Data
Tableau
Tableau is only as good as the data model beneath it — and that is the part we build.
Overview
Tableau is a business-intelligence and data-visualisation platform, now owned by Salesforce, that lets analysts build interactive dashboards and explore data visually without writing much code. You drag fields onto a canvas, Tableau’s VizQL engine turns that gesture into a query and a chart, and you refine from there — filters, calculated fields, level-of-detail (LOD) expressions, and drill-downs. It connects to a wide range of sources: relational databases, cloud warehouses, spreadsheets, and APIs. The product splits into Tableau Desktop for authoring, Tableau Server or Tableau Cloud for sharing and governance, and Tableau Prep for lightweight data preparation.
Most Tableau engagements fail or succeed long before anyone opens Desktop. A dashboard is a view onto data; if the data arriving at Tableau is messy, unmodelled, or scattered across systems, no amount of drag-and-drop rescues it. Our angle is deliberately the unglamorous half: modelling clean, performant data sources; building the pipelines and the warehouse that feed them; and embedding Tableau into products through its APIs. We are engineers who happen to be fluent in Tableau, not dashboard decorators — the difference shows up in every report that loads in under two seconds instead of thirty.
That framing also keeps us honest about what Tableau is not. It is a visualisation and analytics layer, not a data pipeline and not a substitute for a well-modelled warehouse. When a client tells us “Tableau is slow”, the cause is almost always upstream: a query hitting an unindexed operational table, a workbook doing joins that belonged in the warehouse, or an extract that should have been a live connection to a proper star schema. We treat the whole stack as one problem, because that is the only way the dashboards at the top of it stay trustworthy.
Best for — Organisations that need self-service analytics and rich interactive dashboards with real governance — and understand that the payoff depends entirely on the data engineering feeding it.
Why teams choose Tableau
Fast, honest exploration
When the data source is modelled well, Tableau lets analysts ask and answer their own questions in minutes rather than filing a ticket and waiting a week. The gain is speed of insight — but only over clean, performant sources, which is where our work sits.
Governance that scales to an organisation
Certified data sources, row-level security, and a permissions model mean one governed definition of a metric can serve hundreds of users, instead of everyone maintaining their own conflicting spreadsheet. That is a real advantage over ungoverned tools.
Dashboards that stay trustworthy
We build the pipeline, the tests, and the refresh logic behind the dashboard so the numbers reconcile and stakeholders stop second-guessing them. A BI tool people trust gets used; one they quietly work around is wasted licence spend.
Why businesses choose Tableau
- You need self-service analytics your business users can actually drive, not another backlog of static reports routed through engineering.
- You have data across several systems and want one governed, certified place where the definitions are agreed and the numbers reconcile.
- You value a mature governance and security model — row-level security, permissions, auditability — because your data or your regulator demands it.
- You want a partner who will build and fix the data engineering underneath, not just style dashboards on top of whatever data happens to be there.
What we build with Tableau
The capabilities this technology is genuinely strong at — and what we most often build with it.
Modelled, performant data sources
We design the published data source itself — the joins, the relationships, the aggregation grain, the extract-versus-live decision — so workbooks query a clean model rather than raw operational tables. This is where most Tableau performance is won or lost.
Calculated fields and LOD expressions
FIXED, INCLUDE, and EXCLUDE level-of-detail expressions and well-structured calculated fields let a dashboard answer questions the raw schema does not directly support. We use them deliberately, and we know when a calculation belongs in the warehouse instead of the workbook.
Tableau Prep pipelines
For lightweight shaping we build Tableau Prep flows — cleaning, pivoting, joining, and unioning — but we are candid about its ceiling: real transformation at volume belongs in the warehouse and an orchestrated pipeline, not in Prep.
Extracts, live connections, and refresh strategy
We choose per source between a live connection to a fast warehouse and an in-memory Hyper extract, and we design the refresh schedule and incremental logic so dashboards are current without hammering production databases.
Embedding via the JavaScript and Embedding APIs
We embed Tableau views into your own products and portals using the Embedding API and Connected Apps for seamless authentication, wiring filters and interactions to the host application — and we tell you early where the API’s limits will bite.
