Learn Grid with short text tutorials

Beginner · 5–8 min

Upload your first CSV

Goal: bring a structured CSV file into Telene Grid and review it before analysis.

Best for

New users who want to understand the basic flow from file upload to usable dataset.

What you need

• a CSV file with a header row
• a few columns of structured data
• no highly sensitive or regulated data during beta

Steps

1. Open Telene Grid. Start from the Grid app after signing in.
2. Create or open a project. Projects keep related datasets and workspaces together.
3. Choose Upload from computer. Select your CSV file.
4. Review detected columns. Check names, types, and obvious import issues.
5. Save the dataset. Open the table and scan the first few rows.
6. Move to Workspace. Start analysis once the table looks correct.

What to look for

• did the header row import correctly?
• are numbers recognized as numerical columns?
• are dates recognized correctly?
• are text/comment fields preserved?
• are blank cells expected?

Common mistakes

• choosing the wrong worksheet, range, or header row when importing a spreadsheet
• having extra title rows above the header
• mixing numbers and text in the same column
• using private or sensitive data during beta

Next step: run Describe on one or more important columns.

Beginner · 8–12 min

Describe a dataset

Goal: use Describe blocks to understand important columns before deeper analysis.

Best for

Users who want a quick first read of a dataset.

Steps

1. Open a dataset. Use a dataset that has already been uploaded or created.
2. Go to Workspace. The workspace is where analysis blocks are created.
3. Drag a column into the workspace. Choose Describe if prompted.
4. Review the generated block. Look at the summary, chart, missing values, and interpretation.
5. Repeat for important columns. Start with the columns most relevant to your question.
6. Add useful summaries to your report. Keep only what helps explain the dataset.

For numerical columns

• valid and missing values
• mean and median
• spread and outliers
• histogram shape

For categorical columns

• most common categories
• rare categories
• dominant categories or imbalance
• missing values

Common mistakes

• jumping to prediction before understanding the columns
• ignoring missing values
• treating a coded category as a true number
• reporting a chart without checking the sample size

Next step: choose Compare, Relate, or Grid Analyst based on what you want to learn.

Core analysis · 10–15 min

Compare groups

Goal: compare a numerical outcome across two or more groups.

Example questions

• does satisfaction differ by plan type?
• do test scores differ by teaching method?
• is revenue different across regions?
• are response times different before and after a change?

What you need

• one numerical outcome column
• one categorical group column

Steps

1. Open your dataset. Go to Workspace.
2. Select the numerical outcome. This is the value being compared.
3. Add the categorical group column. This defines the groups.
4. Choose Compare. Grid creates a comparison block.
5. Review the chart and summaries. Look at group sizes and group differences.
6. Read the diagnostics. Check assumptions, effect size, and caveats.
7. Add the finding to your report. Keep it only if it answers a useful question.

What Grid may check

• number of groups
• group sizes
• missing values
• outliers
• distribution shape
• whether the comparison appears independent or paired

Good report wording

Average satisfaction appears lower in the Basic plan than the Pro plan. The difference should be interpreted with caution because group sizes are uneven.

Next step: use Relate or Predict if you want to understand what else may explain the outcome.

Core analysis · 10–15 min

Relate two variables

Goal: explore whether two variables are related.

Example questions

• is age related to income?
• is support rating related to churn risk?
• is plan type associated with churn?
• is usage related to satisfaction?

Steps

1. Open your dataset. Go to Workspace.
2. Drag the first column into the workspace. Start with the column you care about most.
3. Add the second column. Choose Relate when prompted.
4. Review the visual. This may be a scatterplot, grouped summary, or association table.
5. Review the method. Grid chooses based on column types and intent.
6. Read the interpretation. Look for strength, direction, caveats, and practical meaning.

What to look for

• direction of relationship
• strength of relationship
• outliers or influential points
• sample size
• whether the result is practically meaningful

Common mistakes

• assuming relationship means causation
• ignoring a curved pattern
• letting one outlier dominate the result
• reporting correlation without looking at the scatterplot
• treating coded categories as continuous numbers

Good report wording

Higher support rating is associated with higher satisfaction. This relationship does not prove that support caused satisfaction to increase, but it suggests support experience may be worth investigating.

