The Co-op Dashboard Nobody Builds: 9 Metrics a Short Food Supply Chain Administrator Should Track Across 38 Producers

Why co-op-level analytics falls into a gap nobody builds for

Agritech analytics splits into two camps: farm management tools that measure yield, soil and irrigation for a single holding, and enterprise supply chain platforms built for pallet volumes and national distribution. A 38-producer co-operative sits between them, and its most valuable dataset, the one linking producers to shoppers to deliveries, has no default dashboard.

The gap matters because the co-op administrator is the only person in the chain who can see across producers. A grower knows their own lettuce sold out on Thursday. They do not know that four other growers also listed lettuce that week, that the co-op oversupplied by 40 heads, and that the shoppers who wanted spinach found none. Only the aggregated view answers that, and answering it is the administrator's actual job: not growing, not driving, but deciding what the co-op should offer, from whom, and in what quantity.

The other reason this dashboard rarely exists is that most co-ops never had the data in one place. When orders arrive by WhatsApp, prices live in a spreadsheet and payouts get reconciled by hand, there is no queryable history. The metrics below assume you have moved to a system that records every listing, order line, substitution and delivery outcome. That is the precondition, not the insight.

The nine metrics, grouped by the decision they drive

A co-op dashboard should be organised by decision, not by data source. Nine metrics cover the four recurring decisions an administrator makes: what to list, who to source from, how much to promise, and where the operation is leaking money. Each one should be readable in under ten seconds and drillable to a producer or product.

Supply-side metrics tell you whether your producer base can deliver what you sell. Demand-side metrics tell you what shoppers actually want versus what you happen to be offering. Fulfilment metrics tell you whether the promise survived the van. And the mix metric, unsold rate by producer and by category, is the one that most directly converts into money left on the table or thrown away.

1. Fill rate by producer: percentage of committed units a producer actually delivered on the day. The single best predictor of whether you can build a box around them next week.
2. Listing consistency: how many weeks out of the last twelve a producer posted stock at all. Distinguishes a reliable weekly supplier from an occasional seasonal one.
3. Producer concentration by category: what share of your tomato volume comes from your top supplier. Above roughly 60% in a staple category, one hailstorm empties your shelf.
4. Category sell-through rate: units sold divided by units listed, per category, per week. Your clearest signal of over- or under-supply.
5. Unmet demand signals: searches, out-of-stock views and abandoned baskets for items you had zero stock of. The metric that tells you what to recruit producers for.
6. Basket composition drift: how the average order's category split moves week to week. Catches seasonal turns two to three weeks before your gut does.
7. Repeat purchase rate by cohort: share of shoppers from a given signup month who ordered again within 30 days. The health check on the whole operation.
8. Substitution and shortfall rate: how often an order line was swapped, reduced or refunded after the shopper paid. Where trust quietly erodes.
9. Delivery success on first attempt: percentage of drops completed without a redelivery, a return to depot or a cold-chain write-off.

How to read supply-side metrics across 38 producers without accusing anyone

Fill rate, listing consistency and producer concentration are diagnostic, not disciplinary. A producer with an 80% fill rate is not unreliable; they are probably a small grower who lists optimistically. The correct response is to change how you plan around them, not to send a warning email. Sourcing decisions improve, relationships stay intact.

In practice, this means segmenting producers rather than ranking them. Anchor producers with high fill rate and high listing consistency are the ones you can build a standing weekly box around. Seasonal specialists appear for six weeks a year with something nobody else has, and their low consistency score is a feature. Volatile suppliers with strong products but erratic delivery belong in surplus listings and add-ons, not in the guaranteed box. The dashboard's job is to tell you which bucket each of your 38 sits in this month, because the buckets move.

Producer concentration is the metric administrators most often skip and most often regret skipping. If one grower supplies most of your potatoes, the risk is not just weather. It is a hospital stay, a broken tractor, or a better offer from a wholesaler. Tracking concentration by category converts a vague worry into a recruitment target: you need a second potato grower before September, not eventually.

Turning demand data into product mix changes, not just charts

Demand metrics only earn their place if they change next week's listing. The workflow is short: read sell-through by category, compare it to unmet demand signals, then adjust either the quantity you accept from producers or the producers you recruit. A category with 98% sell-through and repeated out-of-stock views is a sourcing gap, not a success.

Sell-through above roughly 95% consistently means you are leaving revenue on the table and, worse, teaching shoppers that your leafy greens are usually gone by Tuesday. Sell-through below 70% means someone's crate is going home or into the compost, and if the same producer appears in that number three weeks running, the honest conversation is about volume, not quality. Neither number is actionable alone. Paired with unmet demand, they turn into a specific instruction: ask two growers for 30% more spinach, tell the third to hold back on courgettes.

