If it is boring, repetitive and mundane; automate it. If it needs intelligence and skill then hire someone and get out of their way. Don’t hire someone to do the leg work. Hire someone who can own the task. Take a part of your role, break it off entirely and hand it over. Don’t micromanage their charter, it’s their charter now, not yours!

If it is tedious to run and manage the CRM don’t get a “CRM assistant” to do the leg work of managing the CRM. Automate it. If the CRM needs upkeep beyond that and requires intelligence hire someone who will manage the revenue process not the CRM.

If you are feeling naturally drawn to hiring people to “assist you” rethink if you are ready to be a manager. Your job as a manager is to be selfless and help other people be great at what they do. Their job is not to help you, your job is to help them!

If you need a gut check on if you are hiring an assistant or assigning them part of your role, ask yourself if you would be okay if they reported to someone else. If the answer is no you are hiring an assistant.

This was originally a post shared internally on the Nanonets slack. We help businesses automate manual processes. If you need help automating tasks try our product out at nanonets.com

Introduction

The landscape of business software has undergone a seismic shift over the last decade. Previously, the focus was on Systems of Record—think ERP (Enterprise Resource Planning), CRM (Customer Relationship Management), HRMS (Human Resource Management System), and more. Companies found their competitive moat in owning these monolithic systems that stored critical data. Fast forward to today, and the moat is increasingly built around workflows rather than systems of record.

Examples of Workflow-Centric Ecosystems:

• Sales Funnel: A lead moves from a marketing platform like Mailchimp to a CRM system like Salesforce, to a quoting tool like CPQ software, and finally to the billing system. It's not just the data that's important, it's the nuanced process connecting these systems.

• Customer Support: A customer's query could start in a chatbot, move to a support agent through Zendesk, be discussed in internal communications on Slack, and result in an engineering ticket in JIRA.

• Inventory Management: A product's lifecycle may start from a supplier managed in an ERP, to an inventory management system, to an e-commerce platform, and finally to a shipping partner.

• Recruitment: A candidate’s journey through an ATS (Applicant Tracking System), to interviews scheduled via an HRMS, and finally to onboarding managed through a dedicated onboarding portal.

Costs of switching

Let’s talk about the migration of these 2 types of software. There could be multiple reasons for migrations of SaaS tools, the main reasons for any migration are the following

1. The existing tool doesn’t solve the intended usecases

2. The existing tool has become too expensive

3. The new tool helps replace 2-3 of your existing tools

However when it comes to migration we have to account for a few things.

Replacing a Workflow Tool is Like Rewriting a Legacy Codebase

Think of your Workflow Tool like a massive library of code. Changing it is like rewriting a legacy codebase. Folks who have done this understand what an insane initiative this is in practice.

Workflow Tools deeply integrate into your processes. Teams customize them to fit their needs. Replacing them disrupts workflows, demands retraining, and causes downtime. It's like rewriting years of code.

Changing a CRM is Like Changing the Underlying Database or Library

Now, consider changing your CRM. It's similar to swapping the core database of an application. CRMs manage customer relationships but aren't as deeply tied to daily workflows as Workflow Tools. Changing CRMs involves data migration and system reconfiguration but causes less disruption. While certainly not easy.

The Data Migration Revolution: Thanks to Language Models

In the above analogies, worklow tools are like the code that glues various databases and libraries together, also known as Systems of Record. It is much easier to change and have drop in replacement for the latter.

Another thing to consider is with regard to migration is the introduction of LLMs.

Moving data between systems used to be a cumbersome task, involving transformation processes and often a change in schema. Today, with the aid of Large Language Models, data schema migration has become much easier. The models can understand the schema from one system, translate it, and help to map it to another system, drastically reducing the time and effort needed.

The Workflow Challenge

However, as straightforward as it is to move data, it's much more complex to move workflows. Here's why:

1. Customization: Every business tailors its workflow to specific needs. Custom-built integrations between systems are not easily replicated.

2. Human Element: Workflows often have manual approval steps, decision-making forks, or require specific timing. These nuances are often undocumented and deeply ingrained into a team’s process.

3. Dependencies: Workflows are often dependent on conditional triggers, time-based actions, and parallel processes that are specific to the software tools in use.

4. Error Management: A faulty workflow could result in bad data, duplicated effort, or missed opportunities. Maintaining the workflow is a constant task that involves debugging, iteration, and updates.

Workflows once implemented in an organization can potentially shape the everyday working of 1000s of people. These workflows eventually become habits in the org. The failure of switching comes down to not being able to change these habits very easily.

The New Moat: Owning the Workflow

The Critical Role of Workflow Orchestration

With this growing arsenal of tools, the real challenge becomes workflow orchestration—how to make these 200 different tools talk to each other, work in harmony, and, most importantly, automate actions across them

Owning this orchestration layer, therefore, becomes incredibly strategic. The companies that can best facilitate the flow of data, automate actions, and improve processes between these tools will own the next generation's competitive moat.

Coupled with the following additional benefits that workflows provide, owning it becomes even more important

1. Vendor Lock-in: Companies providing workflow automation tools, like Zapier or Integromat, are becoming increasingly indispensable. They glue together different systems of record, creating a new form of lock-in.

2. Workflow Analytics: Companies like Nanonets are building on this by not only facilitating workflow but also providing deep analytics into workflow efficiency. The analytical layer adds another level of indispensability.

