Skip to content

Datavisualization

Dashboards, public charts, data explorers: Witify and its data division Kashika Studio design the tool where your readers find the answer themselves, on your real data.

Talk to an expert

Two open laptops showing interactive charts that compare scenarios
For
Organizations, institutes and companies in Quebec
Technologies
D3.js, Vue, Nuxt, Laravel
Team
Montreal, with Kashika Studio

01 At a glance

Tools the public uses, not just analysts

Two public tools designed by Witify and its data division Kashika Studio: specialist data, read by a public that is not.
  • More than 1,400

    charts configured for the Observatoire des tout-petits Six types, drawn by one engine built for this data.

  • More than 200

    Observatoire indicators, held in a single data model Sorted into 13 themes, from a simple count to a survey percentage.

  • 151

    scenarios to compare in the Pathways Explorer 269 series, 14 territories and five horizons, from 2020 to 2060.

Figures taken from the Observatoire des tout-petits and Trottier Energy Institute case studies.

02 When

When the numbers exist but nobody reads them

The problem is rarely a lack of data. It is the distance between the file and the person who needs to understand it.
A large open-plan office where employees work at their screens
  1. The report nobody opens

    Forty pages of tables, sent out every quarter. Management reads the summary. The rest is never looked at.

  2. The dashboard where everything is the same size

    Thirty indicators side by side, with no target and no comparison. Nothing draws the eye, so nobody stops to look.

  3. Data that is published, but only as a download

    The file is online. To use it, you have to open it, sort it and cross-reference it yourself: you have to be an analyst.

  4. The chart rebuilt by hand every month

    Someone pastes in the new figures, adjusts the scale, exports the image. A shifted cell goes unnoticed.

  5. An estimate drawn as a certainty

    The chart shows neither the margin of error nor the change in method. Readers conclude more than the data allows.

  6. The tool that waits for a developer

    Every new set of results goes through a programmer. An update takes weeks, and the public goes back to the files.

Are your analysts exploring their own numbers? Power BI or Looker Studio are built for them, and we will point you to those tools. Custom work starts when the reader is no longer an analyst.

A public that reads without training, a margin of error to show, a chart that has to live inside your website or application: that is where a custom tool earns its place. Our data analysis service helps settle the question.

03 What we build

From the internal dashboard to the public chart

Your teams use it to decide, the public to understand. Either way, the tool reads your data at the source, with no copy to keep up to date.

01

Management dashboards

Your organization’s indicators on a single screen, fed by your systems rather than by a copied export.

Management software

02

Embedded public charts

The chart lives on your website, with its title, source, notes and exports. It follows your data when it changes.

Observatoire des tout-petits

03

Data explorers

Hundreds of series to filter, compare and enlarge, for a public that wants to find its own answer.

Pathways Explorer

05

Reports and fact sheets

Reports exported to PDF or Excel, a fact sheet per region, each with its notes and source. The numbers come from the same data as the online tool.

06

An admin to publish from

Import, check, publish: your team updates the tool without waiting for a developer.

See how


Our other data services

Visualization is the last link. Upstream, strategy organizes the data and analysis tells what it holds.

04 Reading

A good chart is understood before it is explained

Readers move from the overview to the detail. Each screen helps them take the next step and says how far the data can take them. The examples use fictitious data.

Overview

Spot it before reading it

One cell per region and per month, darker as the wait grows longer. The problem shows before a single number is read.

Gap

Compare to the target, not just the trend

The target stays grey, the result in colour. The eye goes straight to the gap, with no mental arithmetic.

Compare

One scale for every region

Each region has its own panel, on the same scale. Pointing at a month marks it in every panel at once.

Precision

The margin of error, drawn

Each bar carries its confidence interval. The North has few respondents: an asterisk points to the note under the chart.

Source

The source travels with the chart

Notes, method breaks and source stay attached to the chart, all the way into the image, CSV or Excel export.

05 Method

From the reader’s question to the live chart

Questions come before the chart, mockups before the code. You approve each step’s deliverable before the next one begins.

01 Discovery

We start from your readers’ questions.

Meetings with the tool’s future readers (management, teams, the public) and with the people who produce the data. Each question is written down in the person’s own words, then ranked by importance.
Fig. 01 The readers’ questions, ranked by priority

02 Analysis

We look at the data as it arrives.

For each source: its format, its frequency, its quality and its gaps. We then know what the tool will be able to state. We also know what will need a caveat.
Fig. 02 The data sources and their role in the tool

03 Data model

We design the data structure.

