Dashboards and scheduled reports on Power BI, Metabase or Looker Studio, with agreed metric definitions your own analyst can extend
Business intelligence and reporting work joins accounting vouchers, CRM records and spreadsheet tabs inside Power BI, Metabase or Looker Studio, using metric definitions agreed in writing first. Role dashboards and scheduled packs follow. The model is documented. Your analyst can extend it. Share your requirements and we will say which tool fits.
- Every measure has one written definition
- Tool choice is matched to budget and team skill
- The model is documented for your own analyst
Reporting suits owners who want one screen of figures they trust, such as a distributor tracking branch margin and dealer ageing or a manufacturer setting production against dispatch. A clinic group watches chair occupancy and collections. A services firm follows consultant utilisation. The tool changes with budget and team skill. Agreeing what a sale means across departments is harder than choosing software, and that agreement always comes first.
Where Reports and Meetings Fall Out of Step
Different Figures From Sales and Accounts
Sales counts the order, accounts counts the invoice, and a third view excludes returns, so review time goes on reconciling figures that should already match. Decisions get deferred. Nobody trusts the screen.
Reports Tied to One Person
One person exports from the accounting software, pastes into a workbook and formats the weekly pack by hand. When that person goes on leave, the meeting runs on last week’s numbers and the pack stalls for days.
Dashboards That Go Unopened
A screen crowded with charts can miss the questions people actually ask, and a dashboard that answers none of them stops being opened after the first few weeks. Usage drops quietly. Nobody complains.
Figures That Arrive Late
Data is assembled after the month closes and reviewed mid-month, so a problem from the first week keeps running until that review takes place. Timing is the real fault.
Data Spread Across Four Places
Sales sit in a CRM, billing in the accounting software, stock in a spreadsheet and collections in bank statements. A margin question per branch means opening four sources and matching rows by hand. It is slow work.
Licences Without a Working Model
Power BI seats are paid for while the report reads an Excel export saved on one desktop, so replacing that machine stops the refresh. The licences keep billing. Nothing updates.
Numbers Still Requested by Phone
The quickest route to a figure is a call to the accountant, who then opens several files and ledgers to answer a question one screen should settle. That is the gap.
The dashboard is the last stage. Agreed definitions and reliable extraction from each system come before it, and the screen is then built around the few questions its reader opens it to answer.
Agreed Definitions First, Then Role-Based Dashboards
Work starts with the questions managers ask and the figures they dispute, and each measure then gets one written definition, agreed with the accountant in the room. Sources are connected and a model holds those definitions. Dashboards follow, one per role, in a tool that fits the budget and the skills already on the team.
One Definition per Measure
The definition covers what counts as a sale, when a lead becomes an opportunity and whether returns are netted. It is built into the model, so no report computes it differently. Written first.
One Dashboard per Role
The owner’s view carries a handful of headline numbers and a trend, a branch manager sees one branch, and a sales head sees the pipeline. Each role gets the first questions it asks.
A Tool That Fits the Team
Power BI suits businesses already on Microsoft with analysts in place, Metabase suits teams able to write some SQL, and Looker Studio suits small needs on small budgets. Each has a limit.
What Changes When Reporting Is Structured
What the business gets, before any talk of technology.
Meetings Work From One Set of Figures
Everyone opens the same dashboard, built on the same definitions, and reconciling three spreadsheets stops being the first agenda item of the review meeting. The argument moves to decisions.
Report Packs Refresh on a Schedule
Packs refresh overnight and land in inboxes before the meeting starts. Delivery no longer hinges on one person being free to build them, whether that person is on leave or on a site visit.
Data Refreshed on an Agreed Cycle
A collection shortfall or a branch falling behind appears in the next refresh and does not sit unnoticed until a monthly review, which is how most late surprises begin. Problems surface early.
Managers Filter Their Own Views
Filters by branch, product, period and salesperson let a follow-up question be answered on the screen itself, without a separate request to the accountant or the analyst.
Data Entry Problems Are Listed
Invoices with no branch tag and items with no category appear on an exception list instead of being silently dropped. Whoever made the entry can then fix it at source.
Your Analyst Can Extend the Model
The model, definitions and refresh schedule are written down. Your own analyst can add a new measure without raising a change request, because nothing about the model is hidden.
What Business Intelligence and Reporting Work Covers
Everything from the first discovery session to production and long-term upkeep.
Reporting Requirements and Question Discovery
Sessions with the decision makers collect the questions they ask now, and also the ones they stopped asking because the answer took too long to produce.
Metric Definition Workshop
The accountant and department heads settle one definition per measure, and it is written down. Two definitions of revenue undermine any reporting tool, whatever its price or its brand.
