ServerVultr

Dataflows Gen2 vs Data Factory in Microsoft Fabric: What’s the Difference? I have been asked this question several times in recent training sessions on Microsoft Fabric, so I jotted some notes down here.

Microsoft Fabric brings together the best of Microsoft’s data engineering, data integration, analytics, and AI capabilities into a single unified platform. For many teams adopting Fabric, one of the first questions that arises is:

“What’s the difference between Dataflows Gen2 and Data Factory Pipelines?”

Both can move, transform, and prepare data. Both live inside the Fabric experience. And both can be scheduled, monitored, and orchestrated. However, they serve different purposes, offer different strengths, and work best in different parts of the modern data lifecycle.

This post explains the key differences and provides practical examples to help you choose the right tool for your scenario.

What Are Dataflows Gen2?

Dataflows Gen2 are Fabric’s low-code data preparation and transformation solution. They are built on Power Query, the same engine used in Power BI and Excel, giving analysts and citizen developers a familiar, friendly interface.

Key Characteristics

• Low-code / no-code: Drag-and-drop transformation steps rather than writing SQL or Python.
• Power Query based: Ideal for data wrangling, cleansing, merging, shaping, and enrichment.
• Works well for mid-volume data: Excellent for business data preparation and M-code transformations.
• Outputs straight into Fabric: Can load data into Lakehouses, Warehouses, and KQL databases.
• Accessible to analysts: You don’t need a data engineering background to use it effectively.

When to Use Dataflows Gen2

Dataflows Gen2 shine in scenarios such as:

• Self-service data preparation for analysts building semantic models.
• Ingesting business application data (Excel files, SharePoint lists, Dataverse, SQL).
• Quick transformations such as splitting columns, merging tables, cleaning text, or deduplication.
• Prototyping datasets before handing them over to engineering teams.

If you know Power Query, you’ll feel at home immediately.

What Is Data Factory (in Fabric)?

Fabric’s version of Data Factory combines two things:

1. Pipelines – orchestration and data movement.
2. Dataflows (Power Query) and Notebooks (Spark) – heavy-duty transformation for engineers.

It is Microsoft’s full data integration and ETL/ELT platform, now tightly integrated into Fabric.

Key Characteristics

• Enterprise-grade orchestration with pipelines, triggers, and dependency management.
• Powerful connectors for large-scale ingestion, especially from cloud and on-premises systems.
• Supports Spark notebooks and data engineering workloads.
• Handles high-volume, complex pipelines.
• CI/CD friendly and suited for production data engineering.

When to Use Data Factory

Data Factory is designed for more complex engineering tasks, such as:

• High-volume ingestion from operational systems, APIs, or files landing in cloud storage.
• ETL/ELT using Spark notebooks, SQL scripts, and pipeline activities.
• Orchestrating multi-step workflows, including branching, loops, and conditional logic.
• Copying terabyte-scale datasets from Azure SQL Database, Synapse, ADLS, AWS S3, Oracle, and more.
• Building production-ready pipelines with monitoring, retries, and error handling.

If you are familiar with Azure Data Factory, this will feel like its next evolution within Fabric.

Dataflows Gen2 vs Data Factory: How to Choose?

Here is a simple way to think about it:

Choose Dataflows Gen2 when:

• You want low-code data shaping.
• Business analysts are preparing their own datasets.
• You need simple ingestion or transformation.
• The data volumes are small to medium.
• The source systems are Excel, SharePoint, Dataverse, or SQL.

Choose Data Factory when:

• You are building enterprise pipelines.
• You need orchestration, scheduling, and dependencies.
• You are working with large or complex datasets.
• You require Spark, notebooks, Data Engineering, or SQL pipeline logic.
• Data movement needs to be integrated into CI/CD or operated at production scale.

A Combined Approach

In many organisations the best approach is both, working together:

• Data Factory pipelines handle ingestion from source systems into the Bronze layer.
• Dataflows Gen2 then apply transformations to shape and enrich the data for the Silver layer or the semantic model.

This layered approach provides scalability, governance, and flexibility while still enabling self-service analytics.

