Carlota Perez has been on the reading list. Her frameworks on technological revolutions are insightful and struck a chord when layered on top of what’s happening in AI.

Creative Destruction in Engineering

The phrase "creative destruction" often gets used loosely, but Perez gives it real texture. When a new technology revolution arrives, it doesn't just replace what came before, it collides with it. The old infrastructure, old habits, old business models, all still running, all suddenly awkward.

“Each technological revolution begins with a group of core industries, usually involving some energy source or another all-pervasive input, a new infrastructure and a few main products and processes. From there it spreads to the most closely connected industries, forming a strongly interactive constellation.”

This kind of cascading diffusion is happening now, and the technology today has its clearest applications in engineering.

Our software teams are talking about the new practices emerging in terms of “level” systems, a way to frame the disruption wave and provide for an end target - which we are seeing deployed at various scales across the portfolio. Very on-brand for an industry with a high affinity for games.

References

• Stripe’s minion system is one scaled system people are watching on the internet: https://stripe.dev/blog/minions-stripes-one-shot-end-to-end-coding-agents

• Lee’s list of L3 and L4 (Spec Engineering) tools moving fast in late 2025 and 2026.

  https://github.com/stars/leegold/lists/ai-engineering

What are people building at L4?

Having played around with Claude Code - it seems possible to hit L4 out of the box with subagents. Keeping it turning continuously (at least for investing work) seems a pretty tall ask. I get in engineering there is an endless amount of code and features to write but from what I am hearing on the ground - this L4 “engineering warp-speed machine” is a hard beast to feed with work constantly.

I’d love to chat if you have seen an L4 machine in action.

Judgment, Creativity and Design

Right now we have a sense of how these new systems should work. Each software R&D company will have to adapt or risk falling miles behind. What remains hard is knowing what to point the warp-speed machine towards.

“AI compresses the distance between idea and execution. It does not compress the distance between good judgment and bad judgment. It amplifies whatever you point it at. Strong teams with clear strategies will get faster and more focused on what matters. Weak teams with vague strategies will get noisier and more distracted. Clear thinking compounds. Confused thinking unravels. The gap does not close. It widens.”

Disruption is here. Let’s go.

There's been an image from a Goldman Sachs research report making the rounds across the internet. The sentiment I'm hearing from people looking at this is that the price reaction of software companies is indicative of future earnings collapse, to near zero, as expressed by the lagged collapse of earnings in the chart below.

I think the story missed here is one of dispersion. Below is stock price data from three individual newspaper companies during that era. You’ll see from 99-02 on the left most chart a period of consistent performance relative to each other. Post 2002 to the end of 2024 (matching the starting period of the Goldmans chart above) - you see a precipitous decline below 100 for all three stocks which bottomed at the GFC before enormous dispersion in returns across the three companies. The star performer Schibsted (now Vend) fully embraced the internet and came out with a market leading portfolio of internet classifieds and marketplaces that cannibalised their print business.

I remember my time at Citi. The junior team members spent hours manually extracting data from tables. Inputting numbers from PDFs, one by one, for financial analysis. For some reason, we didn’t have reliable OCR or extraction tools internally. This was in 2019!

I got so frustrated I wanted to build my own micro-service: email out an extract, process it with a Lambda using the latest document extraction tech, and email it back to myself. Unfortunately, that would have been a massive security breach. Easier than trying to navigate internally to buy Adobe Pro though…

There were other instances of this kind of remarkable lag behind true tech capabilities. I automated our weekly newsletter mail-out using VBA—recreating a live document linked to Factset, publishing to PDF, and drafting emails, all within Excel. We could have used Mailchimp.

Once I discovered VBA’s connectivity to Outlook, I found it quite amusing to flood my colleagues calendars with hundreds of 5 minute long meetings 😂

Jokes aside - it was the first time I felt a marked difference in rate of change between the world of work and my personal life. No one in the org could buy or change practices that quickly. Although, reflecting now, I wonder if people would have easily concluded the juice was worth the squeeze.

Back then there were far fewer software vendors than today. Now we have hundreds of vendors across every vertical vying for attention. The technology to replace and improve business has always been available. The challenge is navigating the fog and driving change. That’s why enterprise software isn’t going to implode overnight.

