Expert System — Update 30 June 2016

Expert System has developed proprietary and patented semantic software to extract meaning from text. It can extract entities, relationships, sentiment and categorise the subject matter. This enables it to address the significant commercial opportunities that are presently available for solutions that can draw insights from the 80% of data that are thought to be unstructured. It listed on AIM Italia in February 2014 and, in September 2015, completed the acquisition of French competitor TEMIS for a total consideration of €12m. The acquisition increased Expert System’s total workforce to c 220 people worldwide and expanded its overseas sales from 17% to today’s 58% with offices in the US, Canada, UK, Spain, Italy, France and Germany. From these offices it serves a large number of blue-chip clients including Bloomberg BNA, Chevron, Eli Lilly, the International Monetary Fund, Raytheon, Shell, USA Today and Vodafone.

2015 was a year of significant investment for Expert System as it completed the purchase of TEMIS, opened additional offices in the US and increased its R&D spending. As a result, the company moved from a post-tax profit of €89k in 2014 to a loss of €3.3m, although it remained profitable at the adjusted EBITDA level, which fell from €2.3m to €1.2m. We have adjusted our forecast to reflect the incremental near-term investment and allowed additional time for current initiatives to begin to bear fruit. As a result, we have reduced our 2016 revenue forecasts from €28.8m to €27.9m and reduced our 2016 forecast adjusted EBITDA from €4.3m to €2.5m. We have also introduced 2017 forecasts for the first time and expect adjusted EBITDA to double to €4.9m in 2017 as the company begins to benefit from the operational leverage resulting from its larger size.

Expert System operates in a high growth industry and its geographic expansion serves to further strengthen its growth potential. It presently trades on a 2016e EV/EBITDA of 23.8x falling to 11.9x in 2017e. This places it at a premium to linguistic services peers, but towards the middle of the valuation range of more highly rated data analytic technology peers. Our DCF valuation suggests a value of €2.78 per share. These valuations also fail to reflect the value in Expert System’s stake in the unconsolidated ADmantX business, which was valued at €10m in a 2015 funding round. Other private funding rounds in the enterprise search space also suggest that Expert System could be worth more to a strategic buyer looking to leverage its independently recognised technology capabilities.

Competition: Expert System has a number of patents relating to its technology. However, there are a number of large companies offering solutions based on a variety of approaches. These rivals may produce solutions that are more effective than Expert System’s. Business model transition: Expert System is transitioning from a perpetual licence model (with somewhat lumpy revenues) to a more SaaS subscription-based model. It may also have to adapt its business model to suit different geographic markets going forward. The level of profitability may be higher or lower than the company’s legacy offerings. R&D investment risk: Expert System must invest in R&D to stay at the forefront of technology in its industry. There can be no guarantee that this investment will yield a satisfactory return. Integration risk: Expert System’s acquisition of TEMIS is a material one and introduces typical post-deal integration risks including higher than expected restructuring costs, failure to realise the anticipated synergies and potential business culture frictions. Exchange rate risk: Expert System is an increasingly international business with costs and revenues in a number of currencies. As a result, the company is more exposed to fluctuating exchange rates. The company mitigates this risk by diversifying its cost base with offices based locally.

Expert System has developed proprietary patented cognitive technology, which can extract meaning from unstructured, text-based natural language. This technology is at the heart of its Cogito software solution, which is designed to “read and understand the way people do” by placing words in context. The technology is based on artificial intelligence algorithms, which can distinguish between the different components of a sentence (ie nouns, verbs, object and subject). The technology has a number of practical solutions, as true understanding of the text enables the efficient discovery and extraction of relevant components that can then be organised and tabled in a way which draws out insights that ultimately support data-backed decision making. Ultimately, for data to be analysed it must first be understood. Expert System’s technology allows vast quantities of unstructured text-based data to be understood at speed. The commercial possibilities are wide ranging as the technology can be applied to every industry in any instance that deals with data in text form.

Expert System was formed in 1989 and its three founders, Stefano Spaggiari (CEO), Marco Varone (CTO) and Paolo Lombardi (R&D director) remain at the business to this day. The company aimed to bring technology to the field of linguistics, a discipline that at the time still consisted of extremely manual approaches. Expert System was successful in licensing its early technology to Microsoft for use in its Microsoft Word software programme. It developed its early work into one of the first semantic analysis platforms, which became the basis for its patented Cogito technology. Over time, Expert System’s Cogito technology has been expanded to incorporate more languages (presently 14) and to address specific language nuances in multiple disciplines and industry segments. Having achieved success and profitability in its home market of Italy, the company expanded overseas to take advantage of a growing international need. In February 2014 Expert System raised over €12m in the largest AIM Italia IPO to that date. It has used the proceeds to further invest in R&D and to expand its international footprint, particularly in the US which, although highly competitive, is the world’s largest market for semantic search applications.

