Where the money actually goes on a project.
A development project settles its material price months in advance, then loses margin to things no budget line owns.
The allowance you already budget
Every rate analysis carries a wastage factor, and everyone in the room agrees it is normal. In Indian practice those allowances run to roughly 10% on river sand, M-sand and site-mixed concrete, 8% on clay bricks, 5% on cement and 3% on TMT steel, following CPWD DSR.
The point is not that these numbers are wrong. It is that they are the floor. Rework sits on top of them, running at about 5% of project cost on average and 12–15% on a bad job, according to the Construction Industry Institute’s RS-153. Materials are 40–60% of project cost to begin with, so a few points of avoidable loss on top of an already-negotiated rate is a larger number than most price renegotiations will ever recover.
None of this appears as a line item, and nobody owns it. That is what makes it survive year after year.
Five bottlenecks
Each of the costs above traces back to the same thing: finding out late. These are the five places where a project loses the time it needed.
- 01
The shortfall is found on site
Nobody knows the material is short until the work in front of it stops. Every other cost on this page starts here.
- 02
Three to five days to confirm one order
A manager calls a dealer, the dealer rings three yards, each quotes blind. The answer arrives after the window that needed it, and the yard that lost learns nothing about why.
- 03
Nobody re-sequences the freight
A two-day slip on one delivery is not matched by a change to the dependent trades. The pour window is missed, the next is seven days out, and the crew is paid to wait.
- 04
Approvals wait for a person to be free
Prices, discounts and substitutions queue behind whoever can sign, in a chain that moves in days while the site moves in hours.
- 05
The record is in a messaging thread
What was promised, what arrived and who agreed to it live in chat. A dispute then takes two clerks several days to reconstruct from evidence nobody filed.
Two worked scenarios
Illustrative.
A two-day slip that would have cost a week
- Situation
- Rebar fixing for a podium slab is sequenced behind a steel delivery. The supplier’s dispatch slips by two days.
- Today
- The site learns of it on the morning of the pour. The window is missed, the next is seven days out, and the crew is paid to wait.
- With Zelleo
- The slip is detected against the programme. The agent re-asks the panel for the quantity in the new window, proposes a re-sequence of two dependent activities, and the project manager approves. The pour moves by half a day.
- Measured by
- Slip per pour window, and crew idle hours against the baseline.
Ordering against the programme, not the gap
- Situation
- A high-reorder trade with several yards on the approved panel and a price ceiling per grade.
- Today
- A shortfall is spotted on site, a dealer is called, three yards quote blind, and the order lands three to five days later at whatever the spread allows.
- With Zelleo
- The requirement is computed from the programme and sent to the panel at once. Offers inside your ceiling are accepted automatically; anything outside it stops and waits for a person.
- Measured by
- Waste above the allowance, and time from requirement to confirmed order.
The supplier’s side
The same coordination gap costs the yard at the other end of the call. Up to 40% of truck time in India runs empty, on NITI Aayog and RMI figures, and the driver, the diesel and the loan are paid either way. Confirming a single order takes three to five days, and a yard that quotes blind and loses is told nothing about why.
When a builder’s schedule moves, Zelleo asks the yard directly, with the grade, quantity, delivery window and site access already attached. The yard accepts, counters or declines against rules it wrote once, and the freight attaches to the accepted slot in the same step. Price sits inside a floor the yard sets and is one of six inputs in the allocation, alongside distance, stock on hand, the window promised, the grade required and committed load. Four of those six improve by operating well rather than by discounting.
A dealer who carries credit, breaks bulk, holds stock and absorbs quality disputes is doing real work and keeps their margin. What the system removes is the hand paid purely for the introduction.
The ninety-day pilot
One project, one trade, measured against your own last project rather than our forecast.
- Days 1–14BaselineFive waste numbers from your last project, set with your own quantity surveyor in one afternoon. Nothing else changes.
- Days 15–60Run liveOne trade on one site. Your panel, your limits, allocation on every load, with your usual process running beside it.
- Days 61–90CompareThe same five numbers, measured by the same surveyor, against the baseline you gave on day fourteen.
| Measure | Typical today | Pilot target | Status |
|---|---|---|---|
| Rework | 5% of project cost | 3% | To be measured in the pilot |
| Waste above the allowance | 8% of ordered quantity | 5% | To be measured in the pilot |
| Inbound freight running empty | 40% of truck time | 32% | To be measured in the pilot |
| Slip per pour window | 2 days | Half a day | To be measured in the pilot |
| Materials as a share of cost | 40–60% | — | Established |
| Rework, worst case | 12–15% of project cost | — | Established · CII RS-153 |
We choose a high-reorder trade deliberately. Ninety days of fixtures gives four data points; ninety days of steel or aggregates gives forty.
What we believe
Construction procurement did not get slow because the people in it are slow. It got slow because the chain is human at every link, and a human chain cannot re-sequence three suppliers and a transporter in an afternoon. Every attempt to fix that with another dashboard has added a screen for someone to check rather than removing the checking.
What changes now is that software can hold a negotiation rather than record one. A schedule becomes a demand signal, an agent can ask a panel at once instead of a dealer calling three yards in sequence, and the answer arrives before the shortfall rather than after it.
We think the firms that adopt this first will not describe it as AI. They will describe it as having stopped losing the week.
Stated plainly. Zelleo is at pilot stage. The established rows above are third-party figures with their sources named; the targets are what a pilot is designed to settle on your project, measured by your own surveyor.