Customers want to buy clarity

The opportunity I see coming down the turnpike: helping companies transform and adopt AI.

Customers want to buy clarity (and change management).

Right now, the large software vendors are selling AI features and agents by bundling. “Notion + AI chat.” We’re using this internally already. The Notion environment probably has everything I need to adapt the EVP business and drive productivity improvement. But I’m still running AI workflows outside of Notion on Claude Code. No one has driven the change management yet to lock us in.

Imagine where the issues sit in other industries. A transformation approach to selling might be a strong differentiator. Use your software as a centrepoint to build AND transform the business’ workflows. Use your organisation’s domain expertise to select what matters and offer a breadth of tools/integrations.

What impacts rate of change within a company?

Complexity of the organisation. Size, number of departments, and scale of the tech stack all slow things down. How quickly can you move a 10,000-person org versus a 100-person org?

Scope of service. Entirely digital service businesses, such as marketing or design agencies, may have a fast rate of change. White-collar versus blue-collar versus mixed matters. Offline businesses like service providers, farmers, and construction may have a slower rate of change.

Level of digitisation. This one I’m torn on. Companies that haven’t already adopted tech are more of a blank slate for AI workflows. At the same time, they don’t have the same familiarity and foundations to build from—data repositories, for instance.

It’s war-time again in product

This diagram shown to me by Glen Kruger at Raymond James articulates very neatly a feeling I’ve had about the new war-time in the product world.

New AI-native software competitors are offering completely new experiences (characterised by the agency Y axis) which emerged in the form of things like legal drafting in 2024, high fidelity transcription/note-taking in 2025, to agentic systems in 2026. They’re all competing to be more that just the above. That more is a fully fledged system that contains, put simply, a lot of data and a lot of automations.

The incumbents are coming at it from the other direction. How do we incorporate these game changing new use cases into our platforms.

What an exciting time.

When code is free, the question stops being “can we build this?” and becomes “should we?”

What becomes scarce when code is abundant?

There’s this economic concept called complementary goods. Cod and salt. Cars and fuel. When one gets cheap, the other gets valuable. The salt cod trade was one of the earliest global commodity businesses - cod needed salt for preservation, and for centuries the two were economically bound at the hip.

Complements work in ecosystems too. I’m writing this from a cafe that sells coffee, food, space, and retail product. Make the coffee cheap enough and foot traffic rises. The margin migrates to the croissants.

Where the heck are you going with this Allen?

Bear with me. I promise there’s a point.

I think this framing helps explain how the era of code abundance might shift value to complements.

From an engineering team’s perspective—if anyone can generate the code, the valuable complement is knowing what to build and in what kind of system. If anyone can write content, the bottleneck shifts to taste. The new scarcities are human ones. Taste: who decides what’s good when everything is abundant? Judgment: who makes the call when options are infinite? Access to relationships that can’t be replicated. Meaning, context, narrative.

Vibe Coding

So what does this look like in practice? Enter vibe coding.

I’m shocked at how quickly these platforms are progressing. But I find myself intuitively quite bearish on durability. Most of what people whip up won’t survive contact with production.

That said, I have product teams and leaders using them as an excellent medium for prototypes—and some are pushing further. The SprintSuite team have started moving Lovable into production-grade, company-wide use. They’re rebuilding their entire CSM and reporting stack on it, and it seems to be working. I’m starting to hear about smart operators at insurance firms and consulting shops building their own reporting dashboards, diligence co-pilots and custom CRM tools too.

In the vibe coded software market, the complements that matter are security, maintainability, and scalability. Avenir has a nice diagram showing the trade-offs for software buyers. The interesting bit: total cost of ownership.

Slash the price of dashboards and workflow tools to near-zero, and you push the burden of system design, security, maintenance, and scaling onto whoever’s building. In low-stakes or small-scale environments? This might work great.

Where my mind goes: vibe coding will disrupt industries with high internal tech capability and fragmented tech stacks. You can imagine the internal IT departments creating their own automations and deploying their own applications because the ability to maintain and secure is there already.

Top right quadrant are where vibe code apps might actually be deployed at scale.