3. Data Enrichment: Workflows often improve the data by adding context or by amalgamating data from multiple sources. The enriched data is often more valuable than the original data, making the workflow even more critical.

One-Click Switch but No Easy Workflow Transfer

Software vendors are making it incredibly easy to migrate data with "one-click" switches. A new tool comes along, promising a sleeker UI or a groundbreaking feature, and the temptation to switch is hard to resist. Migrating the data might be simple, but the intricate workflow that runs your business isn't so easily moved.

In conclusion, as the number of specialized software tools balloons, the workflows that connect them are becoming the critical backbone of the enterprise. While migrating data has become simpler, workflows are harder to replace, and their orchestration becomes more complex yet invaluable. This makes workflow management and orchestration the new competitive moat, dictating not just operational efficiency but also strategic advantage.

Let's take a closer look at some of the key differentiation in the metrics of these two types of companies at IPOs.

NRR

A notable difference is the net revenue retention (NRR) rate, which measures the ability of a company to retain and upsell existing customers. SMB companies have an average NRR of 105%, while Enterprise companies have an average NRR of 127%. This is likely due to the fact that Enterprise companies have a smaller customer base with larger contract values, which makes it easier to achieve higher NRR rates. SMBs focused SAAS companies tend to acquire customers more cheaply but also churn them as easily.

S&M Expenses

Sales and marketing expenses as a percentage of revenue is another key differentiator between SMB and Enterprise companies. SMB companies have an average spend of $55M, which represents 17.7% of their revenue, while Enterprise companies have an average spend of $73M, which represents 19.8% of their revenue. This indicates that Enterprise companies spend more on sales and marketing than SMB companies, likely due to the longer sales cycles and larger deal sizes associated with Enterprise sales.

Growth Rates

Finally, it's worth noting the growth rates of these companies. SMB companies have an average growth rate of 50%, while Enterprise companies have an average growth rate of 65%. This is likely due to the fact that Enterprise companies have larger deal sizes and longer sales cycles, which means that they can achieve higher growth rates with fewer deals. Although this might also be because of a few outliers (Snowflake)

Growth Rates vs NRR

We see an interesting pattern here. For Enterprise focused companies Growth rates are directly proportional to their NRR but the correlation is quite weak for SMB focused companies.

Revenue Multiples

Investors tend to like Enterprise and SMB companies equally with similar Revenue multiples (although Snowflake does skew the numbers towards enterprise)

So what is really different?

If we look at the normalised differences (subtract the mean divide by the standard deviation), we see that the largest delta is in NRR and everything else is a distant second. The sales and marketing expenses are the next in consideration. What is least effected by who you sell to? It’s the companies valuation, so if you sell to SMBs or Enterprise as a founder and investor you are going to make the same amount of money.

Thanks to Dishank for helping prepare this.

Mental Framework

Every task or product decision should be put into one of 4 buckets:

1. Not important.

Don't do it.

2. Not important. But needs to be done.

Do the bare minimum

3. Important

As good as the best. Copy the best.

4. Most important.

Innovate to make sure it’s better than the best.

Isnt this obvious?

It is mostly. However there is one notable exception. Average. If it's looking like what you are going to do is average you are better off doing the bare minimum.

If a task is between important and not important go either way do as good as best or bare minimum. Don't do average.

Apple as an example:

Apple is probably the most intense when using this framework. A great example is their history with headphones. For years Apple shipped below average headphones with their iPhones and iPods. Apple's headphones never competed with the likes of Bose. There was nothing new, nothing innovative about them. They needed to be there for Apple's core products to function and they were. They fell into the bucket of ‘2. Not important, but needs to be done.’ Apple even encouraged you to buy third party headphones.

Once Apple noticed a new innovative product that was gaining traction (the Bragi Dash on kickstarter, and possibly before this) they created a best in class product the Apple Air Pods. This suddenly took their headphone offering from below average to best in class. You could say they were copies of the best or they were better than the best but they certainly weren't average. This had become a category that was now important to Apple. Once the Air Pods took off Apple pushed it to the next level moving it to most important creating the Air Pods Pro leaving everyone in their dust.

Other examples of products from Apple in each category

1. Search - Don't care, won't do

2. iCloud/Safari- I'd argue it's the bare minimum they can get away with

3. Airpods Max - as good as the best

4. MacBook/iPad - better than the best

Even if you don't agree with the categorisation you'll probably still agree that Apple has below average products and best in class products but they don't create average products. That is why they have the market dominance they do.

Even Apple fails occasionally:

One example of Apple failing to follow this philosophy is in their maps product. Apple Maps really is a very “average” product which came nowhere close to competing with Google Maps. It's very existence necessitates doing more than the bare minimum which would have been to show Google Maps pre-installed. Word is that they are taking this shortcoming seriously and looking to launch a new redesigned Maps soon.

Using this framework daily

This was originally a slack post shared with the team at Nanonets. We use this on a daily basis when scoping tasks. We ask ourselves:

Do we need to do the bare minimum or best in class?

Depending on the the answer:

A) If bare minimum then the follow up question is do we even need to do it?

B) If best in class should we copy the best or be better than the best?

If you don't hold yourself to this standard you will mostly end up shipping average features that lead to average products.

Note: everything about your product need not be best in class. Only the the features that you matter to your customers and to you as a product/company, which also depends on and will change with the stage you are at.

I'm the CEO at Nanonets, we are building AI to help machines see the world - starting with documents.

I write about startups, products and technology.

Subscribe now