Contract catering customer feedback: a buyer’s spec that proves SLAs

Contract catering contracts are rarely lost because your food is terrible. They’re lost because you can’t prove what changed, where it changed, and how fast you acted when something went wrong. That’s the gap a proper contract catering customer feedback programme closes.

A headline score is a comfort blanket. It tells a story you want to be true, but it doesn’t survive a monthly meeting with a facilities lead who has photos of a dirty condiment station at 12:40, five complaints about the vegan option, and a senior leader asking why “queues are back”.

This post is a practical buyer’s spec: how to separate diner and client audiences, when to trigger each survey, what to measure for SLA evidence, how to escalate critical issues instantly, and how to report by site, counter and daypart without drowning managers.

The real problem: you’re collecting scores, not evidence

Most contract caterers already have feedback. It’s just not built for governance.

A typical setup looks like this: a QR code on a table links to a long form, you get a trickle of responses, you report an average for “food” and “service”, and you paste a few comments into a deck. Then the client asks a question you can’t answer: “What did you do about speed at peak last week, and did it improve?”

SLA conversations need three things your average survey doesn’t produce.

Specificity. “Service was slow” is useless. “Hot counter queue exceeded 8 minutes between 12:05 and 12:25 on Tuesday” is actionable.

Traceability. You need to tie feedback to a site, a counter, and a time window. Contract catering lives and dies on dayparts.

Response time. Clients judge how fast you spot issues and fix them, not whether you can generate a quarterly average.

If your current approach can’t show “here’s the issue, here’s the root cause, here’s the fix, here’s the before/after trend”, you don’t have SLA defence. You have a morale-killing scoreboard.

Separate your audiences: diners and client contacts are not the same respondent

Contract catering has two feedback audiences with different incentives, and mixing them ruins both datasets.

Daily diners: operational truth, fast and unfiltered

Your everyday diner feedback is where you find the small operational failures that create churn: portioning, temperature, queue management, menu fatigue, cleanliness drift. This is where a canteen satisfaction survey belongs, short enough to complete in under a minute, and structured enough to analyse by daypart.

Example: an NHS site can run a decent breakfast but fall apart at lunch when agency staff are covering tills. Diners will tell you that within 24 hours if you ask the right way.

Keep the diner stream anonymous by default, and say clearly that responses go to the catering team rather than the employer. In a workplace, candour rises the moment diners know HR is not reading.

Client stakeholders: governance, confidence, and “did you do what you said?”

Client contacts answer a different question: “Is this contract being managed well?” They care about comms, responsiveness, visibility of improvements, and whether issues are being taken seriously. They’re also influenced by what they hear second-hand.

Example: a facilities manager doesn’t queue for lunch every day. They hear “queues are a joke again” in the lift. Your job is to show measured reality and corrective action, not argue with anecdotes.

The buyer’s spec should state clearly: two distinct survey streams, two sets of measures, two reporting views. One platform, but not one blended dataset.

Triggering: post-service micro, weekly pulse, monthly governance

The fastest way to kill response rates is to ask everyone everything, every time. Contract catering needs a cadence that respects habit.

Post-service (always-on): catch issues while they’re fixable

This is your always-on QR flow at point of exit, table talkers, digital screens, receipts, or intranet tiles. Keep it tight: one score plus a comment, then adapt based on what they mention. This is where you capture diner feedback that can still be fixed before tomorrow’s lunch.

Operational example: “No forks out again” at 12:10 is a same-day fix, but only if it reaches the right person quickly.

Weekly pulse: measure the basics without survey fatigue

A weekly pulse, sent via an employee mailing list or intranet, is where you ask about the fundamentals you need for trend lines: availability, perceived value, and “did we get it right this week?” It’s also a better place for rotating questions (for example, “How clear were allergen labels this week?”) without hammering daily respondents.

Monthly client review: governance questions, not menu comments

The monthly client survey should map to your meeting agenda: responsiveness, issue handling, comms cadence, confidence in compliance, and whether agreed actions landed. It should not be the place where you learn the salad bar has been messy for three weeks.

Procurement note: ask vendors to show you how they manage multiple triggers and audiences without duplicating contacts or creating reporting chaos.

What to measure to support SLAs (and how to phrase it so it’s measurable)

SLA reporting falls apart when measures are vague. “Speed” is vague. “Queue time at hot counter at lunch peak” is measurable.

Your buyer’s spec should require the ability to report these by site, counter, and daypart:

• Speed at peak: perceived wait time and, where possible, structured time bands (“under 3 mins”, “3 to 6”, “6+”).

• Availability: “Were your first-choice items available?” plus an option to name the missing item.

• Cleanliness: separate front-of-house from condiment stations, tables, and clearing speed.

• Dietary handling: confidence in allergen labelling, staff knowledge, and cross-contamination concerns.

• Value perception: this is not “price”. It’s “did it feel worth it today?”, which predicts complaints even when pricing is fixed by the client.

• Menu fatigue: the same diners eat with you every day, so track menu variety and repetition separately from food quality. “Are you seeing enough variety this week?” is a different question from “How was lunch today?”, and conflating them hides a slow leak.

A concrete way to spec it is to insist each SLA theme has a metric, a drill-down question, and a reporting slice. For example:

If a vendor can’t show you this kind of structure live in a demo, they can’t help you prove SLAs.

Turn vague comments into actions: don’t accept “food was bad” as data

Open comments are where the truth sits, but most platforms treat them like a quote bank. That’s fine for marketing. It’s useless for operations.