Tool Selection and Licensing Advice
Power BI, Metabase, Looker Studio and Superset are compared on budget, team skill, data volume and who needs to see what. The lower running cost is weighed first.
Data Source Connection
Accounting software, ERP, CRM, e-commerce platforms, payment gateways, spreadsheets and databases are connected through a scheduled extract in place of a manual export. A failed load raises an alert.
Data Modelling and Semantic Layer
Tables are joined and each measure is defined once, with hierarchies for branch, product and period built in. Two charts on one page cannot compute a measure differently, because definitions live in the model and nowhere else.
Data Cleaning and Standardisation
Branch names, item categories, customer spellings and date formats are brought to one standard on load, and rows that fail are listed on a report instead of being discarded. Nothing vanishes.
Dashboard Design and Build
Role-based screens answer the main questions in one view and stay readable on a phone, while drill-down leads a manager from a headline figure to the vouchers behind it.
Scheduled Reports and Alerts
Report packs go out by email or WhatsApp before a meeting, and alerts fire when overdue balances or a reorder point cross a threshold that managers chose beforehand. Nobody has to open a dashboard to find out.
Row-Level Security and Access
A branch manager sees one branch and a salesperson sees their own accounts. Purchase rates and margins stay with the people entitled to see them, in dashboards and in scheduled packs alike. Access follows the user.
Existing Dashboard Rescue
For reporting that was built and then abandoned, the review asks why nobody trusts it or opens it. The model and refresh are repaired, and unused charts are removed. Useful pages stay.
Training for Your Team
Sessions cover the people who will maintain the reports and the managers who will read them, including adding a measure, changing a filter and reading a drill-down path. Both groups attend separately.
Documentation and Handover
The model, definitions, refresh schedule and known exceptions are written down and handed over, so the reporting stops depending on one analyst remembering how it works.
Where Reporting Is Used
The processes this service is most often asked to set up and automate.
Distributor Reviewing Branch Margin
A distributor with accounting software, a dealer ordering portal and a collections sheet needs them joined overnight. The owner’s view carries sales, margin, collections, overdue balances above sixty days and stock value, each drilling down to branch and then to dealer. Detail is one click away.
Manufacturer Comparing Production With Dispatch
Shop-floor entries record what each line produced, billing records what left the gate, and a comparison per line sets planned, produced and dispatched quantities side by side. Delays carry a reason tag. The gap is visible.
Clinic Group Tracking Occupancy and Collections
A clinic group with several branches on one practice management system needs the same figures each morning. The daily pack covers consultations, new and repeat patients, chair occupancy by slot and collections against billing, while the front desk gets unpaid balances listed by patient.
Services Firm Tracking Utilisation
Time entries, project budgets and invoices sit in three places. The dashboard compares hours logged, billed and collected by consultant and by project, and flags any project running past budget. Overruns surface early, while a conversation with the client can still change the outcome.
Retail Chain Reconciling Store and Head Office Sales
Store sales do not match head office figures because stores count a sale at billing while head office nets returns in the following month. A definitions workshop settles the rule. The model then carries it.
Small Firm With One System
A business with one system, a few dozen transactions a day and two readers of numbers may not need a platform. A nightly export into a shared workbook with pivot views can serve, and the choice is revisited when a second source arrives.
Is this the right service for you?
It suits businesses that have outgrown manual processes and want a system they can rely on.
Owners Who Request Numbers by Phone
Where the quickest route to today’s collections or dispatch position is a call to the accountant, reporting is not yet in place. That holds whatever software is being paid for.
Businesses With Data in Several Systems
Sales sit in a CRM, billing in the accounting software, stock in a spreadsheet and orders on a marketplace. The value comes from joining these into one model, not from each system producing its own separate report.
Companies Already Paying for Power BI
Licences are bought and seats assigned, yet little has been built. The connection often points at a desktop export instead of a real source, so scheduled refresh cannot work at all.
Multi-Branch and Multi-Location Operations
Comparison between branches is the main purpose, and each manager needs a personal view that hides other branches’ performance and margins. Row-level rules do this.
Teams Preparing Reports by Hand
A capable person spends each Monday morning exporting from the accounting software and pasting into a workbook. Structured reporting replaces that weekly chore, frees the person for analysis and builds the pack without anyone touching it.
When a Spreadsheet Is Still Enough
One system, a few dozen transactions a day and two people who read numbers rarely need a platform. A nightly export into a shared workbook with pivot views carries no licence cost. A real tool becomes relevant when a second source must be joined, when several people need their own filtered views, or when a branch manager must not see everything. Weekly preparation eating a person’s time is another sign. The assessment states which side of that line a business sits on.
