We need to talk about fees

June 7, 2024

No items found.

I had a meeting with a design leader of a London-based Architecture practice last month, it went something like this …

Me:

“AI Image models open up new avenues for creative experimentation during early stages of projects. Custom text models will allow us to create specialised agents that can analyse large bodies of project documents on our behalf, presenting data and knowledge back to us instantly and on demand”.

Another architect, in response:

“OK, I’m intrigued but doesn’t AI just make everything quicker? And if it’s quicker — doesn’t that reduce how much fee we can charge?”

Firstly, I share this concern and I think the core argument is correct, but only if we hold to the assumption that we should continue to price our work exactly as we do today, into the future.

My conclusion is that to reverse the current downward trend and at the same time prepare for greater levels of automation in the future, we need to stop selling time and start selling value instead. A fundamental re-evaluation is needed about how we communicate our value and charge for our work and I think the work to bring about this change has to start now.

The Trend from Percentage to Lump Sum

Over the past decade, the profession has increasingly come to define our value to clients simply via the predicted team-resource costs that we estimate at the outset of projects, rather than being paid in relation to the value that our designs create be that for example as fee per home delivered, or as a percentage of construction which was favoured when I first started out.

When I asked my tutor about fees at university I was sent to look at the old RIBA Mandatory Percentage Fee Curve which was removed in 1986 by the Office of Fair Trading (now the CME). I believe that the fee curve was deemed anti-competitive, but to many architects, it now looks like the holy grail.

While I’m not in favour of price fixing by institutions (that’s illegal), I don’t understand why there would be a problem with anonymised (but accurate) data sharing as a way of establishing dynamic consensus around current pricing and a code of conduct that prohibits irresponsible levels of undercutting substantially below market norms.

From my recent conversations with many practices, we are clearly sitting at a low ebb in the market cycle and Architects and the fee landscape is frankly, abominable. We really need a low tide mark from which to build up their self-worth again — and I believe a reinstated benchmarked fee curve would help us to do this.

Nowadays, I find percentages are mostly being used only sense-check resource plans benchmark as a kind of loose KPI. While they remain part of the “client lingo” and are used during negotiation, it is more often the case that they don’t actually make their way into the contract wordings of our appointments, which are usually expressed as a Lump Sum figure against each stage.

You can see why this shift from Percentage to Lump Sum has happened; many clients don’t like percentage fees, because generally build costs will inflate over time and therefore so will fees.

Over a multi-year project this can be a challenging budget line item for clients, particularly when you backdate increases to previous RIBA stages where build costs had been previously underestimated. All in all, a resource-derived Lump Sum fee structure will mean the client is likely spending less on fees overall.

By the same measure, you can also see some clear reasons why a shift away from Percentages to Lump Sum contract agreements has led to worsening conditions in the profession over the same time, particularly since interest rates have risen to tame inflation:

1. The costs of goods and services have risen by 22.3% since the beginning of Covid (figures from the Bank of England), yet many contracts that we work to were agreed before this point. I don’t know an architect in the UK that would report fee rates having risen at all during the same period and this means a painful combination of dramatically increased practice costs and flat or declining fee income over the same period.

2. Estimation of team resource spend at the outset of a project causes routine underestimation due to a strong cocktail of optimism bias and undercutting, both of which I would argue architects have a natural weakness for due to the culture of our training, but that’s a whole other article.

3. Resource-based fee estimation often means it is harder to recover the costs of abortive work when the brief changes or when the programme extends beyond our control. We can sometimes recoup some losses through fee claims, but it’s an uphill struggle.

4. Projects are regularly placed on indefinite hold without notice, or subject to rolling postponement which can go on for months on end and we aren’t paid for this downtime risk. Sometimes we are given a start date that the team is prepped for, only for the rug to be pulled at very short notice. This situation becomes more acute when the market is flat and means we need team members on standby but we often can’t bring in actual fee income for a large proportion of their time.

5. Architects are often being asked (and agree to) work at risk or for free during the enablement phase of a project. The request usually comes in like this: “We’re looking at a new site, can you run a feasibility study, if it’s a goer the job is yours”. When you ask the client if there is a fee, the answer is no and this is ok, because: architecture!

Overall this means that a substantial percentage of fee earning team time is actually spent on project time that doesn’t generate income or on design competition work that also isn’t paid but costs a fortune — all of which compounds losses if you are already a low-margin business model. Over time, people decide they can forgo a 30% profit target on resource based lump sums, so they drop to 20% and then 10% — because any work is better than no work. Pretty soon, practices are grinding under poor working conditions, flat wages and exhausted teams.

Very simply, resource-based lump sum fees lead to lower overall income than percentage based fees over the lifetime of a project and the gradual erosion of profitability which means widespread professional decline.

“Fees#2” — created in Stable Diffusion using two control nets and a custom LoRA. For full prompt settings on images featured in this article, please scroll to the end (credit Arka Works).

Software Products Taking Market Share by Time-Saving

As we observe the post-GPT landscape, we can observe a suite of AEC products that are gradually targeting a market share that is currently dominated by professionals who sell services.

