Expert System — Five-year growth strategy unveiled

Expert System recently published its new five-year plan, covering the period FY20–24. The company laid out its new strategy, which puts its natural language understanding and processing capabilities at the heart of a new platform designed to encourage more widespread adoption of natural language-based technology within the enterprise.

The company is hosting a webinar on 3 July to provide more information on the plan.

The company has outlined an ambitious plan, targeting revenue of €100m by FY24 with an EBITDA margin of 22%. It has presented detailed forecasts on the progression to this target, based on the expected return from an increased level of spend on R&D (to develop the full end-to-end SaaS platform) and sales & marketing (to increase awareness and accelerate adoption).

Over the forecast period, the company expects to spend c €51m on R&D (which we assume is capitalised and amortised) and €117m on sales & marketing.

To date, Expert System has tended to sell its software to enterprise customers, typically early adopters and innovators. Projects are relatively large and have entailed high levels of customisation, resulting in a high contribution to revenues from professional services (36% in FY19). As these are significantly lower margin than software licenses (at best 30% gross margin compared to anywhere from 80% to 100% for software), this has held back the company’s profitability and brought with it the utilisation issues typically seen in professional services businesses.

Selling software to enterprises that requires this much implementation work naturally reduces the addressable market and leads to long sales cycles. As the application of natural language understanding and processing technology to improve the efficiency of business processes becomes more mainstream, the company wants to expand the use of its technology to smaller businesses.

To make it easier for smaller companies with less access to resources to adopt its technology, Expert System is adapting its technology to create a platform that can be accessed by anyone, without needing to be an expert in machine learning or linguistics. The chart below shows the market segmentation approach it is taking, with a freemium model due to be launched imminently.

Providing a freemium model gives potential customers a chance to trial the technology without having to go through a lengthy procurement and implementation process. Some users may stay permanently on the free version of the software, which will provide limited functionality. Others will want to unlock premium functionality by upgrading to a paid version. Expert System sees channel partners as being crucial to the adoption of the platform by users in the mid-two groupings (c 10% of FY19 revenues were generated via the channel). The company will reserve its professional services capability for enterprise customers at the large end of the scale; for these larger customers it will also focus on core verticals (Insurance, Financial Services, Info/Publishing/Media) to provide industry-specific solutions.

The company is also willing to license its NLU/NLP layer to OEMs.

As part of its growth strategy, Expert System has decided that its already market-leading technology can be enhanced by putting it at the heart of a platform that also makes use of other related technologies.

In the AI market, there are a number of companies specialising in providing AI tools and methodologies, such as DataRobot and H20.ai; however, these are generic and do not specialise in natural language understanding and processing. There are many open source libraries focused on the NLU/NLP space, for example, spaCy and NLTK (based on Python – a popular programming language for machine learning) and Stanford NLP and OpenNLP (based on Java). In addition, the large cloud providers support natural language processing via APIs1 (Google Cloud Natural Language, Amazon Comprehend and Microsoft LUIS) and IBM offers its Watson AI platform.

Expert System believes it has the opportunity to provide a platform specialised in NLU/NLP that combines its Cogito technology (which has detailed knowledge graphs for 12 languages) with machine learning tools (proprietary or third party) and open source libraries to provide a comprehensive service that allows users to work with their preferred tools.

For some time now, the company has been shifting to providing its software on a subscription basis. This can be provided on-premise or in the cloud. Under the new strategy, all new licences will be sold on a subscription basis. There will still be a percentage of revenues generated from maintenance contracts on previously signed perpetual licences. Where possible, the company hopes to convert these customers to subscription licensing.

Expert System has outlined when it expects to hit key milestones in its new strategy.

FY20: The company does not expect to see revenue growth this year, as new business is being held back by COVID-19 restrictions. The company has already started adding headcount to its R&D team, which is based in Italy, and setting up the senior sales & marketing team in the US, based in Boston.