The amount of data being produced is growing exponentially, with unstructured data growing at an even faster rate. As a result, data sets have become too large and complex to be addressed by legacy data mining techniques. This has given rise to the specialism of knowledge management within the broader big data trend, with an explosion in the number of companies that seek to address the shortcomings of traditional approaches. Data analytics is just one strand of the knowledge management problem set. Other challenges include the capture, curation, search, sharing, storage, transfer, visualisation, querying updating and security of data. The inherent difficulty of natural language with its ambiguities and double meanings has historically created difficulties for those seeking to use computers to systematically search bodies of unstructured text. More recently, advances in cognitive computing have enabled products to be developed such as those offered by Expert System, which address these shortcomings. As a result, text analytics has become an increasingly large part of the big data market, as can be seen in Exhibit 1. Expert System’s cognitive computing-based software helps to address the problems of search and curation through its ability to understand and categorise large data sets. This makes it much easier for analysts and decision makers to then extract meaning, patterns and insight hidden within unstructured textual data.

Expert System’s core Cogito technology is based on patented IP in the field of semantic technology. Semantic technology relates to the ability for software to derive meaning from data, usually passages of text but also extending to speech and other forms of unstructured data. Semantic technology is important because information in natural language form constitutes a significant portion of the world’s unstructured data. Using linguistic analysis, semantic technology takes into account a language in its entirety – meanings, verb tense or mode, gender, etc and the contextual relationships between words, to mirror the way people process information.

Statistical or stochastic search approaches differ from the semantic approach in that they do not attempt to understand the meaning of text. Instead, they rely on the occurrence and distribution of keywords and they form the basis for popular search methods such as those used by Google, where its PageRank approach algorithms have been extremely successful. The statistical approach has the advantage of being quicker in building the repository of information and analysing a new block of text; however, its results fall short on relevance as it struggles with ambiguity or when words have double meanings due to its inability to recognise context. We discuss the relative merits of Expert System’s semantic approach versus shallow text analytics and keyword-based technologies in our October 2014 initiation report.

Semantic enterprise search stands at the confluence of machine learning, natural language processing (NLP) and big data, all areas of the technology universe that are presently benefiting from increasing profile and rising investment. Technology and market research company Forrester estimates that there are over 200 vendors in the area of text mining and analytics alone. These range from niche companies with single function offerings, to companies with broader, more complete text analytics platforms. Forrester recognised Expert System as a ‘strong performer’ in its Q216 report, The Forrester Wave: Big Data Text Analytics, in which it ranks Expert System as one of only 10 companies, including IBM, HP Enterprises and SAS, which were recognised as leaders in the field. The primary selection criteria included a complete self-contained text mining and analytics platform; broad general-purpose text mining and analytics platform; and market presence as determined by revenues and interest from Forrester clients. This represents a significant third-party endorsement of the strength of Expert System’s product offering.

Expert System’s technology and product offering were also acknowledged by fellow global technology research company Gartner, which recognised it in the annual Gartner Magic Quadrant for Enterprise Search report. Both reports recognise the power of Expert System’s semantic offering, while also noting that it could benefit from further product extensions to complete its proposition.

Technological advances in the area of cognitive computing/NLP have led to increasing investment in both technologies. IBM has staked a significant proportion of its future on the area of cognitive computing through its Watson super computer with an investment of over $1bn to date. At the smaller end of the spectrum, a number of companies in the cognitive computing and NLP arena continue to receive increasing amounts of funding or are being bought out by larger players. Exhibit 2 shows a summary of some of the more noteworthy recent investments.

Expert System’s software can be applied to bodies of text across the spectrum of its 14 supported languages. However, the company has focused its efforts on a few particular industries to both demonstrate the power of the software in these industries and to maximise the value of industry-specific nuances, which it can develop into the technology and leverage across other industry players. Industry-specific knowledge can act as a competitive advantage, as Expert System’s development engineers can programme industry specific jargon, phrases and word associations into the software to increase the relevance of results even further. Industries that have been of specific focus for Expert System include banking, insurance and finance, life sciences and pharma, oil & gas, as well as media & publishing. Expert System is also increasingly developing applications to address the growing opportunity in the national defence and security sector, with ongoing initiatives with major US federal agencies, law enforcement organisations, US Department of Defense, US Department of Justice, FBI and the Defense Threat Reduction Agency.

