Building a TaxTech team

"My employer didn't take kindly to failure. I'm not exaggerating when I say my job was on the line."

Some Context

In 2015, I joined a “startup”.?

But, by the time I joined the ride, Uber had already ridden roughshod over well-established businesses in 61 countries. This ‘upstart’ connected millions of drivers with riders, providing more than 3 million trips in 340 cities daily.?

The expansion was pedal to the metal—the company aggressively launched a new city every five days, and headcount was accelerating to keep pace.

Want a crazy stat? In under eight weeks, a new joiner was more tenured than 25% of the existing staff. Perhaps the hire-fast, fire-fast mentality had something to do with that.

“Always be Hustling” and “Growth at all Costs” were two of the mantras.

Even back then, labelling Uber as merely a “ridesharing platform” was an oversimplification.

It wasn’t only connecting drivers to riders; it was helping to deliver ?? kittens in London, ?? puppies in Pennsylvania, ??mobile phones in Malaysia, ?? chopper charters in Dubai, ?? car leasing out of Singapore, ????? boat trips in Croatia, ?? fast food, ?? autonomous vehicles, flying cars…. you name it.

Nothing was impossible. Everything was tech-driven.

The Tax Team

As one of six full-time tax professionals in the Amsterdam HQ, we supported APAC, EMEA, and LatAm with advisory, market launches, new service offerings, compliance, controversy, planning, optimization, and whatever else came our way.?

I recall that my first call was to support Brazilian income tax. I wondered which part of the “EMEA VAT Manager” job description I misread. This would change as the tax function matured and regional teams developed.

The hours were long, the deadlines short, but the work was relentless and “reliable as running water”.

Despite our business's massive strides in automation and tech, tax compliance was largely manual.

We would pull reports manually, perform reconciliations manually, prepare tax returns manually, and run exception tests—you guessed it—manually.

Doing this for some 60+ markets for close to 200 entities was exhausting. In total, there were roughly 11,000 annual tax-related filings.

In total, there were roughly 11,000 annual tax related filings.

Faced with a to-do list that was growing and not shrinking, our EMEA lead compared us to plate spinners: “Some will fall. Keep the right ones spinning.”

Tax audits, tax data sharing obligations, and transaction-level digital reporting for the company and on behalf of Drivers, Couriers, and Restaurants were on the horizon.

You didn’t need to hop in a DeLorean taxi to realise that we would be swamped in the future. We would either need to hire an army or invest in tech.

Spoiler: we didn’t hire an army.

How And Why Was “Uber TaxTech” Created?

It wasn’t due to strategic foresight.

The Problem

Simply, the risk I dreaded materialised: my usual contact for extracting system data to prepare the quarterly VAT return in time for the tight deadline was unavailable. Neither was the backup—or the backup’s backup.?

Clueless about accessing our data - I was stuck.

What columns to SELECT, FROM which table, and WHERE to find the conditions and filters to apply (SQL joke). I could not access what was needed to submit.

It wouldn't be so bad if the amounts weren't astronomical.

Uber dealt in exponentials and extremes—it's not 2x—it's 10x. It's not millions. It's billions. It's not "that's OK, it happens". It's "you're out".

Our VAT group was processing "substantial amounts". And we were in a quarterly refund position. I had one mission-critical task, and I couldn't do it.

My employer didn't take kindly to failure. I'm not exaggerating when I say my job was on the line. And over something so stupid. So basic.

A Big Bold Bet

I needed a finance professional with VAT/GST knowledge and data analytics skills. Luckily, Uber’s finance team had no shortage of capable, driven, tech-savvy individuals.

One stood out.?

She is a doctor of philosophy in applied mathematics with a specialisation in stochastics (appropriate given the chaos of those days). She obtained the data, improved the process, and removed many manual tasks.

The savings flowing from the improvements were significant.

This example was used to write and present a short business case to our head of tax and obtain approval to create a small squad (of three) exclusively dedicated to driving efficiency via technology.

Initial results were positive - we managed to remove several manual chores to the great satisfaction of colleagues (and department budgets).

I understand the acceptable way to brag about this is to say we "enabled our specialists to focus more on value-added tasks".

The next application to expand was approved. Here's how it evolved:

1. By 2018 we grew to five FTE's.
2. By end-2019 Uber TaxTech had had ten FTE's and its own internal goals.
3. By 2021 - eleven FTE’s with backgrounds in data analytics, data visualistion, machine learning, automation, mathematics, statistics, accounting, and of course - tax.*

*An addition of one FTE in two years is not very impressive... until you recall that this was during COVID when most teams 'downsized'.

None of that would have been possible without the support, belief, and understanding of senior company leadership, who were committed to this team's success from the outset. Our Head of Tax, Francois Chadwick , strongly appreciated tax technology and firmly believed in using it when it made sense.

How did TaxTech Prioritise Work?

A member of TaxTech developed an internal tool (essentially a formula with a user-friendly interface) to score potential projects based on a wide range of criteria (market, volume, local tax landscape, audit history, business priority, the “CEO said so”, etc.).?

The score ranked potential projects in order of priority and suggested which should be undertaken based on estimated value to the company. While only a suggestion, it was fine-tuned over time and provided a framework to settle disputes between competing priorities.

Who Did The Team Report To??

The TaxTech team was a separate function that reported to the Head of Tax. He had an unobstructed and unfiltered line of sight into the team's activities. The blockers, successes, and failures were appraised every two weeks.