Need help applying this in practice?

If your organisation is using Power BI or Microsoft Fabric and needs clarity around architecture, governance, or next steps,
The Data Platform Accelerator is designed to help.

It’s a focused engagement that assesses your current setup and delivers a practical roadmap you can execute.

👉
Learn more about The Data Platform Accelerator

Summary

While Dataflows Gen2 and Data Factory sit side-by-side in Microsoft Fabric, they target very different users and workloads:

• Dataflows Gen2 → Best for analysts, low-code transformations, quick data preparation.
• Data Factory → Best for engineers, enterprise data pipelines, complex ingestion, and orchestration.

Understanding these differences ensures your team uses the right tool for the right job, helping you build efficient, scalable, and well-governed data solutions in Microsoft Fabric.

If you’re teaching or adopting Fabric, this distinction is one of the most important concepts to get right early.

Our 2026 Microsoft Data & Analytics Training Schedule Is Live

We’re excited to share our public training schedule for 2026, covering three of the most in-demand Microsoft data certifications:

• PL-300 – Power BI Data Analyst (3 days) – £995 per seat
• DP-600 – Implementing Analytics Solutions Using Microsoft Fabric (4 days) – £1,200 per seat
• DP-700 – Implementing Data Engineering Solutions Using Microsoft Fabric (4 days) – £1,200 per seat

Across 2026, we will run 24 instructor-led courses, split evenly between UK time and North American time. UK and US courses do not overlap and do not run in the same week, and we’ve planned around key UK bank holidays and US federal holidays.

---

2026 Course Schedule (UK & North America)

All dates below are scheduled in separate weeks for UK vs North America delivery each month.

Month	UK (UK time)	North America (NA time)
January	PL-300 — Mon 12 Jan to Wed 14 Jan 2026	DP-700 — Mon 26 Jan to Thu 29 Jan 2026
February	DP-600 — Tue 10 Feb to Thu 12 Feb 2026	PL-300 — Tue 17 Feb to Thu 19 Feb 2026
March	PL-300 — Tue 10 Mar to Thu 12 Mar 2026	DP-700 — Tue 24 Mar to Fri 27 Mar 2026
April	DP-700 — Tue 7 Apr to Fri 10 Apr 2026	PL-300 — Mon 20 Apr to Wed 22 Apr 2026
May	PL-300 — Tue 12 May to Thu 14 May 2026	DP-600 — Tue 26 May to Fri 29 May 2026
June	DP-700 — Tue 9 Jun to Fri 12 Jun 2026	PL-300 — Tue 16 Jun to Thu 18 Jun 2026
July	PL-300 — Mon 6 Jul to Wed 8 Jul 2026	DP-700 — Tue 14 Jul to Fri 17 Jul 2026
August	DP-600 — Tue 18 Aug to Fri 21 Aug 2026	PL-300 — Tue 25 Aug to Thu 27 Aug 2026
September	PL-300 — Mon 7 Sep to Wed 9 Sep 2026	DP-600 — Tue 15 Sep to Fri 18 Sep 2026
October	DP-700 — Tue 6 Oct to Fri 9 Oct 2026	PL-300 — Tue 20 Oct to Thu 22 Oct 2026
November	PL-300 — Tue 10 Nov to Thu 12 Nov 2026	DP-700 — Mon 30 Nov to Thu 3 Dec 2026
December	DP-600 — Tue 8 Dec to Fri 11 Dec 2026	PL-300 — Tue 15 Dec to Thu 17 Dec 2026

---

Who These Courses Are For

Our public courses are particularly well suited to:

• Data analysts and BI professionals
• Data engineers and analytics engineers
• Consultants and contractors
• Teams transitioning to Microsoft Fabric
• Organisations standardising on Power BI and the Microsoft data platform

We aim for an average of 10 attendees per session, keeping class sizes small enough for meaningful interaction, questions, and discussion.

---

Reserve Your Seat

If you would like to reserve a seat, discuss group bookings, or explore private or tailored delivery, please get in touch.

👉 If you would like to reserve a seat, please get in touch and contact us.