Expert System has spent much of the last year carrying out a number of acquisitions and restructuring some of its divisions to best take advantage of the global opportunity before it.

The September 2015 acquisition of France-headquartered TEMIS transformed Expert System from a largely Italy-focused entity into a company with the ability to be closer to its end-users across the world. This will help strengthen its ability to service its many multinational clients. It also extended the number of languages covered by Expert System to 14, including English, Spanish, Chinese and Arabic. Today, overseas revenues are c 60% of the total (17% before the acquisition), as can be seen in Exhibit 4.

Europe: established Expert System Iberia (December 2014 acquired divisions of ISOCO, Spain). The company also announced the 100% acquisition of French text analytic solutions company TEMIS in June 2015 (completed in September 2015). TEMIS has offices in France (Paris and Grenoble) and Germany.

US: in October 2015 Expert System announced a reorganisation of its US operations designed to enable better focus on its two distinct target markets:

RoW: in December 2014 Expert System established CY4Gate to address the growing ‘cyber defence’ market, where national security agencies are increasingly looking for intelligent methods to address emerging geopolitical risks. The company has also entered into a number of new business initiatives in South America and the Middle East.

Expert System has strengthened its routes to market by signing a number of strategic partnerships. These include: Google for Work (where it was named a Premier Search Technology Partner in October 2015), Google Cloud Platform, Cloudera, MongoDB, Onix, MarkLogic and several technology integrations including Alfresco and Salesforce. These are the most popular cloud and knowledge management providers on the market and will make it easier for system integrators to include Expert System’s Cogito-based solutions as part of broader projects.

Expert System has also been working hard to further develop its relationships with all the major system integration players. This has included educating them on the ability of its Cogito technology to reconfigure data environments and enhance existing toolsets. This work will allow systems integrators and other technology providers to better help their customers to create new applications, adding cognitive capabilities to their information solutions and platforms, providing Expert System with an extremely valuable third-party sales channel.

Reduction of ADmantX stake

ADmantX is a subsidiary of Expert System, which is focused on providing semantic solutions to the advertising market. Its purpose is to provide advertisers with software tools (on a SaaS basis) to automatically determine whether a website is appropriate to display an advert for its products. Expert System reduced its stake in ADmantX in September 2015 by choosing not to take part in the capital raise it was conducting. As a result, its stake fell from 69% to 60%. The pre-money valuation of ADmantX valued Expert System’s stake at €10m. This represents a significant proportion of Expert System’s enterprise value, which we believe is underappreciated by the market. The company has stated that ADmantX is no longer a part of its long-term strategic objectives as the online ad market has specific business characteristics that require in-depth industry expertise and dedicated resources. As a result, while we expect that Cogito to continue to sit at the heart of the ADmantX product offering, we believe that Expert System may continue to reduce its exposure over time.

Expert System has an experienced and technology-focused management team, many of whom were the original founders. Stefano Spaggiari (CEO), Marco Varone (CTO and chairman) and Paolo Lombardi (R&D director) helped co-found Expert System and all have computer science degrees from the University of Modena. Stefano also holds an executive MBA from the Alma Graduate School of the University of Bologna. The management team also has experience in building start-ups; Marcello Pellacani (VP, Strategic Partnerships) and Andrea Melegari (SEVP, Defence, Intelligence and Security) founded web software developer ExpertWeb, which was sold to Italian internet services company Banzai. Luca Scagliarini (CMO) helped found and launch ADmantX in 2010 (within Expert System), where he previously served as CEO, and holds an engineering degree from the Politecnico di Milano, Italy, and an MBA from Santa Clara University, California. Claudio Palmolungo (EVP Europe) leads the European strategic business development. We believe this combination of skills and experience in the senior management team, along with the presence of a number of the original founders, is a strong positive.

No established peer group of publicly listed pure enterprise search companies exists. Many of Expert System’s closest competitors are either privately held or form divisions within much larger multi-disciplinary technology companies. We have produced a list of peers that incorporates language services businesses, global enterprise search peers and smaller Italian software peers.