Next step: use Predict if one variable is clearly an outcome you want to model.

Text analysis · 10–15 min

Turn text responses into analyzable columns

Goal: use text-derived columns so open-ended responses can be analyzed alongside structured data.

Best for

Survey comments, customer feedback, support notes, product reviews, interview notes, and form responses.

Steps

1. Open your dataset. Find the text or comment column.
2. Create derived text columns. Use options such as sentiment, word count, phrase flags, or keyword indicators.
3. Review the new columns. Check a few rows to make sure the results are sensible.
4. Describe the derived columns. Summarize counts, categories, or sentiment labels.
5. Connect text signals to outcomes. Use Compare or Relate with satisfaction, churn, ratings, or groups.

Useful questions to try

• are negative comments associated with lower satisfaction?
• do longer comments come from unhappy users?
• are certain phrases more common in churned users?
• does sentiment differ by plan type?

Common mistakes

• treating text-derived results as perfect
• not reviewing example rows
• using too many keyword flags without a question
• forgetting that sarcasm, context, and domain language can affect sentiment

Good report wording

Comments with negative sentiment tend to appear more often among users with lower satisfaction. The sentiment score should be treated as a signal for review, not as a perfect label.

Guided exploration · 10–20 min

Use Grid Analyst

Goal: use Grid Analyst to surface potentially useful patterns across a dataset.

Best for

Users who want help deciding what to explore next.

Steps

1. Open your dataset. Use a dataset with several usable columns.
2. Go to Workspace. Start Grid Analyst or Find interesting patterns.
3. Review suggested paths. Open the findings that look relevant.
4. Inspect each result. Check inputs, method, chart, interpretation, and diagnostics.
5. Ignore weak findings. Not every suggested pattern is worth keeping.
6. Refine promising findings. Add related columns, filters, or follow-up analyses.
7. Add selected findings to your report. Keep only the useful ones.

What Grid Analyst is good for

• finding possible relationships you may not notice manually
• connecting text-derived signals with structured variables
• suggesting comparisons or associations
• helping build a first analysis path

What Grid Analyst is not

• a replacement for human judgment
• proof of causation
• a guarantee that a finding is important
• a substitute for reviewing diagnostics

Good workflow: use Grid Analyst for discovery, then use your judgment for selection, interpretation, and reporting.

Reporting · 10–20 min

Build a report from findings

Goal: turn selected analysis blocks into a clear report.

Best for

Users who need to communicate analysis to clients, teammates, students, stakeholders, or themselves.

Steps

1. Open your workspace. Review the analysis blocks.
2. Choose a useful result. Select Add to Report.
3. Repeat for other findings. Keep only the findings that help tell the story.
4. Open Report view. Arrange sections and story order.
5. Add or edit notes. Clarify interpretation where needed.
6. Review caveats. Do not hide uncertainty or limitations.
7. Copy the report or share the workspace when ready. Review privacy before sharing.

What to include

• the question being answered
• chart or visual summary
• plain-language finding
• effect size or practical importance where available
• limitations and caveats
• next action or follow-up question

Good report wording

This finding suggests a meaningful difference between groups, but the sample size is small. It should be treated as an early signal rather than a final conclusion.

Trust and safety · 5–10 min

Share safely

Goal: review what is visible before sharing a report, workspace, dataset, or link.

Before sharing, ask

• does this contain personal data?
• does this contain confidential business data?
• are raw rows visible?
• are comments or text responses visible?
• are charts enough, or is the underlying data exposed?
• is the link public or protected?
• does the audience need this level of detail?

Steps

1. Open the report or workspace. Review every included finding.
2. Check visible data. Look for raw rows, private labels, comments, and hidden details.
3. Remove anything unnecessary. Share only what the audience needs.
4. Review share settings. Create a link only when you are ready.
5. Send to the intended audience. Treat shared links carefully.

Common mistakes

• sharing a workspace when a report would be safer
• forgetting that text comments may contain private details
• assuming a link is private without checking settings
• sending exploratory findings as if they are final conclusions

Good practice: when in doubt, share a report with selected findings rather than a full workspace or raw dataset.