Seldom will you see an AI or Generative Design tool being promoted without the key metric of ‘time saved’ being reported as the leading value proposition and in the face of widespread productized automation, any service providers trying to win on a “time-taken” financial model, will be losing the battle. A case in point; the US company Testfit recently reported users were saving 75% of their time on their feasibility studies by using the tool.

These are impressive numbers and assuming you want a US-market-centric pattern book analysis as the basis for your development appraisal, this is probably accurate. As we see more tools like this emerge in the UK, we will likely also observe that they are being marketed directly to clients and they do not permit extensive modification and customisation by designers. If the trend continues to grow, we will probably see a reduction from the client side in their expectations about how long things should take and accordingly, how much they are willing to spend. We may even begin to see architects move ‘in-house’ on the client side to enable more direct procurement of design — indeed I have seen a couple of instances of this happening already.

Social Media Influencers are now tripping over themselves to promote their latest “productivity hack” and report on how many more “10X AI-powered apps” they can squeeze into their working routines, as if the only thing that matters in the world of work, is how long things take.

I am sceptical about how true these clickbait claims are, but I also wonder what we will do with all the free time we win back in the future? Presumably, it can be used for creating more mediocre digital content to feed the internet gods? Or as the practice leader I mentioned at the top fears; maybe there is just less for us to do and be paid for and many of us will be out of the job.

That would be the doomsday view. However, if history is any indication, it seems more probable to me that we will simply use any additional time-saved to do more.

This has been true of every substantial technological shift in the modern age — personal computing, the internet, mobile phones — they have all led to an ever more frenzied sense of productivity. At the same time, we could not have foreseen the new role and job creation caused by these industries — do we think AI be different?

Things can be faster, yes, but are they going to be better? With all this talk of speed and automation, I think architects need to urgently change the narrative about what we are selling.

The tangible outputs that we often cite as justification for our fees include; 2D drawing packs, Stage Reports, 3D Visuals, Area Metrics, Regulatory Compliance checks and we make sure these are all delivered in a timely fashion and with competence. But these traditional markers of good service are likely to become far more common commodities in the future through automation — and that means they’ll likely be worth less.

I would argue that these deliverables have always been by-products of the core value we actually offer as architects — I see things more like this:

Things we sell that are likely to be automated vs things that have greatest value and are harder to automate

My take is that the items in the left column are mechanical and will, with the advent of more Generative Design techniques and AI, become easier and cheaper to deliver over time. These outputs will be more commonplace and therefore lower in perceived value.

The items on the right are far less easily replaced — these are the types of rare attributes that architects can truly excel in delivering and they tend to be very human-centric skills that are contingent on good communication, enduring professional relationships and strategic thinking.

We need to gradually shift the narrative about how architects deliver value to clients over to this column. This is what clients are paying for and it is worth a great deal more than is perceived at present. If we continue to sell our time based on the left column I believe we will be much more vulnerable to replacement technologies.

On the plus side, architects are perhaps blessed that in the digital age, we are still in the business of making physical infrastructure that people depend on. Architecture is at its essence a physical phenomenon responding to human needs and we therefore need to focus on these aspects of our services when we produce fee proposals.

“Fees#3” — created in Stable Diffusion using two control nets and a custom LoRA. For full prompt settings on images featured in this article, please scroll to the end (credit Arka Works).

The Scope of AI in the Workplace is Broadening

A mantra that has been repeated almost to the point of cliche, is that “AI will do the drudgery, so you can focus on the real work”. This line is smart because it’s catchy and who could disagree with this value proposition — do you like grunt work?

Actually, when you start to play out the kind of tasks you might want an AI to perform within a practice you may be shocked at how much of the non-drudgery it will start to help with.

In the startup space, there are products being developed right now that are aiming to transform the way we deliver much more human-native workplace activities than you might expect. These include onboarding, sales, recruitment, training and even management within the workplace, via digital avatars. These digital personas will adopt the likeness of a senior team member and disseminate instructions tailored to each employee with a simple prompt.

Don’t believe me? Here is a video of the tech entrepreneur Reid Hoffman interviewing an AI avatar of… wait for it… himself:

Reid Hoffman interviews himself. Credit: https://www.youtube.com/watch?v=rgD2gmwCS10

You might look at this example and be horrified. You may think to yourself; “surely people will opt for human connection, being in the room, looking a person in the eye?”… I would respond by saying that all truly transformational technology must overcome some kind of massive taboo or norm just like this.

Who would have guessed in 2019 that so many people would prefer to take meetings from home than attend work in person or that critical business meetings could happen via a tool like Zoom? Do we think the demographic shifts in behaviour from the next generation of the workforce will be more or less inclined to ask sensitive questions to a digital avatar, rather than a real person? We may not like the answer, but it doesn’t change reality.

When it comes to AI helping people do things faster, I have seen lots of big outlandish time-saving numbers being proposed by commentators that I respect and as a pure anecdote, I am seeing a general convergence of opinion around the loose target of around 30% aggregated reduction in time spent on typical office tasks, when comparing a Generative AI-centric workflow with a traditional method in the next 5–7 years.

