● Documentation

Velora

Predictive Analysis Workspace — a complete user guide covering login, data upload, forecasting, and every page in the workspace.

11Sections
12Workspace pages covered
User GuideNon‑technical, end‑user focused
Section 1

Getting Started & Logging In

Velora is accessed through a web browser. Once the workspace URL is opened, you will land on the login screen.

1.1 Signing in with email and password

  1. Enter your registered email address (or username) in the Login ID field.
  2. Enter your password.
  3. Click Log In.

Your account must already exist in the system for a password login to succeed. If you don't have credentials yet, contact your workspace administrator.

1.2 Signing in with a connected account

The login screen also offers "Continue with" buttons for single sign-on. Depending on what your organization has enabled, you may see:

Continue with Google Continue with Microsoft Continue with LinkedIn

Selecting one of these redirects you to that provider to sign in, then returns you to Velora already logged in. The first time you sign in this way, an account is created for you automatically using your name and email from that provider — no separate registration step is required.

1.3 Staying signed in / logging out

Once logged in, your session stays active in the browser until you log out or your session expires. Use the Log Out option in the top navigation to end your session. Logging out clears any data you had loaded in that session, so export anything you need first (see Section 9).

Note: If you switch accounts on a shared browser/computer, always log out first — logging in as a different user starts a completely fresh workspace so you never see another user's uploaded data or forecasts.

1.4 First things you'll see after logging in

On your first successful login, Velora shows a short welcome message and confirms that session security tracking has started. You will land on the Overview page.

Section 2

Workspace Navigation

Velora is organized into a set of pages, accessible from the side/top navigation menu. Click any page name in the navigation to switch to it — your active page, uploaded data, and generated forecast persist as you move between pages within the same session.

PageWhat it's for
OverviewLanding dashboard with headline stats (latest Impressions, Engagements, forecast Sales/Revenue, data quality status) and quick links to the rest of the workspace.
Data QualityShows the automated Data Quality & Cleansing report for the most recently uploaded dataset: quality score, missing-value summary, outlier summary, drift summary, consistency checks, duplicate detection, and an editable data grid.
ForecastingDisplays the generated forecast table (Month, Sessions, Posts, Impressions, Engagements, Engagement Rate, Sales, Revenue, confidence range) and the reference table of all Forecasting Variables the model can use.
ChartsVisualizes trends: Impressions Trend, Engagement Trend, Engagement Rate, Seasonality, Impact Multiplier, Forecast Funnel, Sales/Revenue Forecast, Actual vs Predicted, Driver Impact, Correlation, and Feature Importance.
Auto InsightsPlain-language, automatically generated observations about trends, drivers, and forecast outlook based on the current dataset.
Model PerformanceShows the comparison across all trained models plus a time-based backtest: accuracy, precision, recall, F1, and how each model compares to a naive ("no-change") forecast.
Sales PredictionConverts the Impressions/Engagements forecast into a projected Sales and Revenue figure using assumptions set on the Settings page.
Goal SeekReverse forecasting: enter a target Sales or Revenue figure and Velora solves for the driver combination needed to reach it.
Accuracy TrackerAfter uploading real monthly results, compares them against what was forecast: accuracy %, hit/miss, and naive-forecast comparison.
SQL and ReportsPush the current forecast to the connected database, run read-only queries, and download the Forecast CSV, Power BI CSV, or a PDF report.
Notification CentreChronological log of everything that has happened in the session. Admin accounts can also broadcast a message to every logged-in user from here.
SettingsConfigure the model, forecast horizon, sales assumptions, macro/micro adjustment, and LinkedIn integrations.
Section 3

Uploading Your Data

Velora is driven by a monthly CSV file of your marketing/content performance data. Data can also arrive automatically via the LinkedIn integration (Section 10) instead of a manual upload.

3.1 Required columns

Every CSV upload must contain the following nine columns (one row per month). Column names are matched automatically even with different spacing, capitalization, hyphens, or underscores.

ColumnDescriptionExample
MonthCalendar month for the row.Jan-25, 2025-01, 01/2025, January 2025
SessionsWebsite sessions driven for that month.12500
PostsNumber of posts published.18
CreativesNumber of creative assets used.24
LikesTotal likes received.3400
CommentsTotal comments received.210
SharesTotal shares received.150
ImpressionsTotal impressions.185000
EngagementsTotal engagements.9200
Note: The Month column accepts several common formats, including "Jan-25", "Jan-2025", "2025-01", "01/2025", and "January 2025". At least 3 complete monthly rows are required before a forecast can be trained.
Resource: For the full list of every recognized column, its priority tier, and a ready-to-fill CSV template, see the Data Dictionary & Resources page.

3.2 Optional variables

Beyond the 9 required columns, Velora recognises around 30 additional optional columns. If present, they're automatically picked up as extra inputs to improve forecast quality — you don't need to add them, and their absence never blocks an upload.

