drafting the first version, flagging the exceptions. Assistance is the least celebrated of the three and usually the largest.

Questions

What is a value creation plan in private equity?
It is the investor’s written account of how a portfolio company will be worth more at exit than at entry, covering commercial, operational, pricing and technology initiatives, each with an owner, a baseline and a date. It is agreed at entry and revised annually.
What is value creation in private equity?
The increase in a company's worth between entry and exit, and the programme intended to produce it. Returns are ordinarily attributed to revenue growth, margin improvement, multiple expansion and leverage, and only the first two are within management's direct control.
What are the four levels of value creation?
No single list is accepted. Returns are commonly attributed to four sources: revenue growth, margin improvement, multiple expansion and leverage. Some frameworks instead describe ascending levels of ambition, from financial engineering through operational improvement to strategic repositioning. Neither is doctrine.
What is the role of an operating partner in private equity?
An operating partner works with portfolio company management to deliver the plan, as distinct from the deal team that acquired the business. The role combines sector experience with the authority to secure investor resources and to escalate when a workstream stalls.
Where does artificial intelligence realistically contribute?
In processes that are repeated at volume, defined well enough to be judged right or wrong, and already measured. Documented handovers between functions, quotation and estimating support, and the preparation stages of expert work are the usual candidates.
How should benefits be measured?
Against a baseline taken before the change, from a number the company already reports, with a control where one exists. Savings expressed as hours per employee per week should not be counted unless they appear as reduced cost or increased output.
How long before the effect is visible?
For a well-scoped process, one to two quarters. Programmes that promise a portfolio-wide effect within a year are ordinarily describing procurement of a tool rather than a change in how work is done.
What is an example of value creation through automation?
An illustration rather than a client engagement: a distributor quotes by hand and takes two days to respond. The quotation step is specified, historical quotations are used to prepare a draft for review, and response time falls to hours. Win rate is the measure, and it was already reported.
How is artificial intelligence being used in private equity?
At firm level, for document review in diligence, screening of origination material and preparation of reporting. At portfolio level, in specific operational processes such as quotation, scheduling, service records and customer correspondence. Firm-level use is more widespread; portfolio-level use produces the larger effect.

Engagement

We are usually engaged by an investor or an operating partner who wants the technology section of a plan written to the same standard as the rest of it. Research, technical origination and development are carried out by our research partner. Engagements are senior-led and conducted in confidence. We reply personally to every enquiry; write to [email protected].

Related