Indicators, series, regions and values become linked tables, built for the groupings and comparisons the charts will make. Published versions have their place from the start.
Fig. 03 The data model and its published versions

04 Mockups

You try the screens before we code them.

Screens first take the form of interactive prototypes. You work with them, along with a few future readers when possible. A chart that is hard to read is reworked before the first line of code.
Fig. 04 A prototype under review

05 Specification

We describe each chart, and you approve it.

The specification sets out the data each chart reads, its interactions, its exports and its special cases: a missing value, a series that changes method. Development starts from this approved document.
Fig. 05 A chart’s specification, approved

06 Development

We deliver charts every week.

Sprints last one week. Sprintify, our AI-assisted code generation, provides ready-made components, so the team spends its time on your charts. Each one goes to staging, on a copy of your data, where you can try it.
Fig. 06 The sprint board

07 Quality

We test every calculation before launch.

Totals, averages and margins of error have automated tests. Every change goes through a code review, then through those tests. Display time is measured on the real volume, not on a sample.
Fig. 07 A chart change, checked before it goes live

06 Reliability

A public figure must stand up to scrutiny

A chart commits whoever publishes it. These decisions are made at the design stage, not after the first criticism.
  1. Show what the number does not say

    The risk
    A survey of 80 respondents appears with the same confidence as a full count. Nothing warns the reader.
    Our choice
    Margin of error, fragile estimates and changes in method are drawn on the chart, explained in a note and kept in the export. The Observatoire des tout-petits chart engine works this way.
    The gain
    A number you can publish, and that your partners quote for what it is worth.
  2. Precompute what everyone looks at

    The risk
    Millions of values added up on every visit: the first page is slow. Readers leave before they see it.
    Our choice
    After each import, the server prepares the most viewed summaries. The Pathways Explorer keeps 70 of them ready to serve, one per territory and horizon.
    The gain
    The overview appears instantly, even on a large volume.
  3. Keep every published version

    The risk
    A corrected figure replaces the old one without a trace. A partner quotes a value that no longer exists.
    Our choice
    Publishing freezes a version of the indicator, its data and its charts. The working copy stays editable, and the public site only reads the current version. This is how the Observatoire des tout-petits keeps 563 published versions.
    The gain
    Every released figure can be found, dated and explained.

Publish without a developer

A visualization tool moves at the pace of your data. Your team drops in the new files, the tool recomputes its summaries, then they publish. This is how the Trottier Energy Institute team loads each new run of its model. Read the case study

A tool’s admin: an indicator’s files dropped in, the summaries recomputed

07 AI

AI prepares the data, not the conclusions

In a visualization project, AI pays off mostly before the chart, wherever someone retypes numbers from one file to another. In front of the chart, it can also answer a question asked in everyday language: our article on conversational reporting shows what it can and cannot do.

Where it pays off

  • Reading Excel workbooks that each have their own layout, then proposing the changes to make to the database.
  • Matching each label in a file to the right catalogue value: a region, a year, a category.
  • Answering a question about your data in plain language, from the charts and tables already published.

What it changes

  • Less data entry, more checking: the team validates instead of retyping.
  • At the Observatoire des tout-petits, the gain is estimated at about 2,500 hours a year. It is an order of magnitude, not a measurement.
  • A reader who does not know how to filter still gets an answer.

What it does not replace

  • The final say on a public figure: nothing is written to the database before a person approves it.
  • Choosing what to show, to whom and why.
  • The eye of a domain expert, who knows whether a gap is real or comes from a change in method.

Example

A question, an answer with its sources

The assistant answers from the dashboard’s data and cites its sources. When the data does not say why, it says so instead of guessing.

See our AI integration service

08 Launch

A launch checked against your real numbers

A visualization is validated with the people who produce the numbers, before it is shown to the people who read them.
  1. Real data The useful history imported, each series checked against its source.
  2. Expert review Your domain experts compare the first charts with what they know. A number that rings false stands out right away.
  3. First readers Staging opens to a few readers chosen with you, before the public.
  4. Go live The tool opens on the agreed date, every chart approved.
  5. Training Your team learns to import, check and publish, with documentation to back it up.

09 After

Once the tool is live, a partner focused on your readers

A chart is worth what its readers get from it. After launch, we stay on as your technology partner: we read the usage with you, advise you on what is worth adding and keep you up to date on what AI and new data sources make possible. On the technical side, our support and maintenance offer covers updates and monitoring.

01

What your readers look at

Charts opened, filters chosen, exports downloaded: actual usage shows what serves and what could go.

