Skip to content

Cost Take-Out for the Enterprise

Cost taken out of the structure rather than the headcount: governed information, orchestration above the applications, and a baseline before the spend.

Of workers say they spend too much of the day searching for information
62%Of workers say they spend too much of the day searching for informationMicrosoft, 2023
Say they do not have enough uninterrupted time to focus
68%Say they do not have enough uninterrupted time to focusMicrosoft, 2023, same survey
Of organisational data estimated to be unused
55%Of organisational data estimated to be unusedSplunk, 2019

Nobody approves this spend, because nobody can see it.

Two findings from one survey of 31,000 workers across 31 markets. Neither appears on a budget line anywhere.

62% say they spend too much of the workday searching for information

Not a minority complaint and not a training problem. It is what happens when information sits somewhere a person can eventually find and a system cannot reach.

68% say they do not have enough uninterrupted time to focus

The same estate produces both findings, from the same respondents. Every failed search becomes a question to a colleague, and every question is an interruption at the other end of it.

Microsoft, Work Trend Index 2023 Vendor research with a disclosed method: 31,000 workers, 31 markets, fielded February to March 2023. It measures how many people report the problem, not how many hours are lost; nothing recent measures the hours.

The three costs nobody has a line item for

The usual pattern is a round of discretionary cuts, a reported saving, and a run rate back at its old level inside three years.

The spend that was removed was the spend that was visible, and the visible spend is rarely the structural one. Underneath it sits a second category, larger and almost never itemised: three costs, each real, each paid every month, and none of them a row anyone owns.

  1. Storing what you do not need

    Content that is redundant, obsolete or trivial is backed up, replicated, indexed, searched, secured and audited exactly like the content that matters, because no rule distinguishes them. The bill arrives as infrastructure, so it is managed as infrastructure, and an infrastructure team cannot delete what it is not authorised to judge.

  2. Moving information by hand

    Finding the current version, asking who approved it, copying a figure from one system into another. This is not slack. It is work that exists only because the systems were bought separately and never made to talk, and it scales with the number of systems rather than with the volume of business.

  3. Proving compliance twice

    When retention, access and evidence are not enforced by the systems themselves, they are reconstructed by people at audit time. That is a recurring tax, paid in the time of the most senior staff, and it grows with each new obligation rather than being absorbed by the last one.

A cost with no owner is a cost with no advocate for removing it, which is why the last programme took out travel and contractors and left these three exactly where they were.

Governed information is the cost lever

Almost everything above resolves at the same layer, which is why the work is one programme rather than three.

  1. Once

    Retention is defined and applied

    Then

    Storage, backup, indexing and compute fall together

    Content that has passed its rule can be disposed of defensibly, with legal hold respected. The bill stays down, because the rule keeps applying to what arrives next month.

  2. Once

    Information has an owner and a classification

    Then

    Retrieval becomes a system function rather than a human one

    That is where the recovered time comes from, and it is the same change that makes AI worth deploying: a model is only as good as the records underneath it.

  3. Once

    Controls run continuously

    Then

    Evidence is a query rather than a project

    The audit stops being an event, and the compliance work that was being done twice is done once, by the system, as a by-product.

ISO 15489

Structure, not a discount

The reason savings come back is architectural. Where the process logic lives decides whether a cut holds.

Process logic inside the applications

  • Every application you own holds a piece of how the business runs.
  • Every change is negotiated with the system that happens to contain it.
  • Point-to-point integrations that have to be maintained forever.
  • Nothing can be replaced without a project.

Process logic and governance above them

  • Systems plug into a defined structure rather than owning it.
  • Integrations stop being bespoke.
  • Keeping, replacing or retiring a platform is an ordinary commercial decision.
  • The next system you buy adds to the structure, not to the pile.

This is also what makes the saving durable. A discount is renegotiated. A structure that requires less work does not need renegotiating.

A first figure, in your own numbers

Cost take-out in information management comes from four places: content stored past its purpose, systems kept alive for what might be inside them, invoices handled by people, and time spent looking for documents that exist. Put your figures in and see where the money is. The assumptions behind the estimate are underneath, and you can change any of them.

Your estate

The assumptions behind the estimate

Every figure here can be changed. If you know yours, use it. The defaults are placeholders, not benchmarks.

What we do not promise

Large transformation programmes have a well-documented tendency to overrun, and the overruns are not symmetrical: most land near plan, a minority land catastrophically far from it. Any number quoted before the work is scoped is a range, and anyone giving you a single figure is either guessing or selling.

So we do not open this page with a savings percentage. The ones in circulation are either a vendor’s average across engagements nobody can inspect, or a survey of what executives estimate about their own organisations. How much comes out of yours depends on how much redundant content you are holding, how manual your transactional processes are, and how much of your compliance effort is currently reconstructive. All three are measurable in weeks, before anything is committed. That measurement is the first piece of work, and it is deliberately small.

Flyvbjerg, 2022

What you get

Four pieces of work, not four promises. Each one names the work and how you would know it had happened.