What it can do, in plain terms.
Agreed Metric Dictionary
Every measure is defined once in writing. The definition sits in the model instead of being repeated inside each report, so a change is made in one place only.
Scheduled Source Extracts
Automatic pulls from accounting software, CRM, ERP, gateways and spreadsheets run on a schedule, and a failure alerts a named person. It never passes silently, which matters when a stale figure reaches a meeting.
Joined Multi-Source Model
Customer, item, branch and date act as the shared keys, which lets one report span the CRM and the accounting ledgers without anyone merging exports by hand. Joins are documented.
Role-Based Dashboards
Separate views exist for owner, branch manager, sales head and operations, and each answers the first questions of that role while hiding the charts that role never opens.
Drill-Down Paths
A manager moves from the company total to branch, product and individual document, so the cause of a figure can be found without asking anyone. The trail ends at the voucher.
Row-Level Security
Filters are applied per user, so people see only their branch, territory or customer set, and scheduled deliveries follow the same rule. The email pack respects it too.
Mobile-Readable Layouts
Views are designed for an owner opening them on a phone between meetings. They also hold up on a widescreen monitor in the boardroom, where the same layout is reused without edits or rework.
Scheduled Delivery
Report packs are sent by email or WhatsApp on a schedule and reach people before the meeting they are meant for. The recipient list follows the access rules.
Threshold Alerts
Notifications go out when overdue balances cross a set level, a branch falls behind or stock drops under a reorder point. Levels are agreed first.
Exception Reporting
Untagged, unmatched and invalid records are listed on a report. Data entry problems show up and are never quietly excluded from a total, so someone can fix each one.
Refresh Monitoring
The dashboard shows when each source last loaded and whether the load succeeded, so nobody reads a stale view believing it is current. Failed loads stay visible. The timestamp is on screen.
Documented Model
Tables, joins, measures and the refresh schedule are written down, so your own analyst can extend the model without reverse-engineering it. A new measure needs a short entry in the document.
How the work runs, step by step.
From the first conversation to a release your team uses every day. Each step says what you provide and what you get back.
- 01
Questions Before Charts
Interviews with decision makers collect the questions they ask, including ones they stopped asking. Access to the managers themselves is needed, not a requirement list passed down by one person who speaks for everyone.
- 02
Metric Definitions Agreed
A workshop settles what each measure means, with the accountant present and someone who can close a dispute between two departments. People settle this step. Technology does not.
- 03
Tool Selection and Source Access
A platform is recommended against the budget and the team, and the sources are connected through read access to each system. Any source held as a spreadsheet on one machine needs a named owner.
- 04
Model and First Dashboard
The model is built along with one dashboard for one role, kept narrow on purpose. That role reviews it early and reports what is wrong while changes are still small and cheap to make. Narrow first.
- 05
Validation Against Known Numbers
Each measure is checked against a figure already trusted, such as the audited sales register or the ageing report, and the accountant confirms the match. An unvalidated dashboard is rarely believed.
- 06
Rollout, Training and Handover
Remaining dashboards, scheduled delivery, access rules, training and documentation are completed, and a named owner inside the business then maintains the definitions as the company changes. Handover is formal.
What you receive when the work is done.
Everything needed to run, maintain and extend the software without depending on us.
It connects to the systems you already use.
Tools that work stay. The new system is joined to them rather than replacing them.
Accounting and ERP
- QuickBooks and Xero
- Zoho Books
- Odoo
- SAP Business One
- Sage and NetSuite
Sales and Customer Systems
- Zoho CRM
- Salesforce
- HubSpot
- Custom CRM databases
- WhatsApp and call logs
Commerce and Payments
- Shopify and WooCommerce
- Marketplace settlement reports
- Stripe and PayPal
- Bank statement feeds
Files and Databases
- Google Sheets
- Excel on shared drives
- PostgreSQL and MySQL
- Microsoft SQL Server
- CSV drops
Delivery Channels
- Email schedules
- WhatsApp report delivery
- Microsoft Teams
- Embedded dashboards in your own portal
Chosen for reliability and a long life.
Picked to match your scale and to stay maintainable for years.
Reporting Tools
Data Storage
Pipelines and Modelling
Source Connectors
Delivery and Alerting
Why owners choose GullySystem.
Definitions Are Settled First
Many reporting projects stall on disagreement between departments, not on technology. Two departments counting a sale differently will defeat any tool, so the definition is written down before a chart is drawn. This comes first.
