Skip to content

Datamanagement

Duplicate customers, files re-entered every week, years of history locked in old software. Witify and Kashika Studio, its data division, build the databases and data flows that put your data in order and keep it right.

Talk to an expert

Databases, folders and charts laid out on one grid, linked to a cloud
For
SMEs, organizations and institutions in Québec
Technologies
MySQL, PostgreSQL, AWS
Team
Montréal, serving all of Québec

01 At a glance

Databases that grow without losing their history

Two databases designed by Witify. HEC Montréal’s takes in a delivery every week; the Observatoire des tout-petits’ reads workbooks from ten providers. In both, no past data is lost.
  • Over 1.9billion

    dated measurements in HEC Montréal’s MID database Every row keeps the date of the delivery that brought it.

  • 170

    weekly deliveries recorded since August 2023 21 files per delivery, loaded into the database in the same order every week.

  • 97

    data sources imported for the Observatoire des tout-petits Ten providers, over 80% of sources in Excel, each with its own layout.

  • Over 560

    published versions kept At the Observatoire, every release of an indicator stays on record.

02 When

When your data stops adding up

A company’s data spreads quietly: a file here, a piece of software there, a copy made to get by. One day, two reports give two numbers.
An aisle of servers in a data centre, between two rows of cabinets
  1. The same customer, three versions

    One record in the accounting software, another in the sales spreadsheet, a third on the website. Each has its own address. Nobody knows which one is current.

  2. Partner files re-entered every week

    A supplier sends its prices in Excel, a partner its sales in CSV. Someone copies them over line by line, with the mistakes that come with it.

  3. Years of history in an old system

    Access, FileMaker or an in-house tool: years of data sit there, organized in a way nobody really remembers. Our article on the risks of Access in 2026 covers the options.

  4. Totals that depend on who runs them

    Management and accounting arrive at two different totals for the same month. The meeting goes into finding the right one, and the decision waits.

  5. A migration put off year after year

    The new software waits because nobody can promise the history will follow. The old system stays, for lack of a better plan.

If a well-kept spreadsheet is enough, we will tell you. A database earns its place when several people, several tools or several years have to rely on the same numbers.

The move is decided on those criteria, not on the size of the company.

03 What we handle

From the first entry to the backup, your data in order

Data management is the work underneath your software: where each piece of data lives, how it arrives, how it stays right and how you find it again.

01

Databases

A model drawn from how you operate: linked tables, one key per record, a date on anything that changes.

02

File imports

Excel, CSV or partner exports: every row is read and checked. Differences are shown before anything is written.

03

Integrations

Accounting software, online store, CRM: data moves between tools through APIs, with a log of every send.

04

Quality and duplicates

Rules flag duplicates, empty fields and inconsistencies. A person confirms every correction.

05

Migrations

Data from the old system moves to the new one, history included. Totals are compared with the source before the switch.

06

Backups and access

A backup every day, access by role, sensitive data encrypted.


Going further with your data

Once in order, your data can answer other questions. Each use has its own service, with the same team.

04 At work

Where data comes in, moves and gets fixed

Four screens like the ones we build for our clients, each at the exact moment an error could slip into your data.

Imports

Every file received leaves a trace

The log keeps each delivery with its start time, duration and steps. An incomplete delivery is never half loaded: it is marked skipped, with its reason, and the next one goes through.

HEC Montréal: one delivery a week since 2023

Duplicates

Two records, one customer

The system spots records that look alike. A person picks the right values. The orders follow the record that is kept.

The audit log: every merge stays on record

Integrations

A rejected send stays in sight

When the accounting software rejects a record, it is counted and listed with the reason. Nothing disappears quietly.

Distribution MA: eight kinds of documents sent to QuickBooks

Quality

The rules run again every night

Malformed emails, orders without a customer, likely duplicates: every night, the rules reread the database. Gaps are put forward for correction, never fixed behind your back.

05 Method

Understanding your data before moving any of it

Data that is poorly understood migrates poorly. So our projects always follow the same steps, and each one ends with a deliverable you see and approve.

01 Discovery

Learning who touches which data.

Meetings with the people who enter, import and look up the data every day. We note what is missing, what contradicts itself and what gets re-entered. Then we rank it with you by priority.
Fig. 01 Each department’s needs, ranked with you

02 Analysis

Taking stock of tools and data flows.

Software, spreadsheets, databases, file exchanges: we list what each tool holds and where its data comes from. Then we settle its fate with you: it stays and plugs into the database, or its data migrates and it is retired.
Fig. 02 Every tool listed, with the decision that concerns it

03 Data model

Drawing the database before writing it.

The pieces of your operations become linked tables, each with its key. Anything that changes over time, like a price or a status, keeps its date. You approve the model before development starts.
Fig. 03 The tables and their relationships, approved before development

04 Mockups

Approving the entry and review screens.

The screens where your teams will enter, import and correct data are drawn first. Future users annotate them, version after version, until they recognize their own work in them.
Fig. 04 A mockup, commented on by your teams

05 Specification

Writing down the validation rules.