  1. A number before the programme, not after it
  2. Content disposed of defensibly, and the bill that follows it
  3. The structure that stops it coming back
  4. Visibility while it happens, and the same work paying twice

A number before the programme, not after it

The baseline, and it is deliberately small.

A few weeks against your actual estate rather than a questionnaire: what you are storing and what it costs to keep, which processes are still run by hand, and where compliance work is being done twice. Each source of leakage gets a figure and the list gets ranked.

That ranking is the point. It means the sequence of the programme is argued from evidence rather than from whoever is loudest in the room, and it means a board is approving an order of work rather than a promise.

It costs a fraction of the programme it corrects, and if the honest answer is that your estate is in better shape than you feared, that is cheaper to learn here than eighteen months in.

Content disposed of defensibly, and the bill that follows it

The fastest money, because it depends on nothing else changing.

Retention defined and then applied, so content that has passed its rule can go, with legal hold respected and the disposal evidenced. Storage, backup, replication, indexing and the compute running across all of it fall together, because they were all paying for the same material.

It stays down, which is the part that separates this from a clean-up. The rule keeps applying to what arrives next month, so the estate does not refill quietly the way it did after the last exercise.

This is the same classification and retention work the group does for contracts, personnel files and asset documentation. One set of rules for the whole estate, not a second set shaped like a cost programme.

The structure that stops it coming back

Because a saving that is not structural is one you pay for again.

Governance and process logic above the applications rather than inside them. When every application holds a piece of how the business runs, every change has to be negotiated with whichever system happens to contain it, and every connection is a bespoke integration somebody maintains forever.

Inverting that makes the next system you buy plug into a defined structure instead of adding to the pile, and it makes keeping, replacing or retiring a platform an ordinary commercial decision rather than a hostage negotiation.

A discount gets renegotiated at the next renewal. A structure that requires less work does not need renegotiating, which is the whole difference between cost take-out and a budget cut.

Visibility while it happens, and the same work paying twice

Tracked against the baseline, and useful to a second business case.

Value realisation measured against the baseline as the work lands, per process rather than claimed for the programme, so a stalled workstream is visible in the month it stalls instead of at the annual review. That is the discipline the previous programme was missing when its savings quietly reversed.

And everything cost take-out requires is what AI readiness requires: classified content, applied retention, records a system can reach, and a trail behind what it used. The two pages describe one programme with two business cases on the front.

Which means you do not have to choose which to fund. If the AI use case is dropped, the estate is still cheaper to run and easier to govern than it was, and if the cost case is deferred, the readiness work has already been done.

The first step is not a purchase. It is a baseline: a short assessment that tells you where the money is going and in what order to take it out. It costs a fraction of the programme it corrects, and it is the one piece of work that makes the rest defensible to a board.

Get a cost baseline

Cost take-out, asked plainly

  • What is cost take-out?

    Cost take-out is removing cost from the structure of how an organisation runs, rather than from its budget lines. A budget cut reduces spend and leaves the work in place, which is why it tends to come back. Cost take-out changes what has to be done at all: less data to store and protect, fewer manual handoffs per transaction, fewer integrations to maintain. The saving persists because the work that generated it no longer exists.

  • Where does the money actually come from?

    Three places, in roughly this order. Storage, backup and compute spent on content nobody has classified. Time spent finding and rekeying information between systems, which 62 per cent of workers report as too much of their day (Microsoft, Work Trend Index 2023). And the cost of compliance done reactively through manual audits rather than through controls that run continuously. None of the three is a line item, which is why they survive most cost programmes. We do not publish a headline percentage for the total, because the honest answer depends on your estate and anyone quoting you one before measuring it is guessing.

  • Is this a redundancy programme?

    No, and it is worth being direct about it. The savings above come from storage, licensing, integration maintenance and recovered time. Time recovered is time returned to work people were hired to do rather than headcount removed. Where a client does intend to reduce headcount, that is their decision and their programme; ours makes the operating model capable of it, which is different and should not be presented as the same.

  • How long before we see anything?

    The baseline is weeks, not months. Disposal of redundant and obsolete content usually produces a measurable storage reduction in the first quarter, because it does not depend on any other system changing. Process savings follow the sequence of the processes themselves and show up per process as each is done. Structural savings from orchestration are the slowest and the most durable.

  • What if we have already run a cost programme?

    Most organisations have, and the pattern is consistent: the previous programme took out discretionary spend and left the structure intact, so the run rate recovered within two or three years. That is worth knowing rather than hiding, because it tells you where to look. The costs that survived the last exercise are the ones nobody could see, and those are the ones a baseline against the actual estate finds.

  • Do we have to replace our existing systems?

    No. The approach is deliberately neutral about applications: the governance and the process logic sit above them, which is what allows a system to be replaced later without the surrounding work being redone. Replacing systems is expensive, disruptive, and usually optional; the point is to stop the architecture from making that decision for you.