The Lower-Cost Tool Is Considered First
Power BI is not always required. Metabase, Looker Studio or a scheduled workbook may serve at a lower running cost, and the recommendation states which one fits and why it fits your team.
Built Around a Few Questions
A dashboard is designed around what its reader opens it to find out, because extra charts push the useful number out of view. Fewer charts, clearer answers.
Validated Against Trusted Numbers
Each measure is matched to the audited register or ageing report with the accountant checking, because an unreconciled dashboard will not be believed by the people who read it. Trust is earned here.
Bad Data Is Shown, Not Hidden
Records that cannot be matched appear on an exception list. Excluding them quietly gives a tidy dashboard with a wrong total, and the owner then makes decisions on that wrong total without knowing.
Your Team Receives the Documentation
The model and definitions are documented and your analyst is trained on them, so adding a measure never means raising a ticket with GullySystem. The handover pack is the proof.
How the engagement can be shaped.
Each model says what it suits and how the work is scoped. Pick the one that fits, or ask which one we would suggest.
Reporting Assessment
This is for a business that argues over its numbers and has not chosen a reporting tool, so the assessment covers its questions and sources as a defined piece. A BI analyst runs the discovery, the definition workshop and the source review alongside a project lead, with your contact gathering the people who own each metric. A question raised halfway is noted in writing and costed before it is taken on. The report can go to any vendor.
Dashboard Build
Aimed at teams whose questions and sources are known and who want role-based dashboards, scoped by view and charged per phase. A data modeller, a dashboard developer and a tester build them, and your contact checks the figures against the books. A new view comes as a short written note, costed before the model is touched. Handover includes the model, refresh jobs and logins.
Reporting Rescue
The right call for reporting that another vendor built and nobody trusts or opens; the review is charged first, and repair is scoped from its findings. An analyst and a developer take over the existing files, and your contact names the reports people really read. Work found beyond the agreed repair is written down and costed before anyone starts. Unused reports are dropped with your sign-off.
Ongoing Reporting Support
Pick this when measures keep changing as the business grows; a small reporting team works for a monthly fee. An analyst monitors refreshes and keeps definitions current, a developer adds views, and your contact raises any number that looks wrong. Small additions fit inside the monthly fee, while a larger view gets a written note first. Notice ends it; definitions are handed over.
Questions owners ask first.
Plain answers to the questions asked before work starts.
Power BI, Metabase or Looker Studio. Which should we use?
That depends on your budget, your team and where your data sits. Power BI suits businesses already on Microsoft 365 with someone able to learn its DAX modelling language, and it handles large models. Metabase suits teams where someone writes some SQL and self-hosting is preferred to per-user pricing. Looker Studio is free and suits a handful of views over Google Sheets, BigQuery or advertising data, though it becomes awkward beyond that. The recommendation comes before any licence is bought.
Can you pull data out of our accounting software?
Yes. Most accounting software can be read through an API, an ODBC connection or a scheduled export. That lets the sales register, purchases, ledgers, outstanding balances and stock be extracted on a schedule. Practical checks come first. The accounting system must be reachable when the extract runs, ledger names must be consistent enough to match against the CRM, and vouchers need branch or cost centre tagging.
How current will the numbers be?
That depends on the refresh cycle you choose. Nightly is the usual setting, and a refresh at two in the morning gives a complete view of the previous day before work starts. Hourly or near-live refreshes are possible where the source supports them, though they add build and running cost. One test helps. Ask which decision would change if a number were four hours fresher.
Why does everyone in our meeting have a different figure?
Usually it is two definitions, not two mistakes. Sales counts the order while accounts counts the invoice. One view nets returns in the month of sale and another nets them in the month of return. Interbranch transfers may count in one system and not in another. No tool resolves this. A definitions workshop with the accountant and department heads comes before any build.
We already bought Power BI licences and nothing got built. Can you take it over?
Yes. This is a regular part of the work. A common finding is a connection to an Excel export on one desktop, so the refresh broke when that laptop changed. Another is measures computed inside individual visuals, which leaves two charts on one page disagreeing. The review keeps what is worth keeping, repairs the sources and the model, and removes pages nobody opens.
Can a branch manager be stopped from seeing other branches?
Yes, through row-level security, set up from the start. Each user is mapped to the branches, territories or customers they may see, and every report filters accordingly, scheduled deliveries included. Margin and purchase rates are often restricted further. Adding it late is hard. Dashboards and distribution lists were designed for everyone seeing everything, and each would need rework.
Do we need a data warehouse?