Accepted formats, required fields, duplicate rules, what to do with an invalid row: everything is described with testable criteria. This document is the reference: development starts once it is approved.
Fig. 05 An import rule, described and approved

06 Development

Moving forward in one-week sprints.

Each week, part of the database and its imports reaches staging. The code is written with Sprintify, our toolkit: AI-assisted code generation, proven packages and components. Your teams try each delivery with a copy of their real files.
Fig. 06 The week’s board

07 Quality

Testing every rule on real cases.

Valid file, missing column, unreadable date, duplicate: each case has its automated test. Another developer reviews every change, then the full test suite runs again before release to production.
Fig. 07 A change, checked before production

06 Safeguards

Safeguards from the first entry

A wrong value costs little to fix the day it comes in. Six months later, it has already been used to invoice, order or decide. Our choices aim at that first day.
Every change to a record, kept with its date and source
  1. Compare before writing

    The risk
    An import overwrites correct values with those of a badly filled file. Nobody notices.
    Our choice
    The file is first compared with the database: rows added, values changed, rows missing. Nothing is written until a person approves. The Observatoire des tout-petits imports its sources this way.
    The gain
    Every change is seen before it reaches your reports.
  2. Stop rather than load halfway

    The risk
    A half-loaded week skews a series with nothing to flag it.
    Our choice
    At the first unreadable row, the import stops and notes where and why. Once the cause is fixed, it picks up exactly there.
    The gain
    A period is complete, or marked as missing. Never in between.
  3. Keep a trace of every change

    The risk
    A number changes. Nobody knows when, why or what it was before.
    Our choice
    Every change is kept with its date, source and author, in an audit log that stays with the record.
    The gain
    You can find where a number came from and what it used to be.
  4. A copy every day, access by role

    The risk
    A handling error wipes out records. An employee opens data that does not concern them.
    Our choice
    An automatic backup every day, checked regularly. Access by role, sensitive data encrypted, hosting certified ISO 27001 and SOC 2.
    The gain
    Your data can be restored. Each person sees only what concerns them.

The tools beneath your data

Widely used technologies, chosen for the project and for what is already in place. See our technologies

Databases
MySQL, PostgreSQL, every structural change versioned
Files
Amazon S3, storage of the files received
Integrations
REST APIs, QuickBooks, Shopify, depending on the software you already use
Application
Laravel, Vue, or the technology already in place
Hosting
AWS, DigitalOcean, ISO 27001 and SOC 2 certified hosting
Monitoring
Sentry, errors reported in real time

07 AI

AI reads the files, rules guard the database

AI earns its keep where a rule struggles: a messy file, a label spelled ten different ways. What goes into the database still passes through tested rules and a person.

Where AI helps

  • Reading a messy Excel workbook and suggesting where each value belongs: at the Observatoire des tout-petits, the team approves every difference.
  • Matching labels written differently, such as “Mtl”, “Montréal” and “MONTREAL”.
  • Flagging an unusual value so a person can take a look.
  • Querying your data in everyday language instead of formulas: our article on conversational reporting shows how.

What rules keep

  • Keys and relationships between tables.
  • Formats, required fields and thresholds.
  • The totals that must match after every import.
  • A result that is identical from one run to the next, checked by tests.

What it does not decide

  • What is correct: it suggests, a person decides.
  • Writing to the database without approval.
  • Whether it is worth its cost: each use is measured on your files, and kept only if it saves time.

Have an AI project in mind?

Document reading, assistants, automatic classification: our AI integration service first puts a figure on the time saved, then builds what justifies it.

See the AI integration service

08 Launch

From trial import to switch-over

A migration is rehearsed before it counts. Your data runs through the new database several times before the real start. Each trial is compared with the source, number by number.
  1. Inventory What migrates, what stays archived and what is retired, decided with you.
  2. Trial imports The real data loaded into a copy of the new database, then counted.
  3. Gaps explained Every difference has a cause. We fix it at the source, then run it again.
  4. Switch-over On the date agreed together, in a quiet period. The old system is only retired once you give the go-ahead.
  5. Training Your teams learn to import, correct and read the checks. The documentation comes with it.

09 After

Once the database is live, an expert who sees what more it can bring

A database in order opens other possibilities: a report nobody dared ask for, a new source, a task AI can now take on. Witify keeps advising you: spotting the next useful project, then measuring its effect. For upkeep and updates, see our support and maintenance offering.

01

The next manual re-entry, found

We review with you where data is still copied, reconciled or corrected by hand. Then we suggest what to connect next, starting with what costs the most hours.

02

Results, measured

Fewer duplicates, imports with no rework, reports ready sooner: we track what each change delivered, against your own indicators.

03

What is new, sorted for you

New connectors, AI models that can read your files, analysis tools: we sort through them and bring you only what suits your data.

04

A database that keeps up

A new source, a new indicator, an acquisition: the database grows without losing its history.

