Zepth Core · Site Operations

Materials

You cannot miss what you never logged.

Last updated

Zepth Core module

Materials

AI agent built into the module
Per-project stock with an audit trailIssue-to-site workflowsRequisition-checks-stockReorder points from arithmetic

Overview

Site inventory tracks materials from the moment they come through the gate to the moment they are installed: what is in the laydown, where it is, what it is reserved for, and who it was issued to.

It is the other half of procurement, and the half that gets no attention. Our procurement module buys the materials. This one keeps them — and keeping them turns out to be where they are lost.

Why sites lose materials they already own

Three failures, and they compound.

The first is the re-order loop. Stock arrives, goes into a laydown, and becomes invisible — so when the next requisition is raised, nobody checks, because there is nothing to check. The material is ordered again. And at closeout the project discovers it owns a great deal of cable it never needed, purchased twice, now worth its scrap value. A requisition that does not interrogate stock first is not a requisition. It is a habit.

The second is theft, and the mechanism is exactly the sentence at the top of this page: you cannot miss what you never logged. Unrecorded stock cannot go missing, in the strict sense that its absence produces no signal anywhere. Construction site theft is estimated in the hundreds of millions to around a billion dollars a year in the US alone, with under a quarter of it ever recovered — a figure we set out in full on the delivery-notes page. But the estimate is not the interesting part. The interesting part is that a site with no inventory record cannot even participate in the statistic, because it will never know.

And the third is waste, which is the quietest. Material waste runs above ten per cent of purchased quantity on a typical project — the figure sits on our material-procurement page — and site studies routinely find actual wastage at roughly double the allowance carried in the estimate. But waste is only measurable if consumption is attributed to something. Issue material against an activity and the overrun is visible. Issue it as “some rebar for level three” and the overrun is simply the way things are.

What site inventory actually requires

  • Issue-to-site discipline — against a work order and a location, never against a vibe. “Some rebar for level three” is not an issue record. It is a story about rebar. Material drawn against a specific activity and location is what makes consumption per activity computable, and consumption per activity is the only thing that makes waste measurable. Without it, every overrun is invisible and therefore normal — and a site where waste is normal will not reduce it, because it has never seen it.

  • The requisition must interrogate stock first. This is a one-line control and it closes the whole re-order loop. Before a purchase requisition goes out, the system checks what is already in the laydown — and, ideally, in the laydowns of the other projects. The alternative is a portfolio quietly buying, twice, materials it is simultaneously writing off as dead stock somewhere else.

  • Secure storage tiers by pilferability, because not everything walks equally. Copper, cable and power tools disappear. Aggregate does not. Treating both to the same storage regime means over-securing the sand and under-securing the thing that is actually worth stealing — which is how a site ends up with an impressively locked compound and no cable in it.

  • Storage conditions can void an acceptance you already have. This is the one that catches people. Cement ages, so FIFO is not housekeeping, it is compliance. Waterproofing membrane left in UV degrades. Rebar in ground contact contaminates. Material that passed its inspection on arrival can fail on the day it is installed — and the inspection record that said it was acceptable will still be sitting there, correct and useless, describing a condition that no longer exists.

  • Reorder points come from arithmetic, not from instinct. Consumption rate multiplied by replenishment lead time, plus a buffer. Set the alert on that, and it fires when it should. Set it on a low-stock feeling, and it fires when somebody notices — which is generally after the front has stopped.

  • Owner-furnished materials carry the same discipline and more exposure. You did not buy them, and you are holding them. Custody, insurance, condition on receipt, and a record of exactly what arrived — because when an owner-furnished item is damaged or missing, the conversation about who is responsible will be reconstructed entirely from documents, and the party without documents will lose it.

How the shrinkage conversation should go

Received, minus issued, minus what is physically on the shelf, equals shrinkage. It is arithmetic, and it is the only honest starting point.

Which means the count has to happen. Cycle-count by ABC frequency — the high-value, high-pilferability classes often, the aggregate rarely — rather than one enormous annual count that nobody believes and everybody dreads.

And when the number comes out non-zero, investigate it rather than absorbing it. A shrinkage figure that is quietly written off each month is not a control; it is a subscription.

How Zepth runs site inventory

Per-project stock with a full movement audit trail — received, reserved, issued, returned — and issue-to-site workflows that draw material against an activity and a location rather than against a name and a nod.

Low-stock alerts computed from consumption and lead time. Reservations against work fronts, so the material that is spoken for is visibly spoken for. And the whole record joins to the delivery notes that brought the material in and the inspections that accepted it — which is what turns three separate systems into one answerable question.

The value

Why it matters

Materials already owned are visible before the next requisition goes out — so the site stops buying its own cable twice.

Consumption is attributed to activities, which is the only thing that makes waste measurable rather than normal.

Unrecorded stock stops being the norm, so theft produces a signal instead of a silence.

Storage conditions are tracked, so material does not fail on the day it is installed against an inspection record that still says it passed.

Capabilities

What you can do

01

Per-project stock with an audit trail

Received, reserved, issued, returned — every movement attributable, which is the precondition for noticing anything at all.

02

Issue-to-site workflows

Material drawn against a work order and a location, with a signature. Not against a name and a nod.

03

Requisition-checks-stock

The one-line control that closes the re-order loop, before the purchase requisition leaves the building.

04

Reorder points from arithmetic

Consumption rate × replenishment lead time + buffer — so the alert fires when it should rather than when somebody notices.

05

Storage and shelf-life tracking

FIFO on cement, UV exposure on membrane, ground contact on rebar. Acceptance is a condition, not a certificate.