Governance: certified sources and row-level security
We set up certified data sources, a permissions structure, and row-level security so each user sees only their data from a single shared workbook, keeping one definition of the truth instead of a sprawl of forks.
Use cases
Executive and operational dashboards
A governed layer over a modelled warehouse giving leadership and operations teams live, reconciled metrics — with the pipeline and tests behind it so the figures hold up under scrutiny.
Self-service analytics rollout
Certified data sources and training that let analysts across departments explore data safely, without each team inventing its own conflicting definition of revenue or churn.
Embedded analytics in a product
Customer-facing dashboards embedded into a SaaS application via the Embedding API and Connected Apps, with row-level security so each tenant sees only their own data.
Rescue of a slow, distrusted deployment
Diagnosing an existing Tableau estate where dashboards are slow and the numbers are questioned — nearly always fixed upstream in the data model, the extracts, and the pipeline rather than in the workbooks.
When Tableau is the right choice
- Right when a business needs genuine self-service analytics — analysts and domain experts exploring data interactively — rather than a fixed set of static reports a developer regenerates each month. Tableau’s drag-and-drop exploration is its strongest suit here.
- Right when you need rich, interactive dashboards with a mature governance story: row-level security, certified data sources, permissions, and a content server that a large organisation can actually administer.
- Right when the value of BI is high enough to justify per-user licensing, and when the data engineering underneath is already sound or you are willing to invest in it — Tableau amplifies a good data model and exposes a bad one.
- Wrong for embedding a handful of charts into a product at scale on a tight budget: per-user or capacity licensing and the Embedding API’s limits often make a code-based charting approach cheaper and more flexible. Be honest about volume before committing.
- Wrong as a fix for a data problem. If your numbers disagree between systems, your definitions of “active customer” differ by team, or you have no warehouse, buying Tableau will surface those problems faster and more publicly — it will not solve them. Model the data first.
Tableau: pros and cons
Strengths
- Best-in-class interactive visualisation and visual exploration — VizQL makes ad-hoc analysis genuinely fluid for non-programmers.
- Mature governance and administration: certified sources, row-level security, permissions, and a content server built for large, regulated organisations.
- Broad connectivity to databases, cloud warehouses, spreadsheets, and APIs, with the option of live connections or in-memory extracts per use case.
- A deep, long-established ecosystem — calculated fields, LOD expressions, the Extensions and Embedding APIs — and a large pool of experienced analysts to hire from.
Trade-offs
- Expensive. Per-user licensing (Creator, Explorer, Viewer tiers) adds up quickly across an organisation, and the cost is easy to underestimate before rollout.
- It is a visualisation layer, not a data pipeline or a warehouse. Garbage in, garbage out — most “Tableau is slow” problems are data-modelling problems wearing a Tableau costume.
- Without governance it invites dashboard sprawl: hundreds of near-duplicate workbooks, conflicting metric definitions, and stale extracts nobody owns.
- Heavy customisation and product embedding hit platform limits — the Extensions and Embedding APIs are capable but constrained, and past a point a code-based approach is more flexible.
Architecture
We treat Tableau as the top layer of a stack, and we design the whole stack. Underneath the workbooks sits a modelled data warehouse — typically a star schema with conformed dimensions — fed by orchestrated pipelines that extract from source systems, transform to agreed definitions, and load on a schedule. Tableau connects to that model through published, certified data sources, so every workbook shares one governed definition rather than each author rejoining raw tables their own way.
The load-bearing decision is where each piece of logic lives. Heavy joins, aggregations, and business rules belong in the warehouse where they run once and are tested; calculated fields and LOD expressions in the workbook are reserved for genuinely presentation-layer logic. We decide per source between live connections to a fast warehouse and in-memory Hyper extracts, and we structure the Tableau Server or Cloud site — projects, permissions, certified content — so the deployment stays administrable as it grows rather than degrading into a sprawl of orphaned workbooks.