10 Our work

Data in order, at our clients

At HEC Montréal, a research database takes in measurements for thousands of artists every week. At the Observatoire des tout-petits, every file received is compared with the database before import. At Original Garage Motosport, each part is counted in one place, which tells Shopify what can really be sold. Each case study shows the screens and choices that keep this data sound.

View all projects

  • HEC Montréal

    Academic research

    Thousands of artists. Measured every week.

    Custom software

  • Québec’s youngest children. In over 200 indicators.

    Custom software

  • Original Garage Motosport

    Motorcycle parts, online retail

    Shopify sells the kit. The WMS counts its parts.

    Custom ERP

Very satisfied with the service offered by Witify. We developed an inventory management system that has truly transformed my company’s data. Exceptional and fully satisfying service!

André-Pierre Cloutier Co-owner, Original Garage Motosport

11 Questions

Questions and answers

The questions that often come before a first import or a migration.

Another question? Write to us at info@witify.io or call 1 800 334 9031.

How do we move our data off an old system without losing its history?

We start with an inventory: what the old system holds, what must follow and what can stay archived. Each field is then mapped to its place in the new database, dates and versions included.

The data runs through a copy of the new database several times. Each trial is counted and compared with the source. Every gap is explained, then fixed. The old system stays in place until your team approves the switch-over.

If your data lives in Access, our article on the risks of Access helps you gauge the urgency. For a full application, see our application modernization service.

How do we avoid duplicates and mismatches between our software?

First, each kind of data gets a single source: customers in one tool, prices in another. The other software receives a copy, kept current through an integration.

Then, matching rules suggest records that look alike. A person confirms each merge. Imports are checked on entry, then a quality report flags what still needs fixing.

The real starting point is often elsewhere: agreeing with your teams on what an active customer or a discontinued product means.

Where is our data hosted and who can see it?

The systems we host generally run on servers in Canada, with DigitalOcean or AWS, whose hosting is certified ISO 27001 and SOC 2. An automatic backup runs every day. Sensitive data is encrypted.

Each person opens only the screens of their role. Every change is written to the audit log, with its date and author. If needed, we sign a confidentiality agreement before seeing your data.

For compliance questions, such as protecting personal information, see our data strategy service.

Can we keep Excel, or do we need a real database?

Both have their place. Excel remains the right tool for analysis, what-if scenarios and quick fixes. A database becomes useful when several people edit the same records, when history matters or when every file received has to be checked.

Keeping Excel at the entry point is common: the Observatoire des tout-petits’ platform receives mostly Excel workbooks, which it reads and checks before loading them into the database. Our article on the limits of Excel helps sort things out.

What do I need to start collecting data?

First, clear objectives: what you want to discover or solve with this data. Next, the types of data that answer them: demographic, behavioural, transactional and so on.

You also need a place to collect and keep it. Depending on the volume and the number of users, that ranges from an Excel sheet to a database, or even an analytics platform.

Collection must respect the privacy rules of your region: consent obtained, personal data protected.

Finally, you need expertise to interpret and use the data collected: an analyst, a data scientist or a team like ours.

I already have a data warehouse, can you make use of it?

Yes. We connect to your warehouse to extract, transform and load the data your analysis and visualization tools need. Whether it runs on your premises or in the cloud, we work with the infrastructure in place.

If needed, we also help you manage it better: better structured tables, quality checks, access by role.

Is it possible to create visualizations connected to a real-time data source?

Yes. A visualization can update as the data arrives, to track performance, monitor an activity or detect an event.

It takes a configuration specific to each source and close coordination with your teams, so the connection stays stable and secure. Our data visualization service covers this side in detail.

What factors influence the cost and duration of a data project?

Five factors weigh the most.

  • Complexity: large or varied data, or a demanding analysis, takes more time.
  • The state of the data: incomplete or messy data has to be cleaned before anything else.
  • Technology: some tools or integrations add to the cost.
  • Skills in place: a team new to data management will need more support.
  • Scope: the number of questions to answer, sources to integrate and reports to produce.

Our business and technical analysis sizes these factors for your project before it is priced.

Are tools like PowerBI or Data Studio sufficient for data analysis?

For simple to moderate needs, yes. Power BI and Looker Studio (formerly Data Studio) are user-friendly, require no programming and produce good dashboards.

For deeper analysis, such as machine learning, forecasting or very large volumes, you may need tools like Python, R or d3.js. Either way, an analysis is only as good as the questions asked and the data at the source. See our data analysis service.

Which data should come first?

Show us the file re-entered every week, the old system that holds the history or the two reports that contradict each other. An expert looks at them with you and suggests a first step. If needed, we sign a confidentiality agreement before seeing anything.

  • A 30-minute call with an expert, not a salesperson
  • A realistic budget estimate, backed by numbers
  • A concrete action plan, whether you work with us or not
Or write to us

Free. No commitment. Reply within 24 hours.

Names and figures shown in the interfaces are examples.