In an era of abundant code, the role for an entrepreneur moves fundamentally from how to build to what to build.

New capabilities, new systems

As I mentioned in Part 1, one of the core differences in this technology wave is the new capabilities emerging from LLMs and free-flowing code generation. At a fundamental level, software is now expected to do more of the multi-step workflows we used to do with human input. Yes, at the limits, I believe we’ll completely replace some human labor categories. For now, software companies are racing to build systems that do more.

The art is in feature combination. Features create capabilities. Capabilities build systems. Systems produce outcomes. The question becomes: what features and inputs do we need to collapse more of the work?

So what matters now?

Below are the obvious areas of opportunity, according to some very smart founders I work with.

Aggregation of data. Like many I felt the need to export my life into Markdown over Christmas to make the most of Claude code. I can truly feel the power of aggregated customer data now.

I like Matt Brown’s frame of building a context graph as well. He argues that vertical software platforms are advantaged because they typically have the most relevant set of connected data that can trace through and ‘infer’ the why behind a particular decision.

Take platforms like Splose, Fergus, or SprintSuite. They capture industry-specific data models - customer engagement, scheduling, procurement, billing, field ops. All those connected modules create a dataset that traces end-to-end how customers actually run a service business. That’s the context graph in action.

What is your walled garden? I keep coming back to this question with portfolio companies. EVP’s best product thinkers (Rhys Davis at Investorhub and Henry Innis at Mutinex) are pushing on something they call Day Zero value. How do we create a product where customer 1,000 gets materially more value than customer 1 coming in. For them it ultimately comes back to assembly of data they’ve amassed in their walled garden from their customers already and exposing it to new customers in a way which is not possible for an upstart. Look to compound your walled garden now.

Network effects. Network effects businesses (like marketplaces with unique supply) continue to have strong moats. Software can’t shortcut all the time it takes to curate supply on a marketplace as an example. Think about network effects in your roadmap.

Risky business. On security and maintainability: there’s a trust gap for certain critical software workflows. I can’t imagine a generalist internal team fully replacing their cybersecurity stack, their POS and transactional systems, or their core store of customer records. These companies need to ask themselves: do we really want to be in the business of ensuring compliance, maintaining these systems, and incurring ongoing opex ourselves?

That said, there’s probably going to be downward pressure on pricing as we figure out the new balancing act between buy vs build.

The question for every software company now: what’s your salt? What are you going to focus on?

Subscribe now

What do Cod fish and Code have in common? There’s a case study on cod that got me thinking about software technology cycles over the break. Capital Returns covers the development of this delicious fish, it goes something like this...

In the early 16th century, cod was a prized commodity. High in protein and pleasant in texture, it commanded premium prices from consumers across Europe. The Basque fishermen who supplied markets had a geographic monopoly. Catching cod required sailing all the way to the Grand Banks off Newfoundland—a dangerous and capital-intensive journey that few could undertake. The skilled labour, time and capital requirements meant scarcity. Scarcity meant supernormal profits.

Then technology improved. Better ships. Better nets. Better salting and storage techniques. Each innovation increased yield. By the 18th century, what was once a perilous expedition had become an industrialised supply chain. Catches multiplied. Prices fell. The fishermen, the port operators and the processors—each layer of the supply chain saw its margins compress as cod shifted from scarce luxury to abundant commodity.

Where did the value go? To brands, to distributors, to retailers, and ultimately to consumers. The same protein that once graced aristocratic tables became affordable staple food for the masses.

This pattern repeats.

The Pattern in Technology Cycles

Similar dynamics have played out across major technology waves. This piece by Kai Wu first heightened my awareness of these evolutions.

Telecommunications. The internet revolution created enormous value—but not for the companies that laid the pipes. Telecom infrastructure (poles, wires, switches) attracted massive capital investment in the late 1990s. Competition and overbuilding commoditised bandwidth. The economic profits flowed through to the application layer: Google, Amazon and Facebook captured the value that AT&T and WorldCom competed away.

