A customer research repository that answers in customers' own words, and checks every finding is still true this month
Reuse the research you already paid for.
See one case, screen by screen ↓A utility pays agencies every year for move-in studies, digital channel trackers, payment diaries, advisory panels and benchmark reports. The decks land on a shared drive and the transcripts in an agency portal. Two years later a programme team asks the same question, nobody can find the slide, and the study is commissioned again.
When an old finding is found, it is often quoted as if it were still true — a 2022 slide in a 2026 business case, after the app has changed. And a customer's words travel from transcript to deck to email and lose their study, date and participant, so an assurer cannot trace them. Ofwat's PR24 standards ask for customer research that is shared in full and independently assured.
Estimated split for one programme team's evidence request, in working days, by hand and with the solution. The answer itself comes back in seconds; review and sign-off stay with people.
Six agents read every study, answer in customers' own words and check each finding against this month's voices; the head of customer insight signs.
Lumina Utilities, an electricity and water utility, owns 186 customer studies from 2019–2026: $6.4M of research and 41,260 verbatim quotes. On Monday 13 October 2026, the digital channels team wants Kestrel Insight's quote KI-2611 — $68,000 for a new study of expat customers and the app. Before the Head of Customer Insight approves it, an analyst asks what Lumina already knows. Here is what happens, screen by screen, in the working solution.
Nadia types “Why do expat customers avoid the app?” The answer comes back as a wall of customers' own words: 3 findings, 14 quotes from 6 studies, out of 61 matched in 9 studies read, 2022–2025. Each quote names the customer by segment — “Expat, arrived 2023, apartment” — never by name, and carries a chip with its study and year.
“Asked by Nadia Rahman for Daniel Okoro, Head of Digital Channels, before quote KI-2611 for a new $68,000 expat study is approved”
The library holds every study Lumina owns, read into quotes: surveys, focus groups, advisory panels, diaries and benchmarks from four agencies and the customer panel, each with its cost and how often it has been reused. The finding builder matched 61 quotes, kept the 14 most specific and grouped them into three findings, each from at least two studies.
Finding 1: “Sign-up stops at a residence ID they don't have yet” — 6 quotes · 3 studies · 2022–2025
All 14 quotes match their source word for word. A click on a source chip opens the Move-in Journey Study 2023 transcript at Group 3, 00:22:41, the line highlighted, beside the study's facts: Kestrel Insight, 6 focus groups with 48 people, fieldwork Mar–Apr 2023, $62,000, reused in answers 9 times.
“The app wanted my residence ID on the first screen. I'd been in the country nine days. I didn't have one yet.” — Expat, arrived 2023, apartment
The checker read 8,406 voices from the last 30 days — calls, chats, emails and ratings, 13 Sep – 12 Oct 2026. 38 support the three findings and 4 contradict them. Residence ID: 21 support, 1 contradicts. Cards issued abroad: 12 and 2. Passcodes sent to a local number: 5 and 1, rising from 2 the month before. Each live voice shows its channel, date, words and the customer's own rating.
Call, Oct 12: “Moved in last week and the app says residence ID required. I only have my passport.”
“The bill and the app are in one language only”, from the Bill Redesign Research 2022, now has 0 voices supporting it and 6 contradicting it. The app and bill have been bilingual since release 5.2 in March 2025, and six customers this month mention using the Arabic or English switch. The finding stays on the wall struck through, with its reason and a Reverse button for the analyst.
Limit: the still-true checker never edits a finding — it only proposes retiring, and an analyst can reverse it.
The coverage grid shows what Lumina knows by topic and customer group, with the latest year in each cell. Expat business customers have never been asked about the app: 2,940 expat-owned small-business accounts, most paying at the counter. The gap list holds three more, among them water-only customers on estimated bills, and customers on the medical-device list (412 today), asked about outage notices in only one focus group, in 2023.
Gap finder: “not asked yet: expat business customers about the app”
The agency quote asks four questions. Three are marked Known — answered by findings 1, 2 and 3, with today's live voices. The fourth, “Why do expat business owners pay at the counter?”, is the gap. The gap finder drafts top-up NR-0942: an online panel of 300 expat business owners in 3 weeks for $14,000, beside the new study's 6 focus groups and 400-person survey in 9 weeks for $68,000.
Limit: the gap finder never commissions research — every brief goes to the Head of Customer Insight.
From the findings Nadia kept, the evidence writer drafts the summary for Daniel's business case for app release 6.0: one paragraph per finding with its quotes as footnotes and its still-true count, a “No longer true” section, “What we do not know” and a sources table of six studies. Beside it, the checks before signing: 14 of 14 quotes match, 38 supporting voices attached, the retired finding marked “not to be quoted”, no customer names or account numbers.
Limit: the evidence writer never sends without the analyst, and never cites a retired finding as current.
Leila owns the research budget, signs evidence summaries and approves research over $5,000. The sign-off shows the summary, who it is for and the evidence behind it: 3 findings, 14 quotes, 38 live voices. She confirms she has read the quotes and the voices behind each finding, and approves the top-up instead of the agency quote.
“Approve top-up NR-0942 ($14,000) instead of quote KI-2611 ($68,000)” · “I have read the quotes and the voices behind each finding”
The summary is signed 13 Oct 2026, 10:04, and is ready to send to Daniel Okoro. On the Questions page, research not bought twice in 2026 moves from $77,000 to $131,000, against an estimated target of $180,000 for the year — after a solar export credits study ($41,000) and an outage notification study ($36,000) earlier in the year. Every question is kept with what came back: 312 asked in 2026, 61 answered in full from past research.
“Top-up NR-0942 approved at $14,000 instead of quote KI-2611 at $68,000. $54,000 not spent.”