02

What they look for and cannot find

A missing region, an indicator a partner asked for, a comparison the tool cannot make: we recommend the additions that matter most, with the effort each one takes.

03

What AI and new data bring

Open data to cross with yours, an assistant that answers in everyday language, an import that reads the files you receive: we pick out what would move your tool forward, and explain why.

04

An outside view on your choices

Opening your data to the public, adding a BI tool for your analysts, embedding a chart in your application: we tell you what the current tool already allows and what calls for something else.

10 Work

Complex data, readable by everyone

The Pathways Explorer, built for Natural Resources Canada with the Trottier Energy Institute. The data platform of the Observatoire des tout-petits. The screens of the SMR Dashboard, designed with the firm Niji for the OECD Nuclear Energy Agency. The case studies show them screen by screen.

View all projects

  • Net-zero Canada by 2050. 151 ways to get there.

    Web application

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

    Custom software

  • A worldwide industry. One single dashboard.

    Web application

Our visualization tool would not have been what it is without the invaluable collaboration of Witify. The team helped us define the objectives we wanted to achieve with this new tool, and then provided us with clear and effective interfaces. Their team knows how to present data.

Louis Beaumier Executive Director, Trottier Energy Institute

11 Questions

Questions and answers

The questions we hear when numbers need to become readable.

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

What is the duration of a data visualization project?

It depends on the number of views, the state of the data and the intended audience. An internal dashboard connected to clean data takes a few weeks to build. A public platform, with its admin and its versions, takes several months: the one for the Observatoire des tout-petits was delivered in stages, from September 2025 to July 2026.

The analysis at the start sets the scope and the schedule. With one-week sprints, a first chart runs on your real data well before launch.

What data formats do you work with?

Mostly Excel workbooks, CSV files, SQL databases and APIs, as well as JSON and XML. Excel dominates, often with a layout specific to each provider.

The import adapts to files as they arrive, rather than asking each provider to change theirs. This is how the Pathways Explorer imports results in the format the institute’s model already produces.

What technologies do you use to develop your data visualizations?

The choice follows the project. For interactive charts, we mostly work with D3.js, inside Vue or Nuxt interfaces. The data server is often built on Laravel. If your environment relies on other technologies, we adapt to them.

The tool can also feed your existing website through an API: this is how the public site of the Observatoire des tout-petits displays the charts from its data platform.

What is the advantage of a custom visualization versus tools like Tableau, PowerBI, Excel, or Data Studio?

Tableau, Power BI, Excel and Looker Studio, which replaced Data Studio, serve a team analyzing its own data very well. Four questions help you choose:

  • Who will read the tool? An outside audience, without training, needs screens designed for it.
  • Does your data have its own reading rules? A margin of error or a non-comparable series has to show on the chart.
  • Where does the tool need to live? In your website or application, with your visual identity.
  • Who will update it? A custom admin fits the way your data arrives.

The two approaches can also coexist: a well-structured database can feed your analysts in their BI tool and the public in a custom tool.

We already have a dashboard, but nobody uses it. What should we do?

Do not change tools right away. Let us first look at how it is used: the screens opened, the ones people leave at once, the numbers they still go looking for elsewhere. Then we meet a few of its readers to learn the question they have in mind when they open it.

The fix often lies in the first screen: it has to answer that question before anything else. A few reworked screens are sometimes enough, with no rebuild. Our data analysis service can also check that the indicators shown are the right ones.

Can we update the charts ourselves?

Yes, it is planned from the design stage. The tool’s admin is used to import, check and publish, and even to write content: the Trottier Energy Institute writes the six guided analyses of its explorer itself.

At the Observatoire des tout-petits, the team adds an indicator and sets up its charts in a configurator, without a developer. The training given at launch covers these tasks.

How do you choose the right type of chart?

From the reader’s question first, then from the shape of the data. A trend is followed on a line, a share of a whole reads as stacked bars, a comparison between regions needs a common scale. When there is a target, we show the gap rather than two values to compare in your head.

The choice is then checked on the prototypes, before development: a poorly chosen chart shows as soon as someone looks for a specific answer.

How does the tool stay fast with large volumes of data?

By precomputing what everyone looks at. The Pathways Explorer prepares 70 summaries after each import: the overview of all 151 scenarios appears instantly. The detail only loads when the reader asks for it.

Display time is measured on the real volume before launch. Our article on data volume in software explains how to plan for your data’s growth.

Numbers to make readable? Let’s talk.

Show us a report, a file or a dashboard nobody opens. In thirty minutes, we look with you at what your readers should see first.

  • 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.