Design supplier data requests around a clear reporting purpose

Before a supplier receives an ESG questionnaire, test what each requested field will change. A collection round can become difficult to answer because it mixes planning, calculation and supplier assessment in the same form. Give each round one approved output, then decide what the buyer already knows, what it needs now and what depends on a later decision.

This guide is for procurement and reporting teams designing a request before it is sent. The original fictional exercise below prepares an electricity-record availability review. It does not calculate emissions, rate suppliers or determine which disclosures a business must make.

Write the output before the questionnaire

The UK Government Data Quality Framework connects data quality to intended use and recognises that different users can need different things from the same data. Its adoption request concerns central government; the purpose-led approach is useful more widely. Here it informs a voluntary request-design exercise, not a reporting requirement.

Write one sentence that the request must enable the team to complete. Include the decision owner, population and next action. Avoid “collect ESG information”, which leaves the reviewer unable to explain why any particular field is needed.

In this invented case, the reporting lead approves: “For the three supplying sites in our agreed site list, identify which electricity records could support a later extraction for April 2025 to March 2026, and assign the next evidence step.” The output is a planning note with site-level gaps. The buyer has not yet approved an emissions calculation method or a supplier performance comparison.

Keep those possible later uses separate. A statement that bills exist may help schedule an extraction. It does not establish the quantity consumed, its relevance to the buyer’s purchases or the quality of a future emissions result. If the business wants those answers, approve the corresponding brief and evidence requirements.

Give every field a job and a route

The inherited questionnaire has 16 fields. The team tests each field against the approved planning sentence. For this exercise, each field receives one route; the counts describe the form, not respondent effort or a measured efficiency gain.

• Prefill four fields: buyer supplier ID, contracted entity name, the buyer’s relevant order-reference schedule and the existing supplying-site list. Send these for correction rather than asking suppliers to rebuild them.

• Ask five fields now, for each site: available electricity-record types; earliest and latest months available; known missing months or changes in coverage; the responsible business role or existing contact channel; and any access restriction with a proposed review route.

• Open four evidence fields only when triggered: a redacted sample page if the record type is unclear; a meter-list excerpt if site coverage is unclear; month-level references if the stated dates are ambiguous; and a documented sharing arrangement if the proposed review route is unresolved.

• Remove three fields from this request: group-wide emissions total, sustainability rating and net-zero target. None answers this planning question. Retain their disposition in the design log; do not delete existing business records.

The check is 4 + 5 + 4 + 3 = 16. The supplier initially completes five field types per site and checks the prefilled information. Conditional evidence is not automatically demanded from every supplier. A different purpose may legitimately need a removed field, but it requires its own reason and review conditions.

For every retained field, write the consequence of each answer. If any answer would leave the next action unchanged, challenge the field. Also check the opposite direction: can the team make its planning decision when the field is absent? This catches both unnecessary questions and missing information.

Original fictional request-design example. The field counts classify one inherited form; they do not measure time saved, environmental performance or the completeness of an energy inventory.

Test the request with three fictional replies

The buyer’s confirmed list has Supplier S1’s sites A and B, and Supplier S2’s site C. The site list is the case boundary, not a claim that these are all of either supplier’s operations. All organisations, replies and records below are invented.

Site A: the owner describes monthly electricity bills covering all 12 requested months and no known missing month. A business role can arrange an extraction through the existing supplier portal. The planning outcome is “schedule extraction”. No kWh value is accepted yet, and the description of coverage remains subject to checking the actual records.

Site B: the owner reports that electricity is billed through a landlord’s shared service charge. No separate site electricity quantity is identified. Trigger the meter-list or service-coverage clarification and ask who holds the underlying information. The planning outcome is “resolve record ownership and coverage”. A paid service charge cannot be assumed to contain a separately usable energy measurement.

Site C: the owner identifies bills for April to December 2025. January to March 2026 records are not yet available through the proposed channel. The planning outcome is “nine months identified; three-month evidence step open”. Request the location or owner of the remaining records, without filling the gap with zero or calling the site uncooperative.

The note now contains one site ready for an extraction appointment, one requiring a record-ownership clarification and one with a three-month retrieval gap. These are next-work states. They are not a completeness percentage for energy or emissions, and they do not rank the suppliers’ environmental performance.

Run a small pre-send test

Ask a colleague who did not design the form to read the instructions as the supplier. Can they identify which sites to answer for, distinguish an available record from a verified number, and explain what to do when the answer is unknown? Use the three replies above to check that the form can represent each outcome without forcing a false yes/no answer.

In the fictional test, a single “data available” box would have put A, B and C together even though their next actions differ. The team replaces it with the five defined fields and conditional evidence rules. This demonstrates a design defect in the example, not an observed improvement at a client or a tested industry benchmark.

State the return date, accepted response format and existing business channel before sending. Explain who will review the reply and what it will be used for. Request only the relevant non-confidential description at this stage; do not ask for account credentials or unrestricted personal records. If evidence is restricted, agree a suitable route rather than treating restricted access as an adverse finding.

Change the purpose explicitly

Suppose finance later asks for an emissions estimate from these purchases. Keep the availability replies as planning inputs. Create a new, approved request version describing the calculation object, boundary, period, method and evidence needed. The existence of the earlier survey does not approve reuse for a new assessment or establish that the requested calculation is possible.

For a purchased-goods emissions calculation, use the supplier emissions request guide to develop the topic-specific request. When the replies arrive, the supplier boundary and period guide addresses whether the returned dataset matches it. This pre-send exercise ends with a justified request design and an explicit next-use boundary.

Source checked 14 October 2026: Government Data Quality Framework (2020), sections on intended use and user needs. Contains public sector information licensed under the Open Government Licence v3.0. The supplier case, field routes and outcomes are original fictional teaching material.