Basket composition drift is the quiet one. Watch the category share of an average order rather than absolute volume, and seasonal turns show up early. When the salad share starts sliding and root vegetables tick up while total order value stays flat, shoppers have decided autumn started. That is your cue to reshape the standard box and to tell growers what you will be buying, ideally before they have committed their own week.

Forecasting co-op-level seasonal swings from your own history

A co-op does not need machine learning to forecast. It needs two or three seasons of its own order history, indexed by ISO week, plus a note of what disrupted each anomaly. Comparing this week to the same week last year, adjusted for how many active shoppers you have now, gets you most of the way to a usable purchase plan.

The practical method is to normalise. Raw order counts grow as the co-op grows, so a 20% year-on-year rise tells you nothing about demand per shopper. Divide by active shoppers in that week and you get a per-shopper baseline that survives growth. Then layer on the local calendar that generic forecasting tools never know about: school holidays, the week of the town festival, a public holiday that falls on a delivery day, and the fortnight when everyone's own garden produces more courgettes than they can eat and your courgette sales collapse.

Keep an annotation log alongside the numbers. Every unexplained dip that you fail to write down becomes a permanent mystery in your history and a distorted forecast next year. Two lines in a shared document, "week 31: heatwave, three producers skipped harvest", is enough to keep the following year's plan honest.

Building the dashboard: what to instrument first and what to ignore

Start with four metrics, not nine. Fill rate by producer, category sell-through, substitution rate and first-attempt delivery success cover the failures that cost you money and customers this month. Add the demand and forecasting metrics once you have a full season of clean data behind them, because a trend line built on six weeks is noise with a slope.

The instrumentation requirement is stricter than most co-ops expect. Fill rate needs a recorded commitment separate from the recorded delivery, so a producer's "I'll bring 20 kg" has to exist as data before the crate arrives. Substitution rate needs the original order line preserved rather than overwritten when the packer swaps an item. Unmet demand needs the storefront to log a zero-stock view, which nothing captures retroactively. Each of these is a small decision in how the system records events, and each one is impossible to backfill.

Equally important is what to leave off. Vanity metrics like total registered producers, cumulative orders and page views do not drive a decision an administrator can act on this week. Neither does anything you cannot break down by producer or category, because the whole value of the co-op dataset is that it can be sliced by the people and products it describes. If a number cannot be attributed to a producer, a category or a delivery route, it belongs in an annual report, not on the operating dashboard.

Key Takeaways

Farm software measures one holding and enterprise supply chain software measures pallets; the co-op-level view across producers, shoppers and deliveries is the dataset nobody builds a dashboard for.
Nine metrics cover the four decisions an administrator makes weekly: what to list, who to source from, how much to promise, and where money leaks.
Supply-side metrics like fill rate and listing consistency should segment producers into anchors, seasonal specialists and volatile suppliers, not rank them.
Sell-through paired with unmet demand signals turns charts into a specific instruction: buy 30% more spinach, hold back on courgettes.
Forecast from your own history normalised per active shopper, annotated with local events, and instrument commitments and substitutions from day one because they cannot be backfilled.

None of these metrics exist until the order sheets leave the spreadsheet, which is what our 6-week migration plan for moving a co-op's order sheets, price lists and routes into software is designed to fix.

Frequently Asked Questions

How many producers does a co-op need before analytics is worth the effort?

Around eight to ten producers is where the aggregated view starts beating intuition. Below that, an administrator can hold every grower's reliability and every product's demand in their head. Above it, overlapping supply in the same category and unnoticed concentration risk begin to cost real money, and the patterns are no longer visible without recording them.

What is a good fill rate for a small local producer?

There is no universal benchmark, and treating a published one as a target usually backfires with smallholders whose harvest depends on weather. What matters is each producer's own stability: a grower who consistently delivers 85% of committed volume is more plannable than one swinging between 60% and 110%, because you can simply discount their commitments by a known factor when building the box.

Can I build this dashboard in a spreadsheet?

You can build the reporting layer in a spreadsheet, but not the capture layer. The hard part is recording producer commitments, original order lines before substitution, and zero-stock views as structured events at the moment they happen. If orders still arrive by phone and message, that data never exists, and no amount of spreadsheet work recovers it later.

How do I share these metrics with producers without making them defensive?

Share each producer their own numbers plus anonymised category-level context, never a public leaderboard. A grower who sees that their category ran at 68% sell-through across all suppliers understands a request to reduce volume as market information rather than criticism. Individual fill rate is best used in a planning conversation about realistic commitments, not in a performance review framing.

Which of these metrics matter most for a co-op selling to restaurants rather than households?

Fill rate, substitution rate and first-attempt delivery success become far more important, because a chef who receives 60% of an order has a menu problem that day. Demand metrics matter less because restaurant orders are typically standing and negotiated in advance, so the analytical weight shifts almost entirely to supply reliability and fulfilment accuracy.