Know what your customers actually experience.
Mysterya runs mystery shopping end to end. Design the survey, send vetted auditors, review every visit against its evidence, and publish results your teams can act on.
Screens from Saffron Table Hospitality, a sample workspace with two restaurant brands in the UAE and Saudi Arabia.
Results your leadership reads, and your sites act on.
Every wave lands in Echo: scores, reasons, evidence and the actions that follow. The same answers, read six different ways on one overview.
Every area, ranked by what it costs you
Areas sorted by their share of the points, switchable to touchpoint, stage, category, experience, pillar, tier or owner.
Leaders and laggards
Top and bottom five, with the visits a fatal check capped.
Kinds of check, side by side
Cleanliness, speed, staff behaviour: open one to see the areas where it failed.
One score per department
The same answers regrouped by the team that can fix them.
What auditors wrote
Every comment, by location and area, with its tone counted.
The photos behind the numbers
Failures first. Open one to see the answer it proves.
One set of answers, as many charts as you need.
37 ways to read the same scores, from country maps and sunbursts to flows, fail-together chords and hexbins. Everything below is drawn from the real results of the Saffron sample workspace, using the same chart code Echo runs.
Every score, weakest first
Pick what to rank by. The colour is the score band, so the bars that need a plan stand out before you read a number.
- Exceptional90+
- Strong80+
- Good70+
- Needs work55+
- Critical<55
- Seating area63.3
- Upselling65.7
- Queue management66.2
- Kitchen & food quality70.8
- Order taking71.9
- Menu & signage73.8
- Exterior & entrance73.8
- Host stand76.6
- Digital presence79.2
- Website & app booking83.3
- Packaging85.7
- Dining room service86.8
- Sommelier & bar91.1
- Reservations desk92.6
- Cashier & payment93.1
- Kitchen pass93.3
- Restrooms93.9
- Table maintenance97.8
- Payment systems97.9
From the board summary to a single photo.
Cross any dimension with any other: area by tier, location by stage, brand by department. Heat map, bars, waterfall or scatter.
Locations: Every country and venue with its score, spread and coverage, so you know which number is final and which is still filling in.
Visits: Each visit with its score, band and evidence count. Open one for the full report, answer by answer.
Gallery: Every photo from the wave, filtered by failed checks, open actions or what our photo check spotted.
Voice: What auditors wrote, grouped into topics and coloured by tone, as bubbles, a treemap or a sunburst.
NPS and CSAT: Guest surveys from a QR code beside the mystery audit, with what each group of customers told you.
Actions: Findings become numbered tickets with an owner and a due date. The next visit checks whether they were fixed.
Tips: Recommended fixes written from the failing check, its comments and its photos, ranked by points at stake.
Insights: A written report for every wave and brand, and scheduled exports to PowerPoint, PDF or CSV.
Put any two places side by side.
Which venue handles arrival better, which brand loses points at payment, which wave moved. One screen, the same scoring for everything in it.
- Anything against anythingAssets, brands, locations, countries, waves or programmes, two to six at a time.
- Read it any wayBy stage, category, area, touchpoint, experience, pillar, tier or department, as bars, columns or a table.
- Only what you may seeA manager limited to one brand compares inside that brand. Nothing else is loaded.
Ask your results a question.
Type it the way you would ask your analyst. Ask MINT reads the scores, comments and actions you can see, and answers with the places and checks behind it.
- Which location is weakest this wave, and why?
- What do guests at Yas Bay complain about most?
- Which open actions are overdue, and who owns them?
- Answers from your own resultsEvery figure comes from the same scores Echo shows, never from the model’s memory.
- Names the checks and the placesWhich question failed, where, how often and how many points it cost.
- Shared with your teamKeep conversations private, or share them with colleagues who can see the same data.
Every part of a programme, from the first question to the released score.
Your results in your own tools, and in your AI assistant.
A REST API for your data warehouse and BI, and an MCP server for Claude, ChatGPT or any assistant that speaks it. Each key acts as the person who made it, so it sees exactly what they see.
REST API
JSON over HTTPS, 120 requests a minute per key, every call logged.
curl https://www.mysterya.cx/api/v1/results \
-H "Authorization: Bearer my_live_…"- /workspaceThe workspace the key belongs to and what the key can see
- /programmesProgrammes and their waves
- /resultsScores: headline figures and every reading (stages, locations, departments, tiers…)
- /visitsVisits with score and status, newest first
- /visits/{id}One visit: every answer, comment and photo
- /actionsAction tickets
- /photos/{id}A short-lived link to one photo (blurred where it shows people)
MCP server
The same tools Ask MINT uses, for the assistant your team already works in.
{
"mcpServers": {
"mysterya": {
"url": "https://www.mysterya.cx/api/mcp",
"headers": { "Authorization": "Bearer my_live_…" }
}
}
}- summary
- breakdown
- heatmap
- trend
- failed_checks
- comments
- visits
- actions
- list_programmes
- get_visit
- get_photo
- Evidence you can openClick any score to see the photographs, memos and answers behind it.
- A first wave in an afternoonIndustry templates, a guided setup and sensible defaults for everything you do not want to decide.
- Reviewed before it countsEvery visit is read by a reviewer, and any answer you dispute is marked until the review team decides. Upheld, it leaves the score.
- Auditors staff will not recogniseRepeat-visit rules and a large network keep the same faces away from the same sites.
- Many clients, one loginAgencies switch between branded client workspaces without signing out.
- See the fee before you acceptEvery offer shows the fee, the time window, the expected time in venue and any spend.
- Paid every weekApproved visits are paid in the weekly run, with spend reimbursed against the receipt.
- Rules you can readEvery automated check is published in the app. A flag goes to a person, never straight to a rejection.
- An app that stays out of sightSilent photos, voice memos and a one-tap clock screen. It works without a signal.
- Appeals to a real personIf a visit is rejected you see why, and your appeal is read by the program manager.
Templates written for the places your customers go.
Each template comes with touchpoints, scenarios, guidance and scoring already set, so the first wave starts from a tested instrument rather than a blank page.
Checks that protect the score and the auditor.
Every submission is checked automatically, and every flag goes to a person. Auditors can read every rule in the app, and an appeal is read by the program manager, never the reviewer who rejected the visit.
Built to pass your security review, and to keep shoppers anonymous.
Hosted on SOC 2 and ISO/IEC 27001 certified infrastructure, run in line with the GDPR and the region's data laws, and aligned with ISO 20252 for research quality.
How we protect your data →