CategoryOptional columns in that category
Audience GrowthNew Followers, Follower Growth Rate
Website Traffic (breakdown)Organic Sessions, Social Sessions, Direct Sessions, Referred Sessions
Content PublishingPosting Days, Posts Per Week, Consistency Score
Content FormatsVideo Posts, Carousel Posts, Image Posts, Text Posts
LinkedIn Live PerformanceLinkedIn Live Count, LinkedIn Live Attendees, LinkedIn Live Views
Webinar PerformanceWebinar Count, Webinar Registrations, Webinar Attendees, Webinar Attendance, Webinar Attendance Rate
Employee AdvocacyEmployee Posts, Employee Shares, Employee Advocacy Score
Paid MarketingPaid Spend, Boosted Posts, Sponsored Content
Content StrategyThird-Party Posts, Research Posts, Event Posts, Product Posts
Business EventsIndustry Event Flag, Conference Flag, Product Launch Flag

3.3 How to upload

  1. Go to the Overview or Forecasting page.
  2. Click the Upload / Choose File control and select your .csv file.
  3. Velora automatically checks the file's data quality (see Section 4) and, if it passes, trains the forecasting models and generates a forecast right away.

A confirmation notification appears (and is logged in the Notification Centre) once the upload and forecast are complete, including how many months of data were loaded.

Section 4

Data Quality & Cleansing

Every dataset — whether uploaded as a CSV or pulled from LinkedIn — is automatically checked before it is used for forecasting. You don't need to trigger this manually.

4.1 What is checked

  • Missing / null values, including runs of consecutive missing months
  • Statistical outliers in each numeric column
  • Distribution drift between earlier and more recent records
  • Consistency checks: required columns present, correct data types, values within expected ranges, and logical checks (e.g. Engagements should not exceed Impressions)
  • Duplicate or suspicious records (e.g. two rows for the same month)

4.2 Quality score and status

Each dataset receives a Quality Score out of 100 and a rating: Excellent, Good, Fair, Poor, or Critical. A score of 75 or higher (with no missing required columns) is marked Clean and the forecast proceeds automatically. Anything below that is marked Dirty and the workflow pauses for your review.

4.3 What you can do with a Dirty dataset

  • Continue Anyway — proceeds with the data as-is; the forecast is generated but flagged as based on unreviewed data.
  • Clean the Data First — Velora automatically fills/adjusts missing or out-of-range values, removes duplicate rows, and re-checks the result.
  • Edit Manually — an editable data grid lets you correct individual cell values yourself, after which the file is re-checked.

The Data Quality page also lists specific Recommended Actions and plain-language insights explaining how each issue could affect your forecast.

Section 5

Page-by-Page Guide

5.1
Overview

Headline numbers, current data quality status, and shortcuts into the rest of the workspace. Use "Refresh Stats" to recompute the figures.

5.2
Forecasting

Forecast table with projected Sessions, Posts, Impressions, Engagements, Engagement Rate, Sales, Revenue, and confidence range. Includes Rerun Forecast.

5.3
Charts

Impressions/Engagement trends, Engagement Rate, Seasonality, Impact Multiplier, Forecast Funnel, Sales/Revenue Forecast, Actual vs Predicted, Driver Impact, Correlation, Feature Importance.

5.4
Auto Insights

A running list of plain-language observations generated from your current dataset and forecast.

5.5
Model Performance

Compares all models with a time-based backtest: accuracy, precision, recall, F1, and comparison to a naive forecast.

5.6
Sales Prediction

Translates the Impressions/Engagements forecast into a Sales and Revenue projection using your Settings assumptions.

5.7
Goal Seek

Work backward from a target Sales/Revenue number to the driver levels needed to reach it. See Section 7.

5.8
Accuracy Tracker

Tracks forecast accuracy once real results are available. See Section 8.

5.9
SQL and Reports

Push data to your database, query it, and export files. See Section 9.

5.10
Notification Centre

Session activity log and admin broadcasts. See Section 11.

5.11
Settings

Forecast, model, and integration configuration. See Section 6.

Section 6

Settings

The Settings page controls how forecasts are generated and how external connections are configured.

6.1 Forecast & model settings

FieldWhat it controlsDefault
Selected ModelWhich algorithm to use: Auto Select Best Model, Ridge Regression, Random Forest Regressor, Gradient Boosting Regressor, Neural Network (MLP), Adaptive Online Learner (SGD), Extra Trees Regressor, or Hist Gradient Boosting Regressor.Auto Select Best Model
Forecast MonthsHow many months ahead to forecast (1–12).3
Forecast BaseWhich historical point the forecast builds forward from: Last Historical Month, Historical Average, or Rolling Average (3 months).Last Historical Month
Lead Conversion Rate (%)Share of engagement that becomes a lead, used in Sales Prediction and Goal Seek.2.0
Sales Conversion Rate (%)Share of leads that convert into a sale.20.0
Average Deal ValueAverage revenue value of one closed sale.5000
Change RatingManual sensitivity adjustment applied to the forecast.5
Upper / Lower Circuit (%)Maximum allowed swing above/below the model's raw forecast, keeping the confidence range realistic.15 / 15
Next Sessions / Likes / Expected Posts / CreativesOptional manual overrides for next period's driver inputs, used instead of the model's own projection when supplied.blank
Macro Impact (%)Manual adjustment for broad external/market conditions, applied as a multiplier on top of the forecast.0.0
Micro Impact (%)Manual adjustment for internal/company-specific factors, applied as a multiplier on top of the forecast.0.0

After changing any of these, Velora automatically regenerates the forecast using your current uploaded data. You can also trigger this manually with Rerun Forecast.