Hardware. PC manufacturers saw the same compression. In the 1980s, building a personal computer was a high-margin business. By the 2000s, memory chips, hard drives and processors had become commodity components. Dell’s “direct model” was really an arbitrage on the commoditisation happening beneath it. Today, PC margins are razor-thin. The value migrated to software (Microsoft, initially) and then to cloud services.

Surprisingly, previous parts of the supply chain can be re-enlivened, as we see in the current hardware investment cycle focused on GPUs, memory and all their constituent ingredients.

This has me thinking further down the supply chain—to applications.

Code Is Commoditising

My light CS education has me thinking about everything in terms of a stack. What I see today is an acceleration of a trend that has been moving gradually through the application software supply chain: code is commoditising. To extend the thinking further, simple applications fuelled by commodity code are themselves commoditising. This has been happening for a while.

Email is the canonical example. In 1995, email services were paid products. By 2004, Gmail launched with 1GB of free storage—a hundred times what competitors offered. Was this because excess profits had been competed away, or because Google kickstarted a loss-leader strategy to gain distribution? Probably both. The cost to deliver email had collapsed, and the dollar value of email decoupled entirely from the labour cost of its physical predecessor (postal mail).

Form-building software. Early on, building web forms required custom development or expensive enterprise tools. Wufoo entered the market in 2006 and was acquired by SurveyMonkey in 2011. Today, form builders are everywhere—often a table-stakes feature rather than a standalone product. Notion, Airtable and most no-code platforms bundle form creation to fuel their platforms. The result: forms are commoditised, often bundled and increasingly bespoke.

Video conferencing, design tools, CRM—there are countless other examples.

This is the pattern: technology breaks the link between production cost and historical labour cost. For a while, producers capture the surplus. Then competition compresses it. Like the cod industry’s evolution, the supply chain and ingredients required to deliver an end product have become more accessible. Value shifts downstream.

But there is one big difference. The same technology that is commoditising code is also creating entirely new capabilities. The potential for end-product differentiation is far greater than it ever was for substitutable cod.

In my next piece, I’ll share what I see as the second-order effects and some product strategy ideas emerging from this shift.

Subscribe now

Tools & resources you might like:

• Beeper - A chat aggregator that unifies messaging across platforms. I finally batch-respond to WhatsApp, LinkedIn, Messenger, and Slack from one app. It’s owned by Automattic (the Wordpress people), which runs a portfolio of consumer applications.

• Fiscal.ai / ROIC.ai - The Bloomberg terminals of the future. Complete access to financial data with cuts like fwd. valuation, hundreds of ratios and transcripts at a retail investor friendly cost.

• Claude Code - I’ll leave it at that to avoid the hype bandwagon. Very intuitive, even for non-technical use cases. Lenny has some good ideas on how to use it.

One Idea - Double Agents

This extract from Capital Returns offers a sharp take on sales & product pricing that transfers to software.

In a recent lecture given at the University of California, Charles T. Munger described a test he had performed at a number of US business schools. This test involved the Berkshire Hathaway vice-chairman asking the MBA students the following question: “You have studied supply and demand curves. You have learned that when you raise the price, ordinarily the volume you can sell goes down, and when you reduce the price, the volume you can sell goes up. Is that right? That’s what you’ve learned?” The business school students all nod in agreement.

Munger then goes on: “Now tell me several instances when, if you want the physical volume to go up, the correct answer is to increase the price?”

Some students come up with the luxury good paradox, whereby higher prices indicate superior quality which, in turn, leads to greater sales.

Very few students identify Munger’s answer, namely that when the customer is not involved directly in the purchasing decision, then higher prices can be used to bribe the purchaser’s agent and can result in both higher profit margins and sales volumes.

The author captures so neatly some of the effects which I often discuss in SaaS pricing and sales. When you have an agent selling your product, a higher price creates a positive loop and potential for supernormal profits for both the agent and vendor.

Straumann, the Swiss leader in dental implant technology, is used as an example of this effect. Dentists who adopt its implant technique, which replaces the traditional crown and bridge solution, earn higher revenues. Which is also why there are unique training and sales channels to help the dentists sell more of these products. Fun tidbit: Straumann even supports/acquires/promotes SaaS companies that help dentists find more opportunities to sell their products.

In the world of SaaS I’ve observed a couple of agent dynamics at play.