Performance
Almost every Tableau performance complaint traces back to the data, not the tool. A dashboard that takes thirty seconds is usually querying an unindexed operational table, doing joins Tableau should never have seen, or refreshing an extract that should have been a live connection to a proper aggregate. We diagnose with the workbook performance recorder and the underlying query logs, then fix the actual cause — model the source, push aggregation into the warehouse, add the right indexes, or reshape the extract.
Beyond the data, we keep workbooks lean: fewer, purposeful worksheets per dashboard; filters that use context or data-source filters rather than expensive quick filters over high-cardinality fields; aggregation at the right grain; and LOD expressions used sparingly because each one is a cost. The target is dashboards that load in a couple of seconds on real data volumes, because a slow dashboard is one people abandon — and abandoned dashboards are wasted licence money.
Security
Tableau’s governance model is one of its genuine strengths, and we use it properly. Row-level security — enforced through user filters or, better, entitlement tables joined in the data source — ensures each person sees only the rows they are permitted to, from a single shared workbook. Permissions are structured at the project and content level so access is granted by role rather than hand-set per workbook, which is what keeps a large estate auditable.
For embedding we use Connected Apps and trusted authentication so views appear inside your product without exposing Tableau credentials, and we keep data-source connections on service accounts with least-privilege access to the warehouse. Sensitive logic and credentials stay server-side; nothing security-critical is left to a client-side filter that a determined user can remove. Where data is regulated, we design the row-level model and the audit trail as first-class requirements, not afterthoughts.
Scalability
Tableau scales along two axes that need managing separately: users and data. User scale is handled by Tableau Server or Cloud sizing, extract refresh scheduling, and caching — but it degrades fast if hundreds of workbooks each run heavy live queries, which is why a governed set of certified sources over a fast warehouse matters more than any server tuning.
Data scale is a warehouse problem, not a Tableau one. As volumes grow we push work down to a platform built for it — pre-aggregated tables, partitioning, and incremental pipeline loads — so Tableau queries small, purpose-built results rather than scanning raw history. The other scaling risk is organisational: dashboard sprawl. We contain it with governance — certified sources, a clear project structure, ownership, and periodic review — so the estate grows in coherence rather than in clutter.
Tableau integrations & ecosystem
The technologies we most often pair with it — each links to how we work with it.
How we work
We start upstream, not in Desktop. The first phase is understanding the questions the business actually needs answered and auditing the data that would answer them — where it lives, how clean it is, and whether a model exists. Only once we know the data source is sound do we design the workbooks, because building dashboards on unmodelled data is how you end up rebuilding them three months later.
From there we work in thin slices: one real, governed dashboard over one certified source, in production, proving the whole stack — pipeline, model, security, performance — against reality before we scale out. The engineers who build the data model are the ones who build the dashboards and keep them running. That is what we mean by operating what we build: the person choosing an extract over a live connection is the person who will get the support call when a refresh fails, so they choose for the long run, not the demo.
The service behind it
Delivered throughData EngineeringWhat we build with Tableau
The disciplines this technology most often shows up in — from a first build to taking over and stabilising an existing one.
How we deliver
- 01
Discover
We map the system, the constraints and the business it serves — including the parts nobody documented.
Architecture brief
- 02
Architect
Decisions get made, written down and defended before a line of production code exists.
Decision records
- 03
Build
Short cycles against working software. You see progress in the product, not in a status deck.
Shipping increments
- 04
Operate
Monitoring, incident response and iteration. The system is alive, so the engagement is too.
Runbooks & SLOs
Industries we use Tableau in
Domain knowledge changes what gets built. A few of the sectors we know before the first meeting.
Also in Data
Why teams choose us for Tableau
Engineers, not dashboard decorators
We build the pipelines, warehouse, and data model beneath Tableau — the part that actually determines whether your BI is fast and trustworthy. Anyone can drag a field onto a canvas; far fewer can make the data behind it correct and performant.
Senior-led, and we run what we build
The people designing your data model are the ones who operate it, so decisions favour the eighteen-month maintenance reality over the demo. No juniors learning extract strategy on your production estate.
