The Social Impact Report a Co-op Can Actually Defend: Turning Transaction Data Into Local Spend, Jobs and Food Access Numbers

Why most co-op impact reports fall apart under questioning

Most local food impact reports fail because the numbers are borrowed, not measured. A co-op quotes a multiplier from a national study, applies it to its own turnover and prints a figure it cannot trace back to a single transaction. When a funder asks how the number was produced, the trail stops at someone else's research paper.

The second failure mode is the opposite: a genuine internal number calculated once, by hand, in a spreadsheet, from an export that has since been overwritten. It was probably correct in the month it was made. Nobody can reproduce it a year later, and nobody can explain which orders were included, whether cancelled orders were stripped out, or how a producer who moved premises mid-year was treated.

A defensible impact report has three properties. Every figure traces to rows in an operational system, not to an assumption. The method is written down in enough detail that a different person running it next year gets the same answer from the same data. And the boundaries are stated openly, including what the co-op chose not to count. That last property matters more than precision. A funder rarely challenges a modest number that is honestly bounded; they challenge a large number with no visible derivation.

The five numbers worth reporting, and the fields each one needs

Five metrics cover most of what funders, members and local authorities actually ask for: total spend with local producers, percentage of revenue retained locally, producer income and price share, jobs and livelihoods supported, and food access reach. Each maps to fields a co-op platform already stores against orders, payouts and producer records.

The discipline is to define each metric as a query before you calculate it. Write down the table, the date range, the filter and the exclusions. If you cannot express the metric as a query over data you hold, you are about to estimate something, and it should be labelled as an estimate in the report.

Local producer spend: sum of settled payout lines to producers, by period, filtered to producers whose registered premises fall inside your defined local radius or administrative area.
Revenue retained locally: local producer payouts plus co-op wages, plus locally sourced operating costs (packaging, fuel, van maintenance from local suppliers), divided by gross sales.
Producer income and price share: producer payout as a percentage of shelf price per order line, plus average price per unit compared with the producer's own wholesale reference price where they supply one.
Livelihoods supported: producers with sales above a stated threshold in the period, split into full-time-equivalent bands using declared farm labour, plus co-op staff and paid driver hours from delivery records.
Food access: distinct delivery postcodes served, deliveries into areas you have classified as underserved, donated volume and value, subsidised or reduced-price box counts, and pickup points within walking distance of served addresses.

How to compute local spend and revenue retention without inventing a multiplier

Local spend is the sum of what the co-op actually paid producers and local suppliers in the period, taken from settled payout and purchase records rather than from sales. Revenue retention is that sum divided by gross sales for the same period. No multiplier is required, and none should be applied unless it is separately labelled as a modelled estimate.

The mechanics matter. Use settled payouts, not gross order value, because gross order value includes commission, delivery charges, packaging deposits and VAT that never reach a producer. Use the payout period, not the order period, and say which you used. Exclude cancelled and refunded lines, and exclude producer credits issued for quality complaints, or state that you included them. If a producer resells goods they did not grow, either separate those lines or note the limitation, because a funder who knows the sector will ask.

Defining local is a policy decision, not a data decision, and it belongs at the top of the report. A radius in kilometres from the hub, a county or administrative boundary, and a national border produce three different answers from the same rows. Pick one, publish it, and keep it stable year on year so the trend line means something. If you must change it, restate the prior year on the new definition alongside the old figure. Multipliers can still appear, in a clearly separated section, showing what the local spend figure would imply under a published regional model, with the source cited and the modelled figure never mixed into the measured one.

Turning payouts and delivery records into jobs supported and fair pricing evidence

Jobs supported is the metric most likely to be inflated and the one most worth doing carefully. Report it as three separate, clearly labelled components: co-op employment measured directly from payroll and driver hours, producer livelihoods measured as businesses reaching stated revenue thresholds, and, if you want it, a modelled full-time-equivalent estimate that is presented as a model and nothing else.

Producer livelihoods work best as banded counts rather than a single converted number. Say how many producers earned above a threshold that represents meaningful supplementary income, how many earned above a threshold that represents a substantial share of a small farm's turnover, and how many are new to the co-op this year. That statement is fully defensible: it is a count of businesses against payout totals you can show. Converting it to a jobs figure requires knowing each producer's total income across all channels, which you do not hold, so any conversion needs the producer's own declaration or must be flagged as an assumption.

Fair pricing evidence comes from the same payout data seen per line. For each order line you know the shelf price, the commission, the delivery contribution and the producer's share. Report the average producer share of shelf price by category, since it will differ between a leafy green with high spoilage and a jar of honey. Where a producer also sells wholesale and is willing to share a reference price, the gap between that price and the co-op price is the clearest direct-sale gain you can publish. Do not compute it from supermarket shelf prices you scraped; that comparison invites an argument about product grade you will not win.