06

Owner-furnished custody

Condition on receipt, custody and insurance — because that conversation will be reconstructed from documents, by people who were not there.

The workflow

How it actually runs

  1. 1

    Receive to a location, off the GRN

    The goods-received note says what was accepted. Inventory says where it went. A material with no bin is a material with no future.

  2. 2

    Reserve against work fronts

    So that material which is spoken for is visibly spoken for — and the next requisition can see what is genuinely available rather than what is merely present.

  3. 3

    Issue against an activity, with a signature

    Against a work order and a location. This is the step that makes consumption — and therefore waste — a measurable thing rather than a fact of life.

  4. 4

    Cycle-count by ABC frequency

    The valuable and the stealable, often. The aggregate, rarely. Not one annual count that everybody dreads and nobody trusts.

  5. 5

    Reconcile, and investigate the difference

    Received minus issued minus stock equals shrinkage. A shrinkage number that is written off rather than investigated is not a control — it is a subscription.

AI that does the work

How AI changes Materials management.

Duplicate-requisition detection.

A purchase requisition raised for material that is already sitting in a laydown — on this project or another one. This single check pays for the module, and it pays for it repeatedly.

Shrinkage patterns by material class.

Not the total, which tells you nothing, but the shape: which classes, which locations, which weeks. Theft has a pattern; absorption of a monthly write-off does not reveal it.

Dead-stock candidates for transfer.

Material sitting unused on one project that another project is about to buy. The portfolio view is the only altitude from which this is visible, and it is invisible from inside either project.

Plain-language portfolio stock queries.

“How much cable tray do we actually have, across every site?” Which is a question that is trivially answerable and almost never answered.

The engineer’s judgment stays in charge; the AI removes the latency and the blind spots.

Best practices

  • Make the requisition check the stock. One control, and it closes the entire re-order loop — which is otherwise a portfolio buying materials twice while writing the same materials off as dead stock elsewhere.
  • Issue against activities, never against people. Consumption per activity is what makes waste visible, and waste that is not visible is waste that is normal.
  • Tier your storage by what actually walks. Copper and cable and power tools disappear; sand does not. An impressively secured compound with no cable left in it is a real outcome.
  • Treat FIFO on cement as compliance, not housekeeping. Material can fail on the day it is installed while its inspection record still says, accurately, that it passed on arrival.

Dashboards & reporting

Stock on hand by project, location and class, with the full movement trail behind every line. Consumption per activity — which is what turns waste from a fact of life into a number. Shrinkage, by class and location, reconciled rather than absorbed. Dead stock and transfer candidates across the portfolio. Reorder alerts against real consumption and real lead times. And the join back to the delivery notes that brought the material in and the inspections that accepted it.

Live dashboards
Drill-down & filters
Export to Excel / PDF
FAQ

Common questions

How do you stop ordering materials you already have?

Make the requisition check the stock before it leaves the building — on this project, and ideally across the portfolio. It is a one-line control and it closes the entire re-order loop. Without it, stock in a laydown is invisible, so nobody checks, so it is ordered again, and at closeout you discover you own a great deal of cable you never needed and purchased twice.

What is issue-to-site?

Drawing material against a specific work order and location, with a signature — rather than “some rebar for level three”, which is not a record but a story about rebar. It matters because consumption attributed to an activity is the only thing that makes waste measurable. Unattributed, every overrun is invisible and therefore normal, and a site that has never seen its waste will not reduce it.

Read the full answer
How often should site stock be counted?

By ABC frequency — the high-value and easily-stolen classes counted often, the bulk aggregate rarely. Not one enormous annual count that everybody dreads and nobody believes. And the reconciliation is arithmetic: received, minus issued, minus what is on the shelf, equals shrinkage. Investigate that number rather than absorbing it, or it becomes a subscription rather than a control.

How do you calculate a reorder point?

Consumption rate multiplied by replenishment lead time, plus a buffer sized to how badly a stockout would hurt. That is arithmetic, and the alert fires when it should. The alternative is an alert that fires when somebody notices the pile looks small, which is reliably after the work front has already stopped.

Who is responsible for shrinkage?

Whoever holds custody — but the honest answer is that a site with no inventory record has no shrinkage, in the sense that it produces no signal. You cannot miss what you never logged. That is precisely why the record comes first: not to catch anyone, but so that an absence produces a number rather than a shrug.

Read the full answer
Can stored materials fail inspection after they were already accepted?

Yes, and this is the one that catches people. Cement ages. Waterproofing membrane degrades in UV. Rebar in ground contact contaminates. Material that passed on arrival can fail on the day it is installed — and the inspection record saying it was acceptable will still be there, entirely correct, describing a condition that no longer exists. Acceptance is a condition, not a certificate.

Read the full answer

Sources

  • Site-theft estimates (NER/NICB) — anchored with their figures on /modules/delivery-notes/. Referenced here rather than re-stated: they are estimates, and one estimate should appear once.
  • Material-waste research — anchored on /modules/material-procurement/. Site studies commonly find actual wastage at roughly double the allowance carried in the estimate.
  • FMI research — time lost hunting for project information. Anchored on /modules/document-register/.
  • This page carries no stat band on purpose. Every figure it would want is already cited on the page that owns it, and repeating a number across pages presents one finding as several.

Zepth is the construction project delivery platform — it runs construction, procurement and asset management on one record, and does the work: reading the drawings, reviewing the submittals, matching the invoices and flagging the risks, with a human sign-off on anything consequential.

See it on your project.

A short, tailored walkthrough on your real workflow — no generic demo.