The release of the cloud API in July will be the first step in the product strategy, enabling the launch of the freemium version of the platform. The larger product release will be in October.

FY21–22: Next March, the company expects to launch the full SaaS platform. This will have integrations to other machine learning and AI tools, including open source. To build awareness of the platform, the company intends to develop a community of users, and will also work with the community to seek feedback on enhancements to the platform.

The company is targeting revenue growth in the US of more than 100% per annum – in FY19, €4m/12% of revenues were generated in the US, so this would imply revenues growing to more than €16m by FY22 (assuming limited or no growth in FY20). Over the period FY20–22, management forecasts that the business will require additional cash of €23m. The company is planning a capital raise of c €25m in the coming months.

FY23–24: If the plan succeeds, the company expects the platform to be the market leader in AI-based natural language problem solving by FY24. In FY23, the company expects to reach EBITDA profitability, and over FY23/24, generate free cash flow of €8m. By FY24, it expects US revenues to make up more than half of the total and for 40% of revenues to be generated through channel partners.

The ‘Rule of 40’ refers to a rule of thumb used when assessing SaaS companies. This says that if the total of revenue growth and profitability exceeds 40% in a given year, the company is performing well. The phrase was coined by VCs investing in SaaS start-ups to assess the balance between growth and profitability. In the early phases, growth could be well over 40% but the company could be loss-making, as funds are invested in growing the customer base as fast as possible. As revenue growth moderates, investment in sales & marketing and R&D can be slowed allowing for higher profitability (say 20% revenue growth and 20% margins), still satisfying the rule of 40.

We have revised our forecasts to take account of the new plan. This results in a reduction in our revenue and EBITDA forecasts for FY20 and FY21, as the company builds the foundations for its new product-led strategy. We introduce forecasts for FY22. The model factors in the cost base (pre-depreciation and amortisation) increasing by 10% in FY20, 25% in FY21 and 23% in FY22 as the company builds up its R&D team and sales & marketing function. We have modelled an increase in debt to fund the cash shortfall over the period, although we expect funds to be raised via a capital increase at some point later this year.

Since the plan was published, the company has announced that it has sold its 17% stake in Cy4Gate, as the company has just listed on AIM Italia. Expert System’s stake was worth €5.375m and was accounted for within long-term financial investments at a cost of €0.251m. We have treated the gain on disposal as an exceptional financing item.

Our estimates are essentially the company’s targets over the forecast period. We expect to monitor the following key indicators to assess whether the company is on track to meet its financial targets:

In Exhibit 5, we show Expert System’s financial and valuation metrics versus three groups of peers:

As the benefit from the next couple of years’ investment in R&D and sales & marketing does not fall within the two-year period for which peer group forecasts are available, the sales multiples for Expert System will be inflated. On EV/Sales multiples, the company is trading at a discount to the first two peer groups and a premium to Italian peers (which have more modest growth and margin potential). As evidence emerges that the company is tracking the revenue growth and margins targeted in the five-year plan, we would expect the valuation to move more in the direction of the SaaS/subscription software companies.

We have performed a discounted cash flow analysis based on the company’s plan to FY24, and for the following five years, trending revenue growth down to 4.5% by FY29, trending EBITDA margins up to 29% (which results in an EBIT margin of 20%) and reducing capex/sales to 9% by FY29 (compared to 21% in FY20 and 14% in FY24). We have included the FY20–23 share grant and option grant plans, which add 5m additional shares to the existing 41.1m outstanding shares.

Using a WACC of 9% and a long-term growth rate of 3%, we arrive at a per share value of €5.43. We note that this valuation assumes perfect execution of the plan, and we would expect the share price to move towards this value as progress according to the plan becomes evident. This valuation would equate to an EV/Sales multiple of 2.5x and an EV/EBITDA multiple of 11.2x in FY24.

A 1pp increase/decrease in the WACC results in a per share value of €4.33/€6.99.