6.2 LinkedIn integration settings

Covered in full in Section 10.

Section 7

Goal Seek

Goal Seek answers the question: "What would we need to do to hit a specific Sales or Revenue number?" It works backward from your target using the model that is already trained on your data.

7.1 How to run a Goal Seek

  1. Open the Goal Seek page (requires an uploaded dataset and a generated forecast).
  2. Choose the Target Metric: Revenue or Sales.
  3. Enter the Target Value you want to reach.
  4. Choose the Month Offset — how many months ahead the target applies to.
  5. Optionally, lock one or more drivers to a fixed value and let Velora solve for the rest.
  6. Click Run Goal Seek.

7.2 Reading the result

  • The required level for each driver needed to reach your target
  • How that compares to your current baseline (percentage change per driver)
  • Projected Impressions, Engagements, and the achieved Sales/Revenue value
  • A Gap % showing how close the achieved value is to your target
  • A plain-language narrative summarizing what would need to change
Note: If the target is very far from what's achievable within a realistic driver range, Velora flags the result as a stretch estimate rather than a fully feasible plan.
Section 8

Accuracy Tracker (Uploading Actuals)

Once real results for a forecasted month become available, upload them to see how the forecast performed.

8.1 How to upload actuals

  1. Make sure a forecast has already been generated.
  2. Go to the Accuracy Tracker page.
  3. Upload a CSV with a Month column plus whichever of Impressions, Engagements, and Sales you have real figures for.
  4. Velora automatically matches each row to the corresponding forecast month and computes accuracy.

8.2 What you'll see

  • Accuracy % for Impressions, Engagements, and Sales for each matched month
  • A Hit / Miss classification against a 15% tolerance band
  • A comparison against a naive "no-change" forecast
  • The confidence level originally attached to that forecast month
Section 9

SQL, Exports & Reports

The SQL and Reports page is where forecasts leave the workspace — either into your own database or as downloadable files.

9.1 Push to SQL

Pushes the current forecast (and the combined historical + forecast dataset) into your connected database. Each push is stamped with a unique Run ID and scoped to your own login. Pushing also lets your historical data auto-load automatically next time you log in.

9.2 Run SQL Query

A simple query box for running your own read-only SELECT statements against the connected database. Only SELECT statements are permitted from this tool.

9.3 Downloads

ExportContents
Forecast CSVThe generated forecast table (Month, drivers, Impressions, Engagements, Sales, Revenue, confidence range) as a CSV file.
Power BI CSVA combined historical + forecast dataset formatted for import into Power BI or similar BI tools.
PDF ReportA formatted PDF summarizing the forecast, key charts, and headline figures — suitable for sharing outside the workspace.

Each export is timestamped in its filename so you can keep multiple versions.

Section 10

LinkedIn Integration

Velora can connect directly to LinkedIn in two distinct ways, configured separately on the Settings page.

10.1 "Continue with LinkedIn" sign-in

Used purely for logging in to Velora with your LinkedIn identity (see Section 1.2). This does not pull any performance data by itself.

10.2 LinkedIn organization statistics

A separate connection authorizes Velora to pull your LinkedIn Page/organization statistics — Impressions, Clicks, Likes, Comments, and Shares — directly from LinkedIn.

  1. Enter the Client ID, Client Secret, and Redirect URI provided by your LinkedIn developer app under Settings → LinkedIn Stats.
  2. Save the settings.
  3. Authorize the connection when prompted; you'll be redirected to LinkedIn to approve access and returned to Velora afterward.
  4. Use Fetch LinkedIn Stats to pull the latest statistics on demand.

Every LinkedIn pull automatically runs through the same Data Quality check described in Section 4 before the numbers are used anywhere in the workspace.

Section 11

Notification Centre

A running, timestamped log of everything happening in your session: successful uploads, forecasts generated, data quality warnings, export confirmations, connection errors, and security events.

Use this page any time something doesn't look right — it's usually the fastest way to see exactly what happened and when.

11.1 Broadcast messages (administrators only)

Administrator accounts can push a message that appears in every logged-in user's Notification Centre at once — useful for announcing maintenance windows or workspace-wide updates. A broadcast can be cleared by the administrator once it's no longer relevant.