• Consultants and Channel Partners - A $100k/year tool earns a higher commission (and implementation fees) for your partners vs $10k. So long as the value proposition stacks up, or perhaps even if it doesn’t, a positive flywheel emerges.

• Buyers vs Users (Personal Incentives) - Championing expensive enterprise solutions can be a resume builder for IT directors. Higher-priced vendors often provide buyers with exclusive events, training certifications, executive briefings, and networking opportunities that become personally valuable. All of these diverge from the value prop of the product itself.

• Buyers vs Users (Trust and compliance) - Higher prices might signal "enterprise-grade" security to boards and auditors. Expensive tools are easier to justify when explaining why a breach didn’t happen.

So who are you really selling to? Who is a double agent in your market?

Worth Reading - mostly AI stuff

https://karpathy.bearblog.dev/year-in-review-2025/

https://simonwillison.net/2025/Dec/31

https://blog.sparklinecapital.com/wp-content/uploads/2025/10/sparkline-ai-capex.pdf

• Large US tech firms are on track to spend $400bn this year alone (Meta, Google, Amazon and Microsoft)

• Bain estimates that $2 trillion in annual revenues are needed by 2030 to justify these costs

  • Current ChatGPT revenue is at $20bn so 100x in terms of monetisation

  • There are issues with these estimates being so far ahead and fuzzy.

    • We have other driving forces resulting in higher consumption which is the Jevon’s paradox

    • But the compression in cost and potential efficiency gains in algorithms will push these down

• Capital Cycles in the past have more frequently shown boom and bust

  • US Railroad stocks

  • US Telecom stocks

  • It’s very difficult to estimate DEMAND in time as an investor

    • Steer clear of capex boom companies trading on continuation of forward multiple expectations

• AI Prisoner’s Dilemma

  • If any new player goes All-In the others must follow suit and capital wastage is collateral damage

• Infrastructure captured very little of value that was created. The value accrued to applications and consumers.

  • In the dot com bubble telecoms laid 80 million miles of fiber optic cable. 85% of these cables remained unused and the cost of bandwidth fell to 90%.

  • Eventually the dark fiber ended up being consumed - but it wasn’t until the post 2010 period

Succinct Communication

Shiller PE Ratio

The Shiller PE ratio takes the PE by averaging the earnings for the S&P over the last 5 years (inflation adjusted) and comparing it to the price of the current market. It’s reaching levels close to the 2000s. When companies are trading high on PE there is a lot of forward growth expectation built into the multiples.

In search of long-term thinking

Writings

In Search of Long Term Thinking

My mum pinged me recently about forestry — my “future inheritance.” Eight acres by a rush filled riverine inlet in rural Ireland, to be split with my two sisters. Oak and larch. The commercial timeline for a Larch harvest? 60 years from planting. Oak? 120 years. One useful harvest in a lifetime, a second across three generations…

Read more

8 months ago · 19 likes · Finn

Cheeky Pint with Dan Sundheim D1 Capital

Dan makes two stock recommendations to look into.

• Clean Harbors Inc Thesis

  • The company specialises in hazardous waste disposal and industrial services for commercial clients. The essence of his investment thesis covers:

    • Asset base that is difficult to replicate - incinerators are heavy industrial assets which are challenging to establish in the market given environmental and social factors

    • Established supply chain is complex with process powers.

    • Growth in demand that will tie to American re-shoring of manufacturing should drive growth

• Siemens Energy Thesis

  • Producer of gas turbines that are used to power data centers. Backup power.

  • This week I have seen 4-5 mentions of power constraints on data centre roll-out

High Alpha released their 2024 saas benchmarks report recently, if you’re familiar with OpenView this looks to be the continuation of reporting that used to be led by them! If you’re receiving this email you probably know that I’m a nerd for metrics and wanted to share some of my initial thoughts.

Sample overview - check the biases

• 800+ companies across US (62%), Europe (20%) and other countries

• Most of the respondents have ARR between $1 - 20m - great for benchmarking Seed-Series A

• The thing that’s unclear is the skew in ‘quality’ of the respondents. For example, Iconiq’s benchmarking report is skewed towards venture funded companies and represents averages within what is already a high performing group. I think this report probably skews just slightly to venture funded startups (given the names involved) but is much closer to a SaaS market proxy.