Making food access measurable when you only see orders

Food access is the hardest metric to evidence honestly, because delivery records show where food went but not who could not afford it. The defensible version reports reach and provision: distinct addresses and postcodes served, coverage of areas you have classified as underserved using published deprivation or food desert data, donated volume, and counts of subsidised, sliding-scale or voucher-funded boxes.

Postcode-level analysis is the workhorse here. Aggregate deliveries to postcode districts, join those to a public deprivation or rural access dataset, and report the share of orders and volume delivered into the lower bands. Report it as coverage, not as impact: the co-op delivered a stated volume into these areas, which is a fact, rather than the co-op improved food security, which is a claim requiring evidence you do not have. Keep the aggregation coarse enough that individual households are never identifiable in a published report.

Donations and volunteering deserve their own capture path rather than being reconstructed at year end. If surplus routed to a food bank is recorded as a donation order type with the same weight and product data as a sale, the annual figure is a query rather than a memory exercise, and the food waste diverted number falls out of the same rows. Volunteer hours logged against picking or delivery sessions give you a labour contribution figure that pairs naturally with the jobs section. Both are small features that turn an anecdote into a line item.

Publishing a report that survives an auditor and a journalist

Publish the numbers with their method attached. Every headline figure should carry the data source, the period, the definition of local, the exclusions applied and whether it is measured or modelled. A one-page methodology appendix does more for credibility than another chart, and it is what an auditor, a grant officer or a sceptical journalist will turn to first.

Structure the report so measured and modelled numbers never share a sentence. Lead with the measured set: producer payouts, retention percentage, producer counts by band, postcodes served, donated kilograms, volunteer hours. Follow with a clearly headed modelled section if you want multiplier effects or estimated full-time equivalents, citing the model's source and showing the input you fed it. Close with limitations, stated plainly: what you cannot see, which producers resell, which sales channels are outside the platform, where a definition changed.

Then run it again next year on the same definitions. A single-year impact report is a marketing asset; a three-year series on stable definitions is evidence, and it is what turns a grant application or a local authority procurement conversation. This is also why the calculation belongs in the operational system rather than in a consultant's spreadsheet. Plodie holds the order lines, producer records, settled payouts and delivery data in one place, which means these figures are queries over live operational data rather than an annual archaeology project, and the same rows that pay a grower this week produce the impact number next January.

Key Takeaways

Impact numbers are only defensible when each one traces to rows in an operational system, with the period, filters and exclusions written down.
Use settled producer payouts rather than gross sales for local spend, since gross order value includes commission, delivery, deposits and VAT that never reach a grower.
Define 'local' as a stated policy (radius, county or national border), publish it, and keep it stable so year-on-year trends mean something.
Report jobs as banded producer counts plus actual co-op payroll and driver hours; keep any full-time-equivalent conversion in a separate, clearly labelled modelled section.
Capture donations, subsidised boxes and volunteer hours as data at the moment they happen, not as a year-end reconstruction.

For the environmental half of the same impact story, see our food miles report companion piece, which covers route distances, producer coordinates and waste from the same underlying order data.

Frequently Asked Questions

What is a local food multiplier and should a co-op use one in its impact report?

A local multiplier estimates how many times a euro spent locally recirculates in the local economy before leaving it, usually drawn from regional economic studies. A co-op can cite one, but it should sit in a clearly separated modelled section with the source named and the input shown, never blended into measured figures like producer payouts. Funders generally accept modelled numbers that are labelled as such and challenge ones that are not.

How often should a food co-op produce a social impact report?

Annually is the practical rhythm for a published report, aligned to your financial year so the sales and payout figures reconcile with your accounts. If the underlying metrics are queries over live data rather than manual work, run them quarterly internally so you spot a falling retention percentage or a shrinking producer base in time to act. The published version should always use the same definitions as the previous year.

Can a co-op report on food access without collecting income data from shoppers?

Yes, and in most cases it should. Reporting delivery coverage by postcode district joined to public deprivation or rural access data, plus counts of subsidised boxes, voucher redemptions and donated volume, gives an evidenced picture of reach without asking households about income. Collecting income data creates a special-category-adjacent GDPR burden that a small co-op rarely needs to take on.

What data do funders and local authorities usually ask a food co-op for?

Most commonly: total spend with local producers, number of producers supported and how many are new, geographic coverage of deliveries or pickup points, employment and volunteer contribution, and any provision to low-income or underserved households. They also increasingly ask how figures were calculated, so keeping a one-page methodology appendix with each report saves a round of follow-up questions.

How do you calculate the producer's share of the shelf price?

For each order line, divide the producer's settled payout by the shelf price the shopper paid, before or after VAT depending on your stated convention, and average it by product category rather than across the whole catalogue. Categories with high spoilage or heavy handling carry different margins, so a single blended figure hides the detail a producer or funder actually wants to see.