Your spec should demand two behaviours.

1) Intelligent probing that turns opinions into evidence

When a diner writes “food was bad”, the system should ask what “bad” means in operational terms: temperature, seasoning, freshness, portion size, presentation, or dietary suitability. That single follow-up turns a rant into a fix.

In Active Insight‘s case, this is exactly what an AI-adaptive flow is built to do. It changes the next question based on the actual words the diner used, so you don’t ask ten generic ratings. You ask the one question that makes the feedback actionable.

2) Data quality checks that stop false signals

Contract catering has a lot of rushed responses. People mis-tap. They score 2 out of 10 then write “great staff, good lunch”. If you let that stand, you spend the month explaining a wobble that never existed.

A better approach is to cross-check score against sentiment in real time and ask the diner to confirm when there’s a mismatch. It’s a small interaction, but it prevents bad data becoming a client narrative.

Escalation: critical issues need instant routing, not a dashboard you check on Fridays

Contract catering carries risk. When something goes wrong, you don’t need a report. You need an escalation path.

Your buyer’s spec should explicitly include critical issue categories such as:

• Food safety concerns and suspected illness

• Allergen mistakes or cross-contamination risk

• Discrimination, harassment, or staff misconduct

• Accessibility failures that create harm

• Explicit requests for contact

The platform must detect these from comments, not just from a low score, then trigger real-time issue alerts to the right people. Not a generic inbox that nobody owns, but a named route with an audit trail.

Operational example: a diner writes “I asked if it was nut-free, they said yes, then I saw the label after”. That should never wait for a weekly summary. It needs immediate contact capture (if the diner opts in), an internal notification, and evidence that it was handled.

Active Insight flags critical issues automatically and alerts the right people in real time, and it never nudges an unhappy diner towards a public review: drafted reviews are only offered when the feedback would make an authentic one. That matters in contract catering, where reputation is client-facing and operationally sensitive, and where you still need to treat the diner fairly.

Reporting that proves change: site, counter, daypart, and time-to-fix

Most reporting fails for one of two reasons: it’s too high level to be believed, or too detailed to be used.

The spec you want is role-based:

• Unit managers see today’s issues, top drivers, and a short list of what to fix next service.

• Account managers see trends, comparisons, and SLA-aligned narratives for the client meeting.

• Head office sees cross-site patterns: which locations are drifting on cleanliness, which counters are bottlenecks, which menu cycles create complaints.

Insist on multi-site dashboards that can slice by site, counter and daypart, and still roll up cleanly for an account view. If the platform can’t show “Lunch, Hot Counter, Tuesday to Thursday” in two clicks, you’ll end up back in spreadsheets.

For SLA defence, require two additional reporting elements. First, time-to-acknowledge and time-to-resolve on flagged issues, even if “resolve” is a note that maintenance is booked. Second, a simple way to annotate changes, for example “Added second till 12:00 to 13:00 from 10 June”, so trends have context.

That’s how SLA reporting moves from “our score improved” to “queue complaints dropped after we changed staffing, and you can see it by daypart”.

A procurement spec and demo script you can actually use

When you go to market, don’t ask vendors if they “do surveys”. Ask them to prove the workflow end-to-end.

Required capabilities (non-negotiable)

• Two distinct feedback streams, diner and client stakeholder, with separate measures and reporting.

• Adaptive questioning that captures detail without making the survey long.

• Critical issue detection from free text, with immediate escalation and an audit trail.

• Segmentation by site, counter, and daypart.

• SLA-aligned reporting that shows trends and supports meeting packs.

Demo script (make them do this live)

Ask the vendor to run these four scenarios in the demo, using your terminology:

1. A diner says lunch queue was “ridiculous”. Show how the survey captures where and when, and how it reports it.

2. A diner flags an allergen mistake. Show the escalation, the alert routing, and what the manager sees.

3. A client contact completes a monthly survey. Show how it sits alongside diner data without blending.

4. Show a month-on-month view for one site, split by counter and lunch peak, with annotations for operational changes.

If they can’t do that without “we’ll configure it later”, you’re buying promises.

If you want to see what this looks like in practice, Active Insight is built for exactly this kind of contract catering workflow: adaptive diner surveys that stay short, instant escalation for critical issues, and, paired with Service Monitor’s dashboards and monthly AI reports, reporting that stands up in client reviews because it shows what changed and where.

Frequently Asked Questions

What makes contract catering customer feedback different from restaurant feedback?

You’re reporting to two masters: the diner and the client. Restaurant feedback is mostly about returning customers. Contract catering feedback also needs to evidence SLA delivery, responsiveness, and compliance in a governance setting.

How long should a canteen satisfaction survey be?

Aim for under 60 seconds for the core flow. The trick is adaptive follow-ups, so only unhappy or specific comments trigger extra questions, instead of forcing everyone through the same long form.

What should we include in SLA reporting from feedback data?

Include measures tied to operational levers: peak speed, availability, cleanliness, dietary handling, and value perception. Report them by site, counter and daypart, and add time-to-acknowledge for critical issues.

How do we handle serious complaints like allergen mistakes or discrimination?

Treat them as critical issues with instant escalation, not as low scores. You need immediate alerts, a way for the diner to request contact privately, and an audit trail that shows how the issue was handled.

How do we avoid drowning unit managers in comments?

Give them a short daily view: top issues, critical alerts, and quick wins they can fix before the next service. Keep deep analysis for account and head office, where trend and root cause work belongs.