If we discount that expectation by half, a 15% change in productivity would still be very impactful for architects and how we charge clients for our work.

If you’re still not sure about what AI could mean for us all, let’s game-theory out five possible outcome scenarios from here about what AI will mean for the profession:

Future outcomes matched with logical business strategy

In all situations other than option 1, you need an AI strategy.

In 3 out of the 5 scenarios, you will see fees impacted by the implementation of AI longer term which means we need to start changing the narrative around fees and value as a profession.

“Fees#4” — created in Stable Diffusion using two control nets and a custom LoRA. For full prompt settings on images featured in this article, please scroll to the end (credit Arka Works).

The Deflationary Impact of Technology

It’s been said that tech has gradually dematerialised objects and “stuff” over time. Every big step forward causes the gradual eradication of everyday items; the rolodex, paper calendars, the walkman, dictation devices, encyclopaedia britannica, nostalgic curiosities that all fit into a phone now.

As tech squashes more physical things into software, it also allows us to scale our efforts, to reach higher levels of productive output without the same unit economic input of energy. It’s gradual so we often don’t realise it’s happening, but as it does — the marginal cost of production drops too, meaning when we get beyond a certain scale, we can produce more for cheaper.

I would argue BIM-products (like Revit) have gradually done this to architecture. They have enabled us to scale our efforts, to leverage our clicks. Yes they are a real effort to get going, but once you have your systems up and running, incrementally scaling your drafting up to ever more square meterage becomes increasingly easy if there is widespread repetition beyond a certain scale. It also means a medium sized practice can compete for the same work as a larger scaled practice.

Working in this way is just not the same as manually drafting and issuing drawing packs. But have we used this aid to produce things more quickly and make more profit in the process or have we used it to just do more? I would love to see a chart as evidence, but I would guess the number of drawings we produce at tender has risen in direct correlation to this productivity increase, meaning for the architect at least, it hasn’t really proven to be an economic benefit — particularly as we all decided to roll the “BIM Contributor” role into our standard terms of service nowadays. Revit also costs a bomb, in case you didn’t know.

We may extend this analogy to AI — which could have an even greater deflationary impact than BIM. So the question is; are we going to use it to improve our business performance and quality of outputs, or are we just going to use it to do more, for less?

So how do we change the conversation about Fees?

Here are some simple steps we should consider taking as a collective profession if we want to bring about change:

1. In light of AI adoption, we need to rapidly adapt how we communicate about the value of what we’re selling.

See table above.

2. Stop working for free.

When we work for free, we anchor our value at zero. We need to encourage a wide culture of practice that is fundamentally hostile to the idea of free-work. Urge practices to sign up to a commitment that they will not endorse working for free. Encourage employees to call out such practices. Ask clients to sign up to similar commitments saying they won’t ask for it — this should be a basic Social Value commitment from all clients to treat suppliers fairly. I can’t think of another professional service that is so routinely asked to carry out work for free.

3. Teach Part 1s to really value themselves and tell them how to charge for their work properly from day one.

Encourage discussion about fees and the economics of practice at the outset, don’t wait for Part 3. Focus education around building skills in the the “right-hand column” that will be less prone to automation in the future.

4. Anonymously share our fee data via professional bodies for every proposal we send out.

We need a robust cross-sector and scale benchmarked Fee Scale that is widely contributed to and kept up to date.

We don’t talk about fees openly, but we should — the culture has to change because the undercutting is becoming more widespread and it’s harming practice. If the RIBA or ARB don’t want to help us with this, we should do it ourselves. Fees should be shared anonymously with a live database that is accessible to all members, an open ‘order book’ similar to other free markets that operate on the basis of bids and asks.

5. Move back to Percentage or “Value-based” Fees.

Create consensus that in order for the profession to meaningfully survive into the future, we need to move away from the time resource-derived Lump Sum fees and back towards contractual Percentage Fees that are tethered to actual project values with back dated-adjustments to previous stages where costs increase. This shift needs to be endorsed and encouraged by professional bodies.

You need healthy practices to create good architecture and as long as our clients and end-users are human then highly skilled architects will remain best placed to serve them.

But sure, if our clients become some form of AGI, then all bets are off.

👇

DOWNLOAD LINK TO THE IMAGE CREATION TECHNIQUE

🔗

Thank you for reading!

If you enjoyed the images — follow Arka Works on Instagram:

Arka Works

Keir is an AEC Domain Expert operating with one foot in architecture and one in live software development.

He founded Arka Works with a mission to prepare the architecture profession for AI-driven change. He does this by helping Architects, Clients and Startups to effectively apply the latest Generative Design and AI tools to the work they already do, so that they can adapt to a rapidly shifting professional landscape.

Date

Published

Subject

Link

We need to talk about fees

More like this

In the second post of this series we are looking at what I’m calling “Auto Optioneering”.

Generative Design & AI Trends in Architecture Series: “Auto Optioneering”

In the second post of this series we are looking at what I’m calling “Auto Optioneering”.

August 14, 2023