In this guide
Measuring Loss Across Hundreds of Sites
Shrinkage in quick commerce is a network measurement problem before it is a counting problem. A chain of several hundred micro-fulfilment sites cannot be counted the way a warehouse is counted, because sending a team to every site on a common date costs more than the loss being measured. What has to be complete is the high-value and high-theft population at every site, since those lines carry most of the money and are where deliberate loss concentrates. What can be sampled is everything else, drawn so that each site is represented and each category is covered across the network. The distinction that matters is between the site view and the network view. A single site's variance tells you very little, because a site with unusual results may simply have had an unusual week. The same variance appearing at forty sites in one city, or at sites sharing a manager, a supplier or a shift pattern, is a finding.
Why Site-by-Site Counting Fails at Scale
Counting every site the way a warehouse is counted stops being feasible somewhere between fifty and a hundred locations, and the reasons are arithmetic rather than organisational. Coordination cost across hundreds of locations is the first. Every site needs a team, a date, an access arrangement, a system extract and a reconciliation, and the overhead per site barely falls as the number rises, so the total scales close to linearly with locations while the value per site is small. Counts on different dates are not comparable is the second and more damaging problem. A network figure assembled from sites counted across several weeks describes no single moment, and any site-to-site comparison is confounded by the different trading conditions each experienced. Stock moves fast enough in this format that a fortnight's separation makes two sites genuinely non-comparable. Travel and staffing arithmetic closes it. Counting teams have to reach sites distributed across a city or a country, and the time spent travelling between micro-fulfilment locations frequently exceeds the time spent counting in them, which means the cost per verified rupee of stock is far higher than in any conventional warehouse and rises further as the network spreads.
Sampling a Network
Once complete coverage is impractical, the sample has to be constructed rather than drawn casually, and three criteria do most of the work. Selecting sites by volume puts weight where the stock actually is, since micro-fulfilment sites differ substantially in throughput and a sample skewed toward small sites will describe the least of the network. Age matters because new sites behave differently: processes are unsettled, staff are inexperienced, and accuracy is typically worse and improving, so a sample without new sites will understate the problem the business most needs to see. Prior variance is the third criterion, and it is used to revisit rather than to punish, since a site that counted badly last cycle is where a corrective action either worked or did not. Simultaneous counting matters wherever comparability does. Sites intended to be compared against each other, or aggregated into a network figure, are counted on the same date so that the differences between them are differences in performance rather than in timing. What a sample can conclude is a statement about the network within stated limits, and what it cannot do is certify any site that was not visited.
Separating Network Loss From Site Loss
The most useful thing a network programme produces is the distinction between problems that belong to the operating model and problems that belong to individual sites, and reading it wrongly wastes a great deal of management effort. Systemic shrinkage appears across all sites at a broadly similar level, which points at something structural: a picking process that generates errors, a returns flow that is not posted, a damage rate inherent in the packaging, or a system behaviour that misstates stock everywhere. No amount of site-level management will fix it, because every site is doing the same thing and getting the same result. Outlier sites are those materially worse than the distribution, and they indicate something local: a supervision gap, a physical layout problem, a staffing pattern, or genuine loss. Those are the sites worth visiting rather than analysing. Category effects that look like site effects are the trap in the middle. A site carrying an unusually high proportion of a high-shrinkage category will show worse aggregate shrinkage without performing worse at all, and the only way to see it is to compare like categories across sites rather than site totals against each other.
Evidence That Supports a Network Number
A network shrinkage figure is only as good as the comparability of the counts behind it. Count dates matter first: sites counted weeks apart are describing different trading conditions, and aggregating them produces a number that no single date supports. Where a common date is impossible across hundreds of sites, the dates are stated and the aggregation is presented as a period figure rather than a position, which is a weaker but honest claim. Coverage is expressed as a proportion of network value rather than as a count of sites, because sites differ substantially in what they hold and counting many small ones can leave most of the value unverified. A statement that a large majority of sites were covered, alongside a much smaller proportion of value, is the pattern worth watching for. What an auditor will certify follows from both. Verified positions at named sites on named dates can be certified. A network-wide shrinkage percentage extrapolated from a sample is an estimate with stated assumptions, and presenting it as a verified figure asks the evidence to carry weight it does not have.
Running a Programme Across the Network
Choose deliberately between rolling coverage and periodic snapshots, because they answer different questions. Rolling coverage reaches every site over a period and is the only affordable way to touch a large network, but it produces no single dated position and cannot support a statement about the network on any particular day. Periodic snapshots take a smaller sample of sites on a common date, which supports comparison and aggregation properly but leaves most of the estate unvisited. Most networks need rolling coverage for control and a periodic snapshot at the reporting date for the accounts. Staffing and technology follow from the choice. Rolling coverage across hundreds of sites needs either a large field team or scanning capability that lets site staff count reliably under supervision, and the second is only credible where the counts are independently validated on a sample. An external team is worth bringing in for the reporting-date snapshot, for any site where internal counts have produced results nobody believes, and wherever the figure will be published or relied on externally. Stock audit for quick commerce is scoped from the network rather than the site.