The same question, seen by the four people it matters to — what their month looked like, and what it looks like now.
Six agents read every study, answer in customers' own words and check each finding against this month's voices; the head of customer insight signs.
Reads each study — decks, reports, survey open answers, focus-group and panel transcripts — and splits it into quotes with the customer segment, topic and page or minute.
Searches the quotes for a question in plain words and groups the strongest into findings, each with a short headline in plain words.
Reads every quote on the wall back against its source, word for word, with the page or minute.
Matches each finding against the last 30 days of Voice of Customer and counts the voices that support or contradict it.
Lists the parts of a question no study answers, and drafts a narrow brief with a sample and a cost.
Drafts the evidence summary for a business case from the findings the analyst keeps, every sentence footnoted to its quotes.
Signs and approves. The agents propose; a named person decides.
Ask about any past study, or which finding to check against this month's voices. When no study answers, it says so and offers the gap list instead.
Is the residence ID finding still true this month?
Yes. In the last 30 days 21 voices support it and 1 contradicts it, against 19 the month before. The newest is a call on 12 October from a customer who moved in last week with only a passport. The one contradicting voice registered with a passport at a kiosk, which the app then picked up.
What have we already paid for on expat customers?
Seven studies since 2022 heard from expat customers, $236,300 of research — among them the Move-in Journey Study 2023 ($62,000), the Bill Redesign Research 2022 ($57,000) and the Payments Diary Study 2024 ($44,000). Three of the four questions in quote KI-2611 are already answered by them.
Draft the evidence summary for Daniel's business case
Drafted: three findings, 14 verbatim quotes from 6 studies, each with its page or minute, the still-true count from 38 voices this month, and the retired language finding marked so it is not quoted. It waits for Leila Haddad to sign.
Which past findings have customers stopped supporting?
Four findings in the library are now retired, because more voices this month contradict them than support them. The one in today's answer is “The bill and the app are in one language only” from 2022: 0 support, 6 contradict since the bilingual release 5.2. The others are about paper bills (2021), the old payment kiosk (2022) and the usage forecast (2023).
12 screens from the working solution, on its sample data. Pick one to see it large.
How answers are built, how findings are rechecked and who decides are settings, chosen by Customer Insight.
| Setting | Default | Choose from |
|---|---|---|
| Verbatim quotes only | On | On · Off — a quote is shown only when it matches its source word for word |
| At least two studies per finding | On | On · Off — otherwise the finding is labelled “one study” |
| Show the age of older studies | On | On · Off — studies older than 4 years carry their age on every quote |
| Still-true window | 30 days | 7 days · 30 days · 90 days of voices counted for each finding |
| Retire a finding when | At least 5 voices | contradicting voices outnumber support, with at least the number of voices you set |
| Call a finding rising when | On | On · Off — this month's support is at least double last month's |
| Remove names and numbers when a study is read | On | On · Off — participant names, phone numbers and account numbers |
| Who decides | Leila Haddad | Head of Customer Insight — signs summaries, approves research over $5,000 |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate: about 22 studies a year at a $46,000 average, with 1 in 5 requests answered from past research (about 4 studies a year not bought); the days split is also estimated · “typical” = published figures (Drive Research, market research costs: focus groups $7,000–$20,000+ per group; online survey $5,000–$15,000 per 400 responses). Regulatory context: Ofwat PR24 standards for customer research. People, Lumina Utilities, the agencies, studies and quotes named on this page are fictional — characters and sample data in the working solution.
A searchable library of every customer study a utility owns — surveys, focus groups, advisory panels, diaries and benchmarks — that answers questions across them. Customer Research Repository answers in customers' own words, grouped into findings, with the study and year on every quote, and checks each finding against this month's Voice of Customer.
Six agents share the work. The study reader turns each study into quotes with segment and page or minute; the finding builder groups them into findings; the quote checker confirms each one word for word; the still-true checker counts this month's voices for and against; the gap finder lists what no study has asked; and the evidence writer drafts a footnoted summary. People keep, sign and decide.
Every finding is checked against the last 30 days of calls, chats, emails and ratings. A meter shows how many voices support or contradict it, and you can read each one. When contradicting voices outnumber support, with at least five voices, the checker retires the finding and shows it struck through with its reason; an analyst can reverse it.
That is its purpose. Before a study is approved, it shows which questions in the agency quote past research already answers. In the demo, three of four questions in a $68,000 quote were answered, and a $14,000 top-up covered the fourth. Every question and the research it saved is kept on a reuse ledger.
Every quote carries its study, year, method and page, minute or answer number, opens at that point in the source, and is checked word for word before it is shown. Signed summaries keep the quotes and the voices behind them. Ofwat's PR24 standards ask for customer research that is shared in full and independently assured.
Participant names, phone numbers and account numbers are removed when a study is read, so quotes name a customer segment only. Studies under restricted consent answer to Customer Insight only, and the still-true check reads voices without customer names or account numbers.
Yes. The agents search, check and draft; they never commission research and never send a summary on their own. The analyst chooses which findings and quotes go into a summary, and the Head of Customer Insight signs it and approves any new study or top-up over $5,000.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your past studies, agency portals, panel transcripts and Voice of Customer feed — and it goes live once every quality gate has passed. We will walk you through it on a sample of your own studies first.
Every voice read, and each recurring problem named with an owner and a fix.
1-min demoClosed-Loop Feedback ManagementNo unhappy customer left without a call back, and every recovery measured.
1-min demoComplaint Root Cause AnalysisEach complaint trend proved with system data, owned, and fixed at the cause.
1-min demoWe'll run Customer Research Repository on a sample of your own past studies and this month's customer voices.