OK so what’s interesting in growth rates?

The story is basically one of bifurcation in the market.. median growth rates are continuing to trend down year on year. Every year SaaS gets more competitive and if you overlay the buyer sentiment, macro backdrop (which is turning) in the market there are certainly forces making it harder to sell and thereby harder to grow.

But top-quartile growth rates are stronger than before?

Seems like the average SaaS company is being impacted by factors above but the high performers are benefitting more recently from the shift in the market. Looks like the tailwinds are accruing to the AI-led and vertical market companies… as validated in the chart below.

I will also note that the median growth rate of 50% for $1-5m ARR range is lower than I would’ve thought but speaks to how the normal SaaS (non VC) company probably sees the world!

This was the most interesting table in the deck…

The data shows a sweet spot in ACVs at $25 - $50k (which is a pretty wide range). It seems like this might be a hint at the balance between value maximization and deal velocity where you are going lower-market enterprise or mid-market customers and selling them larger deals.

You’re past the point of innovation budget zone so expect adoption and retention to be better and there are a lot of accounts in the market that can pay that kind of sticker price.

The data is probably skewed earlier stage in the sub $100k zones but given the sample size it does seem to show that there is a drop-off point in the selling capability after a certain ACV.

I was a bit surprised to see such strong GRR at the early stages.

My intuition would be that revenue churn might be higher for early stage businesses due to the product still being developed. The counter to that would be you probably found your best ICP customers in the early days and your service levels are particularly high.

A side note, I am believer that low NRR can be massive red herring metric for early stage companies. If you don’t have the right pricing and packaging to create expansion revenue, you won’t even have a shot at NRR.

GRR also improved this year, suggesting that the 2023 effect of budget cuts, churn and rightsizing of SaaS is past us and spend is back! We should be looking forward to the coming sales environment.

In office teams growing faster than remote.. looks like the debate is settled.

And finally.. some ridiculous sleep analysis.

Read the full report

Subscribe now

Why do you care?

As a SaaS founder immersed in navigating your company through the early growth stages, everything seems to be progressing smoothly. Your team is diligently addressing challenges and pursuing deals, often disregarding the need for polished spreadsheets, flawless systems, or comprehensive playbooks. The focus is simply on growth. However, once you secure your Seed or Series A funding, your investors will inevitably mandate, "You need to prepare a budget (or a financial plan, or a strategy)!" The question arises: Why do I care?

This guidebook is the first part of a series on planning from the vantage point of board members and active investors. They are tips learned from guiding several SaaS companies through their formal planning cycles.

Reasons to care

Team alignment. The capital planning process, which includes budget preparation and strategy formation, serves to create alignment with your team and help leaders in your business understand its key drivers. Documenting strategic decisions and translating the laundry list of objectives into its fundamental elements can help your team agree and focus on a set of goals. This extends to your board and helps align your priorities with theirs.

Framework for quantifying decisions. Business actions also need to be measurable and quantifiable. We believe that every great founder needs to deeply understand their P&L in order to run the business. Budget preparation forces you to think in terms of ROI on your actions. 

Getting deliberate on capital allocation. While you're focused on hitting next week and next month's performance targets, you want to make sure you're allocating a portion of your spend, headcount and attention to investments on second and third horizon opportunities to set the business up for long term success.

Dress rehearsal for maturity. Consider it a dress rehearsal for maturity. As you shift from the Seed/Series A phase to a growth stage, it's essential to cultivate institutional rigour and capability within your team. Experience has shown that by the time you reach Series B, investment partners are keen to see evidence that your team can not only devise a sound strategy but also execute it with reliability.

What is a capital plan

When we talk about capital plan in this guide we are referring generally to the more specific and tactical components of strategy as well as the preparation of a budget.

Strategy

The capital plan typically starts with reference to a medium term business strategy which seeks to map out where the business is oriented over the next three years. The strategy can be formulated with the input of the founders, broader executive team and investors / advisors who aim to find the right balance between business expectations and reality.