How Was TaxTech Structured?

We created three sub-groups:?

1. A core tax data team was the gatekeeper of our data and the go-to for anything related to tax information stored in (or scattered across) our systems. They are the specialists who understand tax and our upstream systems.
2. An internal TaxTech matters team - which focussed on supporting our accountants and tax specialists team (e.g. VAT return preparation, audit responses, etc);
3. An external TaxTech team: This team, nick-named CRISP, acted as interpreters between tax and the different product teams (tax calculations, tax receipts, tax invoices, tax summaries, and tax profiles). They helped gather requirements and explained tax matters to tech specialists and vice versa.

The lines between these three teams were blurry - they often worked closely together and supported one another.

How Did TaxTech Track & Monitor Performance?

Everything the team could be converted into a dollar value.

1. Time saved - converted into hours and minutes of average regional salaries. The team also tracked their own performance. For example, the Data team?monitored ticket volumes over time and their ability/time to resolve them.
2. Reduced audits - converted into expected audit fees foregone and time spent (based on historic fines and other internal and external data sources).
3. Reduced tickets - converted into cost-per-ticket savings (we knew the cost of the back and forth of tickets as we were well connected to CommOps).
4. Job satisfaction - tracked in qualitative feedback and measurable over time with headcount churn (feedback from exit interviews was key). While rare, some solutions we deployed infuriated rather than enabled users. On one occasion enough for them to leave. Lessons were learned.

What are some examples of what TaxTech did?

Here are three examples of literally hundreds of projects:

1. Tax return automation—improving the efficiency of tax return preparation, review, and submission through technology, including Alteryx , Tableau , and home-built systems like T2C ("Tax to Compliance").
2. Automated responses to tax authorities—automating the response to tax authority queries relating to individual Platform Sellers (before the days of DAC7). Specifically, the tools built would read and translate the tax authority questions received and extract the necessary information from our systems into a prepared format for review and approval by a regional tax manager.
3. New business tax modelling—the team would model the tax impacts on business (e.g. moving from an agency to a merchant model). When multiple model changes were on the table, the team would compare and contrast how transaction flows would go through each model and offer insights to business of the differences between them. Pretty important for pricing and driver earnings. This required total clarity on CREAM classifications that were the codification of all transactions and on which many transactions, taxes, fees, and tolls were constructed (CREAM, by the way, stands for Cash Rules Everything Around Me - a nod to the Wu-Tang Clan song)

Would I do anything differently?

I'm no Edith Piaf - "No Regrets". I'm more Sinatra - "regrets, I have a few", which I'll now mention. I wish we:

1. Were more strategic. Consider the impact of future obligations sooner and plan (rather than react) on how our function needs to be transformed. Which tasks will consume the most resources? What skills, technologies, and processes must be invested in to ensure a smooth transition? Today, if I were back in the house, I would focus on data accuracy, digital reporting, and learning to prompt large language models.
2. Avoided hacks. We got very good at hacking things together and deploying "temporary" solutions that were "good enough". There is nothing more permanent than a temporary solution. And these temporary solutions fester. Some of the things we hacked together back then still plague teams today.
3. Spread out faster. At the outset we were centralised which was fantastic for the EMEA function we initially supported, but other regions needed help too. We should have supported them sooner.
4. Crafted a clear career path for TaxTech professionals. This came far too late - partly because I didn't have a template to apply to the unbeaten path we were exploring. Drafting job descriptions was challenging enough, but I would have loved to provide clarity on career opportunities sooner.
5. Got a grip on the data. By that, I mean figuring out who owns the data and where the accurate sources are to avoid reconciliation issues in the future. We could not rely on our ERP data because it lacked the necessary detail, but we probably could have leveraged it more.

There is nothing more permanent than a temporary solution

There are many other things that I would do differently, including get a picture of the whole TaxTech team together so I could post it here.

Concluding thoughts

I left Uber in 2021. Also several key members, such as Raphael H. , Lili Glazier , Bessie Pintor, PMP , and Yen Le moved on to bigger and greater things. However, much of the team still exists under the guidance of Zhanna Gres . Edith Zeegelaar , Wesley Bouman and other OG's like Naina Himatsinghka continue to build on the foundations they laid years ago. Others, like Ava Doncheva , moved into FinTech yet still work and support TaxTech on a daily basis.

I hear the team plays a critical role in the company's tax operations globally. In a recent conversation, my good friend and Uber colleague Oleh Fedusiv, LL.M. , LL.M. (EMEA Tax Team) said, "I can't imagine how people got the necessary data in the past without TaxTech".

I instantly thought, "I can. And it's not pretty".

In a recent webinar discussion with an industry colleague from Airbnb, Erich Tschopp, he offered a different (perhaps better) way to measure impact on stakeholders rather than USD (which is what I relied on).

Other leading minds at well-known tech companies confirmed that my experience in shaping Uber TaxTech, although thrilling, was not unique.

Just listen to Liz Gallagher from Meta , Misha Narinsky from 谷歌 and Ha Pham (currently at Booking.com but formerly Uber TaxTech). These leading companies have well-established TaxTech functions (albeit some under a different name).

Should others consider building a dedicated TaxTech team?

Yes. Such teams are no longer a lucky luxury. They are a necessity.

If you want to know more, connect.