Are AI companies the next generation advisory firms?

In a recent article “The New Business of AI (and How It’s Different From Traditional Software)” by Andreessen Horowitz, the author argued that AI business won’t resemble traditional software companies. The main reasons for that are lower gross margins due to cloud cost and ongoing human support, scaling challenges due to the thorny issues of edge cases, and weaker defensive moats due to the commoditization of AI models and challenges with data network effects. The article raised the question of “Software + services = AI?”

The author’s examples seemed to focus more on general purpose AI applications, such as translation, image processing, and natural language generation. As AI companies expand their footprint into larger enterprises and aim to solve more industry domain specific problems, the three challenges highlighted for AI companies will become even larger.

Having worked with a number of AI companies for my clients and having worked in enterprise technology for over 20 years, I would hypothesize that there is an alternate theory on AI companies’ business models. Maybe a significant number of AI companies are just next generation advisory firms, similar to the McKinsey/BCG/Bain (the so-called MBB) and Accenture/PwC/Deloitte/KPMG/EY (the so-called Big 4). Are AI companies actually advisory companies, augmented by software?

Let’s take a step back and look at the business models of the advisory firms such as MBB and Big 4 first. The advisory firms are people businesses and make money by selling individual consultants and developers and they charge by hours. To differentiate the individual experts from the competitors, the advisory firms augment the individual consultants with proprietary tools, proprietary data, scale, and brand. Proprietary data has been a core part of the business model for advisory firms. For example, AON acquired McLagan in 2001 to gain access to wealth management benchmarking data. McKinsey acquired Finalta for its performance benchmark data and best-practice knowledge from 350 banks, insurers, and telcos in over 50 countries. Advisory and system integrators have also created proprietary technology tools to allow their clients faster access to insights and impact. For example, Bain partnered with leading price optimization and managed software provider Price f(x) to create the Bain Pricing Navigator.

However, the advisory firms must rely on the clients to generate the actual impact. There are three key components that clients own but consultants don’t:

• Clients’ own proprietary data — consultants must analyze clients’ own data to be able to make customized recommendations. This is why most consulting engagements started with the current state benchmarking and diagnosis.
• Clients’ own ability to implement the recommendations — in the end, clients must take the PowerPoint and implement them into real products and services and business processes. Consultants can advise on what to change and how to change, but the clients must take the real action.
• Client’s own capability to continue to evolve the recommendation — clients don’t want to make consultants their crutches and keep on paying consulting fees. Often clients want to get consultants to help them build internal capabilities so that clients can get on with the execution on their own.

Given those factors, we can see that there are a number of similarities of advisory firms and AI companies:

• Both need to staff people: AI companies often have to send in people to capture data, test and adjust the algorithms, and help clients implement the created solutions. Unlike software products where the products can be often plug and play, AI companies, similar to advisory firms, must understand how things work today at a client, capture the right data, analyze and make recommendations. People are very heavily involved, even with AI companies.
• Neither owns ALL the data: While both advisory firms and AI companies would claim that they bring proprietary data, in the end, it is the clients’ own data that will really drive the solution generation. AI companies must test and learn their own algorithms with the clients’ data and adjust the algorithms accordingly.
• Both rely on client’s ability to adopt the recommendation and implement the changes: Both advisory firms and AI companies must rely on client’s ability to implement real changes. An algorithm might be perfect on paper, but the algorithm must be incorporated into the client’s business processes and products/services.

To some degree, many advisory firms have already embarked on similar journeys to augment their advisory business models with AI capabilities. They have done this both with organic capability building and with inorganic acquisitions. For example, McKinsey acquired QuantumBlack to build its advanced analytics capabilities, BCG acquired Kernel Analytics, and Bain bought Pyxis. Accenture and Big 4 have done similar acquisitions.

Given this, I would argue that many AI companies are just the next generation advisory firms. Their business model is to provide the experts to clients, augmented with proprietary software, tools, and data.

If this is indeed true, AI companies need to aggressively consider a number of key factors as they continue to build up their businesses:

• Craft the right go-to-market strategy:

As one of my mentors used to say, consulting is a supply business, not a demand business. Clients always have problems that they need to solve. It’s just that often they don’t realize they have a problem. Thus it’s critical to have the right “rainmakers” from consulting firms to be able to ask the right questions and uncover the demand. This is very different from the product-driven go-to-market approach of a software company.

• Build up industry knowledge and pick the right battles

The big potential of AI is to embed AI into all parts of the business for clients. Most of the embedding requires in-depth industry knowledge. This would force many AI companies to have the industry experts on hand and also adapt the general algorithms into industry solutions. Major software product vendors had all realized this before and created industry specific solutions for their customers. AI companies may have to explore the same approach.

• Focus on how to make the changes happen and stick

This is where strategy consulting firms such as MBB really shine. They bring the top management perspectives and understand the right levers that can be pulled to drive changes across the enterprise. AI companies, once shifted beyond selling an algorithm, must possess the same level of capabilities to be able to generate real economic impact out of their AI technical solutions.

• Explore alternative revenue models

One of the biggest challenges for advisory firms is to create recurring revenue, beyond the 3–6 months engagements. As the authors pointed out in the Andreessen Horowitz article, many AI companies are starting to see the same problems. Alternative revenue models must be explored so that AI companies can enjoy the same level of revenue recurrence. For example, could the revenue model be a combination of one-time engagement fees plus recurring maintenance cost? Could the revenue model be a subscription model? Or could the revenue model be outcome based, as a percentage of actual business impact generated? As someone who is now working more as an internal consultant, I would say that clients would much prefer the outcome-based fee arrangements.

According to IDC, the potential corporate market for AI software, hardware, and services are going to be around $58B by 2021, compared with $12B in 2017. This is a vast market that everyone from management consultants to tech providers such as IBM and Google to AI companies are going after. The race is on and AI companies with their specialty expertise have a huge opportunity ahead of them!