Early expressions of interest help us confirm capacity and ensure you get the dates that work best for you and your team. If your prefer team training contact me directly and we can disucss what you need

When the Data Is Right (and I Ignore It Anyway): A Festive FPL Reflection

I’ve been picking my Fantasy Premier League team for a few weeks now using what I’d like to believe is a sensible balance of data, logic, and gut feel. Unfortunately, every now and then, the gut gets a bit too confident, the excitement kicks in, and I convince myself that this is the moment a player finally explodes into life.

This week’s confession: I wasted a transfer on Isak.

On paper, the logic wasn’t completely mad. He scored, he looked sharp enough, and I told myself that the first goal would open the floodgates. Classic FPL optimism. The kind that ignores everything you’ve learned over years of playing the game and replaces it with vibes and hope.

The problem? The data never agreed with me.

Despite the goal, the underlying numbers never really shifted. The xG didn’t spike, the involvement didn’t suddenly increase, and the minutes picture wasn’t as convincing as I wanted it to be. I chose to ignore that because it felt like the right moment. Unsurprisingly, the data won that battle.

To make matters worse, Isak hasn’t really made the impact I hoped for, and with Ekitike now firmly in possession of the shirt and actually scoring goals, it’s becoming harder and harder to justify holding him. If I’m being brutally honest, if I were Arne Slot, Ekitike would be my starting option right now. He’s converting chances, influencing games, and doing exactly what you want your striker to do.

And FPL, at its core, is a game about minutes and impact.

So that naturally leads to the next question: if not Isak, then who?

Going Back to the Numbers

This is where my Power BI model earns its keep. I track xG, xA, shots in the box, touches in the penalty area, and a few other useful indicators that help separate “good FPL picks” from “players who look busy but don’t return”.

With Christmas approaching and fixtures piling up, I thought it would be a bit of fun (and vaguely sensible) to look at which players are currently overperforming and underperforming their xG. Not as gospel, but as context.

Overperformers are often riding a hot streak that can cool quickly. Underperformers can be traps… but they can also be opportunities if minutes and fixtures line up. This is where judgement still matters.

And yes, I say all this while fully aware that I’d just ignored the data the week before.

The Wolves Problem (and Why It Matters)

I’m writing this having already made my transfer for the week, and Wolves played Manchester United on Monday night. As a Wolves fan, that match was painful. As an FPL manager, it was alarming.

Wolves somehow managed to make a distinctly terrible Manchester United side look competent. That tells you everything you need to know about where they are right now. They look beaten before games even start, confidence is low, and structurally they’re all over the place.

Which brings us neatly to Arsenal.

Wolves are away to Arsenal, Arsenal are in strong form, and this has all the makings of one of those games where things could get ugly quickly. Arsenal at home against a side with no belief is exactly the kind of fixture I want to attack over the festive period.

So I made the move.

The Transfer: Isak ➝ Gyökeres

Out goes Isak, in comes Gyökeres.

The thinking is fairly simple. He looks fit again, the minutes should be there, and Arteta (or “Lego Head”, depending on your preference) will want him involved. Arsenal will dominate the ball, create chances, and if Gyökeres gets service early, this could be one of those statement games.

Is it risk-free? Absolutely not.

But this feels like the right moment to take a calculated punt, especially with Manchester City and Haaland facing a tricky trip to Palace. That fixture has “awkward, frustrating, low-return” written all over it.

Captaincy Roulette

Which brings me to the captaincy.

I’m seriously considering captaining either Gyökeres or Rice this week and backing Haaland to blank. It’s uncomfortable, it goes against rank-protection instincts, and it could backfire spectacularly — but this is also the time of year when calculated risks can pay off big.

Playing safe over Christmas is often how you get quietly overtaken.

Looking Ahead

The other big positive is squad flexibility. When the gameweek rolls over, I’ll have five free transfers and one still in hand before the next deadline. That puts me in a strong position heading into the festive chaos — rotation, injuries, surprise benchings, the lot.

The key lesson this week? Trust the data — but also understand why it’s saying what it’s saying. Ignoring it entirely rarely ends well, but blindly following it without context is just as dangerous.

