Transactional apps (OLTP)
App databases behind orders, bookings, accounts and payments.
Top 3 risks we test first
- Transactions that save halfway
- Two users updating the same record
- Business rules enforced in the database
[ Data platform types ]
An order database fails when two people save at once. A warehouse fails when a join drops rows. A dashboard fails quietly, with a number that looks right. We start with the risks that belong to your platform, and test those first.
App databases behind orders, bookings, accounts and payments.
Top 3 risks we test first
Central stores of history used for reporting.
Top 3 risks we test first
Raw and curated data in open files and tables.
Top 3 risks we test first
Jobs that move and reshape data between systems.
Top 3 risks we test first
Events and changes processed as they happen.
Top 3 risks we test first
Charts and KPIs leaders make decisions on.
Top 3 risks we test first
Moving to a new engine, version or schema.
Top 3 risks we test first
Old on-premise data moving to a cloud database.
Top 3 risks we test first
Customer, product and supplier records shared across systems.
Top 3 risks we test first
Apps built on document, key-value or wide-column stores.
Top 3 risks we test first
Data prepared to train and feed models.
Top 3 risks we test first
One database shared by many customers.
Top 3 risks we test first
[ What we test ]
Data problems rarely start where they show up. We group our testing into five layers, from how data is shaped to who can reach it, so you can see exactly what each test protects.
How data is shaped
Tables, columns, data types, keys and indexes checked against the design, and schema changes checked before they ship.
Primary and foreign keys, unique and check constraints, and defaults tested so bad data can’t get in.
Logic inside the database tested with known inputs and expected outputs, including errors and edge cases.
Create, read, update and delete actions from the app checked in the database, along with the business rules behind them.
More on this service →How data travels
Extract, transform and load jobs tested step by step: mappings, joins, lookups, type conversions and failed runs.
Row counts, totals and sample records compared between source and target, so nothing is lost, doubled or changed.
Every record, relationship and piece of history checked after a move, plus a tested rollback.
Change data capture and event streams checked for lost, repeated and out-of-order events, and for lag.
What people read and trust
Completeness, accuracy, consistency, validity, uniqueness and timeliness measured against agreed thresholds.
Dashboard and report figures checked against hand-written queries, including filters, dates and drill-downs.
How it holds up
Slow queries found and measured, query plans and indexes reviewed, and improvements re-measured.
More on this service →Many users and jobs reading and writing at once, checked for locks, deadlocks, timeouts and lost updates.
More on this service →Backups restored for real and recovery time and data loss measured against your targets.
Replicas checked for lag and consistency, and failover tried to see what happens to live traffic.
Database engine and driver upgrades tested for changed behaviour, deprecated features and performance shifts.
Who can reach it
Users, roles and permissions checked so each account reaches only the data it should, with access logged.
More on this service →Every input that reaches a query tested for SQL and NoSQL injection, manually and with scanners.
More on this service →Personal data in test and analytics copies checked for masking, and deletion and export requests checked end to end.
Data handling tested against the GDPR, HIPAA and SOX controls your auditors ask about, with evidence.
More on this service →[ Databases & data quality ]
Data can load without errors and still be wrong. We agree a threshold for each of six quality dimensions with you, check every one automatically and report a score, with the rows that fail.
No missing rows or empty required fields.
Values match the real-world source.
The same fact agrees across every system.
Formats, ranges and codes follow the rules.
No duplicate customers, orders or products.
Data arrives when people need it.
[ Process & deliverables ]
You always know what we’re doing and what you’ll get. Every query, script and report belongs to you.
We map your sources, pipelines, stores and reports, and the business rules your data must follow.
We agree mapping and reconciliation rules, quality thresholds and a safe test data approach.
SQL checks, pipeline runs and report comparisons, with every issue logged as it’s found.
Performance, concurrency, security and recovery testing where your scope includes them.
Data checks wired into your CI or scheduler, fixes retested and the release or migration signed off.
[ Tools ]
Most data testing is careful SQL. We add tools where they make checks faster, repeatable or automatic, and we’re happy to use the ones you already have.
See the full list on our technology stack page →
[ Case Studies ]
Products where the data had to be right, written up by the engineers who tested them.
3,000+test cases traced to requirementsRead the case study
100%of survey question types covered by data-capture checksRead the case study
“QAonTOP has provided at least 20 times the value. Their testing service is outstanding and has greatly streamlined our process.”
[ Ways to work ]
Choose the engagement model that matches how your data changes. You can switch models as your platform grows.
Not sure which fits? Find your model in four questions or compare all six.
[ FAQ ]
{{ f.a }}
[ Contact us ]
✓ Free one-week POC✓ NDA before access✓ A QA lead replies the same day
[ Keep exploring ]
Every browser, device and layer
02Contracts, data and error handling
03Tenancy, billing, scaling and releases
04Installers, updates and OS behaviour
05Accuracy, guardrails and agents
Also useful: ERP & CRM testing · All testing services · Fixed-price packages · Estimate your QA cost