Expert System is investing heavily in its future growth, and as a result we believe that its present EBITDA level understates the company’s future steady state earnings power. Taking 2017e EBITDA as our base, it trades on an EV/EBITDA of 11.9x. This places it at a premium to language service peer Nuance Communications. SLI Systems is trading on a recovery valuation, although Expert System trades at a premium to SLI on an EV/Sales basis. We believe that Expert rightly trades at a premium to language service companies due its higher growth potential and the larger technological component of its business, particularly when compared to SLI Systems. When compared to its enterprise data peers, Expert System’s valuation sits in the middle of the pack. We believe its valuation reflects the proprietary nature of Expert System’s core Cogito technology and the significant ongoing investment in R&D, which we believe will enable it to take advantage of the high growth nature of the end-markets it is seeking to address. Expert System’s valuation also fails to recognise the value in its non-consolidated stake in ADmantX, which was valued at €10m during ADmantx’s 2015 fund-raising round, a significant proportion of Expert System’s present EV and 40c of its present €1.84 share price.

We have also produced a DCF valuation analysis. Although we recognise that this is very sensitive to the inputs used, it can often better reflect the value in a business after it has successfully scaled up. Our DCF valuation is based on our best estimates (9% WACC, 25% long-term EBITDA margin, with revenue growing at a CAGR of 13% from 2015 to 2026 and capex/sales falling to 5% from 30% in 2015) and suggests a value per share of €2.78. A 1% decrease in the WACC would increase the valuation to €3.67, while a 1% increase would reduce the valuation to €2.15. Our long-term EBITDA margin is contingent on Expert System continuing to derive an increasing amount of its revenues from its SaaS-based standardised model.

2015 was a year of significant investment for Expert System as its management successfully internationalised the business to offer a broader platform from which to drive growth in the coming years. As a result, although the company continued to be profitable at the adjusted EBITDA level (€1.2m), it slipped to a net loss of €3.3m in 2015, from a profit of €89k in 2014. We have adjusted our 2016 forecasts to reflect the ongoing investments in marketing, people and product, and introduce 2017 forecasts for the first time. Details can be seen in Exhibit 6 below.

2015 was a year of significant investment for Expert System. Much of this relates to significant investment in capex, which was up 44% from €3.9m in 2014 to €5.6m in 2015, and the €3m upfront cash component of the €12m TEMIS acquisition. Operating cash flow was also negative in 2015 (-€2.2m), largely due to the investment in building its US footprint, including hiring staff and the opening of two new R&D facilities.

Expert System increased its level of funding in 2015 to ensure that it was able to make the necessary investments in its future growth. This consisted of a mixture of convertible bonds, which increased from zero in 2014 to €5m in 2015 resulting from its fully subscribed July issue. The bond is long term in nature (9.5 years) and was issued at a very attractive interest rate of 4%. The company also increased its level of bank debt from €7.0m to €16.3m during the period, with a further €7.3m owed to other lenders. The net impact of the negative operating cash flow and the significant R&D and capital investments during 2015 resulted in Expert System moving from a net debt position of €2.8m to €12.4m at the end of 2015. However, it should be noted that the company retains the liquidity to meet its short-term financial obligations, with €11.2m of cash on the balance sheet. Expert System has also signalled that it intends to shore up its balance sheet and access further growth capital via a €5m equity raise to be completed later in the year (but not yet included in our forecast).

We have slightly reduced our 2016 revenue forecast from €28.8m to €27.9m (48% y-o-y growth) to reflect a slightly longer sales cycle. Expert System has carried out a lot of work in recent months to introduce its upgraded product suite and APIs to system integrators. While they have received positive feedback, conversion to tangible sales activity is taking longer than we had previously accounted for. As a result, our adjusted 2016 EBITDA forecasts are down from €4.3m to €2.5m.

We have also introduced 2017 forecasts for the first time and expect revenues to grow by 17% versus our revised 2016 forecasts to €32.5m. We expect 2017 to be the year that much of Expert System’s recent investments really begin to bear fruit. We particularly note the potential for contract wins in the US government division. Just one significant win in this area would lead to a material upgrade to our forecasts. However, given the inherent difficulty in forecasting the timing of any deal, we have taken a conservative approach and chosen not to model any win ahead of its announcement. We believe that our 2017 forecasts are achievable without a significant government contract win; with the company beginning to benefit from the operational leverage effects of increased size and a higher proportion of licensing revenues, we see its adjusted EBITDA rising to €4.9m with the EBITDA margin rising from 8.8% in 2016 to 15.1% in 2017.