Honest about cost and fit
If per-user licensing makes Tableau the wrong call for your use case — high-volume embedding on a tight budget, say — we will point you at Power BI, Looker, or a code-based approach before you spend the money.
We fix the real problem
When a deployment is slow or distrusted, we diagnose upstream rather than reshuffling worksheets. The cause is nearly always the data model, the extracts, or the pipeline — and that is exactly where we work.
Typical timeline
- 01
Data audit and metric definitions
One to two weeks establishing what the business needs to know, agreeing metric definitions across teams, and assessing the state of the data that must answer them.
- 02
Model and first certified source
Two to three weeks building the modelled data source and the pipeline feeding it, so the foundation is sound before any workbook is drawn.
- 03
First governed dashboard
One to two weeks delivering a real dashboard end to end in production — with row-level security, performance tuning, and refresh scheduling proven on live data.
- 04
Rollout and handover
Expanding certified sources and dashboards, setting up governance and permissions, and training your analysts so the estate stays coherent and your team can own it.
How pricing works
- Fixed-scope builds for a defined deliverable — a modelled data source and a set of governed dashboards — quoted once we have audited the data and agreed the metrics.
- Data-engineering engagements for the pipeline and warehouse beneath Tableau, priced by the modelling and integration work rather than by the dashboards on top.
- Audits and rescues of an existing deployment that is slow or distrusted, priced by assessment — the fix is usually upstream in the data model and the extracts.
- Note that Tableau’s own licensing (Creator, Explorer, and Viewer tiers, per user) is paid to Salesforce and is separate from our fees — we will help you size it honestly before you commit.
Hire Tableau engineers
Need Tableau capacity on your own team? We embed named senior engineers into your existing team — reporting to your leads, working in your rituals — so you add capacity without a hiring cycle.
Hire Tableau engineersCommon questions
Our Tableau dashboards are slow. Is that a Tableau problem?
Almost never. In the large majority of cases a slow dashboard is a data-modelling problem — a query hitting an unindexed operational table, joins that belonged in the warehouse, or an extract that should have been a live connection to an aggregated star schema. We diagnose with the workbook performance recorder and the underlying query logs, then fix the cause upstream. Tuning the workbook alone rarely fixes a problem that lives in the data.
Do we need a data warehouse to use Tableau well?
For anything beyond a small deployment, yes. Tableau is a visualisation layer, not a data pipeline or a store. It can connect directly to operational databases and spreadsheets, but doing so at scale is slow, fragile, and produces conflicting numbers. A modelled warehouse gives you one governed definition of each metric, fast queries, and dashboards that reconcile. Building that layer is the bulk of what we do — the dashboards are the visible tip of it.
How does Tableau compare with Power BI or Looker?
Tableau leads on interactive visualisation and visual exploration and has a very mature governance story. Power BI is typically cheaper, especially if you are already invested in Microsoft, and strong on modelling with DAX. Looker is code-first — its LookML models the semantic layer in version-controlled code, which suits engineering-led teams. The right choice depends on budget, existing stack, and how code-based you want to be. We will give you a straight comparison rather than defaulting to whichever we know best.
Can you embed Tableau dashboards into our own product?
Yes — we embed views using the Embedding API and Connected Apps for authentication, wiring filters and interactions to your application and using row-level security so each customer sees only their data. We are also candid about the limits: at high volume or with heavy customisation, per-user licensing and the API’s constraints can make a code-based charting approach cheaper and more flexible. We will size both honestly before you commit to embedding at scale.
How do we stop Tableau turning into hundreds of duplicate dashboards?
Governance, designed in from the start rather than bolted on once the sprawl has happened. Certified data sources give everyone one agreed definition to build from; a clear project and permissions structure controls who can publish what; and named owners plus periodic review retire stale content. Without this, Tableau’s ease of authoring works against you and the estate fragments into conflicting, unmaintained workbooks. We set the governance up so growth stays coherent.
Building on Tableau?
A technical conversation with the engineers who would do the work. If we are not the right fit, we will say so on the call.