Not usually, and not at first. With one or two sources and moderate volumes, reporting tools can read them directly or through a light staging database. A warehouse becomes relevant when several sources need joining or when reads slow down a live system. It also matters when history that the source overwrites must be kept, or when transformation logic outgrows a few SQL views. Building one earlier adds cost and no benefit.
What do the licences actually cost to run?
That depends on the tool and on how many people need access, and the running cost is stated in the recommendation before any commitment. Power BI is priced per user, with capacity options for larger deployments. Metabase has an open-source edition you host yourself alongside paid editions. Looker Studio is free with paid options at scale. Superset is open source. Cloud hosting for the data and the time to maintain the reports are further costs to plan for.
What if our data is messy?
It usually is. That never blocks the work. Branch names spelt several ways, items without a category, invoices with no cost centre and duplicated customers are all common. Whatever can be standardised is standardised on load, and whatever can be matched is matched. The remainder goes on an exception report, not into the bin, and that list tells your team which entries to fix at source.
Who maintains it after you finish?
Your team, if you want that. The model, joins, measure definitions, refresh schedule and known exceptions are documented, and whoever will own the reports is trained to add a measure or change a view. A small monthly support arrangement is available for new requirements. Nothing here needs GullySystem to stay.
How is this different from your data engineering service?
The two overlap, and the difference is where the weight sits. This work covers the reporting layer, meaning the choice of tool, the agreed definitions, the model and the dashboards. Data engineering covers what sits underneath when sources get complicated, such as pipelines, warehouses, transformation and history. For many small and medium businesses, the reporting layer over direct connections is enough. When it is not, the two pieces of work run together.
What drives the cost of a reporting project?
The sources to be joined, and how clean they are, set the cost. One system with tidy masters is a smaller job, while several systems that disagree on customer names and branch codes need more data work. Other factors are the number of roles needing their own view, row-level security, whether a warehouse is required, the history to load, scheduled delivery and alerting. The cost is confirmed once the sources are reviewed and the definitions workshop is done.
What should our first dashboard show?
A small set of headline numbers and one trend, built for one person. Start with whoever makes the most decisions, usually the owner, and cover the questions they currently ask by phone. What was sold, what was collected, what is overdue, what is stuck and how that compares with last month. A narrow first dashboard gets opened daily. A broad one is opened once and forgotten.
Read before you decide.
Practical notes on the questions this service raises, not product marketing.
Power BI vs Metabase vs Looker Studio: Which Suits Your Business
Power BI, Metabase or Looker Studio? A plain comparison by data source, team skill, licensing and who will maintain the reports once they are built.
Why Sales and Accounts Report Different Numbers, and How to Fix It
When sales, accounts and the dashboard each show a different sales figure, the cause is usually definitions. How to trace the gap and agree one rule.
How to Plan a Dashboard Built on Tally Data
Before connecting Tally to Power BI, Metabase or Looker Studio, settle the questions, voucher tagging, extract route, refresh cycle and checks.
How to Build Dashboards That Help Managers Make Decisions
Build a manager dashboard around decisions, not data: name the decision, pick one measure, set a threshold, add context and agree who acts on each panel.
More guides on this page’s topic
- What is a semantic layer and why dashboards need one
- Metric definition document: a template for agreed business numbers
- How to set up scheduled report emails and alerts people act on
- What to ask a BI consultant before they build your dashboards
- Why your Power BI report is slow and how to find the cause
- Why dashboard figures change after a data refresh
- Why managers stop opening the dashboard and how to rescue it
- How to set up row level security so each branch sees its own numbers
- Which reports to keep in accounting software and which to move
- Import or live query: how should a BI tool read your data
Let’s Build the Right Software for Your Business
Tell us about your current operational challenge, spreadsheet bottleneck, or software requirement. An experienced engineer will review your workflow and reply within one business day.
- No obligation consultation
- Senior engineer reviews your brief
- Your operational details stay 100% confidential
Start with your requirements
Seven short steps that tell us what you need. A senior engineer reads them before anyone gets in touch.
Share your requirementsRelated services.
Data Engineering and Business Intelligence
The pipelines, warehouse and transformation underneath, for the point where sources get too complicated for direct connections and the reporting tool starts to strain. It sits below the dashboards.
Data Migration and Moving Off Spreadsheets
Dashboards are only as good as the masters behind them. This covers cleaning and consolidating the customer, item and branch data that feeds the reports, before any chart depends on it.
API and System Integration
Getting data out of systems that were never designed to share it, on a schedule and without a manual export. A failed pull raises an alert for a named person.
AI and Intelligent Automation
Once the numbers are agreed, assistants that answer questions about them and flag what changed without anybody opening a dashboard.