The conversations that feed into the fabled strategy may happen formally at a strategy offsite or informally but generally takes the form of an agreed framework and set of objectives to strive towards over three years.

The point is to tease out a number of issues that inform the initial ‘envelope’ of inputs that create a budget. 

• What are your expectations around growth?

• What are your other key stakeholder’s expectations around growth?

• What is the market telling us about things like financing and buyer behaviour?

• What are your objectives?

• How much are you willing to spend?

The answers to these questions form the first set of inputs for your financial model. Below is an example hierarchy of goals that are informed by these initial conversations and paints the picture of a typical company that is seeking to maximise market ARR growth and market expansion while managing burn to a certain level. Note that this is an example hierarchy for a single year short-term plan.

The inputs almost always have to reference a burn and growth target. The value is in starting the conversations around what’s more important with respect to the current stage of business. Is revenue growth or customer growth more important? Is revenue growth or preserving productivity and cash burn more important? ARR in existing markets vs new markets?

Note this hierarchy is focussed more on financial objectives as they contain the outcomes that are assessed by investors and relevant at a ‘board level.’ The objectives you communicate to your team when trying to deliver a revenue outcome will sit a level or two below with leading indicators such as pipeline, registrations or engagement. See example of operational framework here:  Amplitude North Star Framework

Budget

The budget is a model that brings together the objectives of the company and attempts to map out the required resource allocation for each function and productivity required to achieve them.

The goal of the budget is to confirm where the business is going to spend its money, what growth it’s aiming for and the implications on its cash balance. We see different approaches to budgeting modelling that depend heavily on the stage and sophistication of the business. This is a longer topic we will cover separately, for now we’ve shared some helpful resources on budget modelling below. 

I personally like budgets to break down the atomic assumptions which drive the revenue forecast (such as sales people, number of leads and conversion rates). Even if sometimes the data is limited and there’s variability, an approach which helps you stress test the level of required activity at the base level is preferred.

The budget can crystallise opportunity costs and ensure dollars are spent on the areas with a best chance of ROI. For example, what is the real impact of spending more on R&D vs on sales? Can we translate R&D investment into a more attractive ROI than additional sales team members? 

“Thinking about the budget in terms of ROI and payback on investment was some great feedback we got from EVP. I’d urge founders to add this to their thinking. Consider your spend based on when and how a dollar can be created or what end metric can be moved, be it conversion, lead generation or onboarding.” - Henry Innis

Here are a some good resources on budget model preparation:

• OnlyCFO SaaS modelling templates

• MostlyMetrics 5 part guide to Annual Planning (includes templates)

Piecing it together

As you navigate the journey from Series A to B, the complexity of your business will naturally increase. It's at this juncture that you'll likely recruit a financial expert—be it a Head of Finance, CFO, or similarly experienced individual—to help manage this complexity. Their primary role will be to act as a conduit between you, the founder, and the management team, meticulously capturing the operational details and translating them into a comprehensive budget.

The management team works with finance to produce their laundry list of requirements that they believe will enable them to hit their targets. These might include:

• Number of new heads needed to hit targets

• New tools required to be more productive

• In the case of marketing, likely spend required to generate leads

• Other large buckets of cost such as contractors

VP of sales might say: “to hit $4m of new ARR this year I think we need to double our AE count to 10 and increase our lead volumes by 1.5x.”

Here is where the finance team earns its stripes by grounding the asks in reality. Is that a productive use of resources? What other costs are incurred to support 10 AEs? What is the quota per rep and how does that track to what we did last year? Do we need to hire sales operations now? What else do we need to increase our lead volumes?

I view the finance function as a balancing force, one that rigorously challenges each department's proposals, aligning them more closely with practical expectations. A prudent finance professional might adopt a conservative stance—playing the 'devil's advocate' to a founder's optimism—to better manage the risks of underachievement.

If your org doesn’t have this muscle yet internally, which is common at this stage, we’d suggest you lean on your investors to play that role and provide oversight for a short while.

Investor perspectives + cheat sheet

At the end of and during this whole process you’ll get feedback from your board about how the plan is looking and feeling. Understanding how your investors are thinking about it will give you the foresight to prepare and consider elements of your budget in the right way.

Subscribe now