Description

Marchex has lost money for years with the stock performing poorly.  In 2020, things got worse as Covid affected their business.  However, they divested their declining Local Leads business in October leaving a call analytics software business that has growth prospects.  In particular, technological advances in natural language processing and artificial intelligence has led to an emerging "sales engagement" application that is higher value. Marchex has a competitive advantage in selling these new sales engagement applications to their existing call analytics customer base that is the largest in the industry. The heavy R&D spend and the high expenses from integrating strategic acquisitions related to development of these sales engagement applications continues, however.  The years of losses will continue for at least a couple more quarters, and adoption is not assured. Management that owns 15% of the company promises to sell if this three-year development  project doesn’t result in profitability in the next couple of years.  Recent private-market transactions for competitors in this emerging space were well more than twice the current Marchex valuation.  The company and the largest investor together tendered for and purchased 5 million shares each at $2.15 closing in early October.  Patient investors could make multiples if the R&D spend and acquisitions pay off, and perhaps do okay in a sale even if they do not.

Call analytics software was initially pretty rudimentary.  It monitored how long callers were on hold and how long they would wait before hanging up.  It could record how many calls came in at different hours of the day and whether the operation was short-staffed at certain hours resulting in excessive hold times and lost business.  In recent years, artificial intelligence has aided computers in recognizing and understanding human language resulting in devices like Siri and the Google Home speaker.  Call analytics software can now create transcripts of the calls for a tire shop.  Employees using inappropriate language or not following the sales script can be detected.  Large conversation data sets can be analyzed to show that employees that ask for and then refer to that caller by name result in a higher number of appointments being made.  With that insight, training can teach employees to ask for names and the software can make sure they do by monitoring future calls. With even further progress, artificial intelligence is now being used to detect intent and assess customer sentiment from the tones in the conversation.  Historically, call analytics analyzed calls to gain insights to help improve employees future call performance.  Real-time assistance to the employee is being tested now, however, with these features known in the industry as “Sales Engagement” tools.  Marchex Sales Rescue product detects using AI trained from prior data sets that a caller had intent to make a service appointment, for example, but when the call was transferred to the service department nobody picked up.  Sales Rescue notifies employees that this occurred so they can immediately call the prospective customer back and make the appointment.  Many more AI assisted sales engagement applications are in pilot-testing or development, both by Marchex and competitors.

Two industry reports are useful to investors.  The Forrester Wave Sales Engagement Q3 2020 report characterizes Marchex as a “Contender”.  It concludes “While the vendor brings deep automotive consultative expertise to the table — it has nine of the 15 OEMs and 21 of the 34 brands in the US as its customers — the platform lacks the breadth and depth of some table-stakes features present in other vendors. Marchex's reference customers view the vendor as a strategic partner that is in tune with their needs but would like to see the vendor enhance its sales training and coaching capabilities. Marchex is a natural fit for any automotive or automotive value chain business, companies in the home services industries, and those in other traditional industries”.  The report identifies competitor Outreach as a “Leader”.  It is a privately-held leader in Sales Engagement serving technology companies that recently completed a $50 million Series F round at a $1.3 billion valuation.  Marchex shareholders can hope for a fraction of Outreach’s valuation by winning the emerging higher value Sales Engagement business from their existing analytics customers in the niches they dominate.

The Opus Conversational Intelligence report is the 2nd useful industry review report.  It defines the opportunity noting “Businesses have stepped up investment in technologies that improve the conversations they carry on with both prospects and customers. They are successfully leveraging their investment in speech and text analytics, natural language processing and machine learning in order to extract meaning and recognize the intent of each individual”.  Marchex is categorized as one of the four leading companies, #1 overall, and received the best rating in all four categories evaluated.  The commentary notes Marchex’s  superior transcription accuracy (better than Watson), omnichannel capabilities after incorporating the texting application from the Sonar acquisition, and the largest customer base (over 1300 companies).  The report notes the $75 million investment from Morgan Stanley in December 2019 for Calminer, one of the four “leaders”identified.  Marchex investors relation lead, Trevor, confirms Calminer is “great”, but he notes they generally do not directly compete with Marchex.  Calminer primarily serves call centers who are less attractive customers as they mostly look to cut costs.  The second highest rated company, $50 million estimated revenue Invoca, raised $56 million in Dec 2019 at a $225 million valuation.  Invoca is serving U.S. Bank, Dish Network, and others using AI to analyze conversations and is close in size to the $46 million run rate of the remaining Marchex  conversational analytics business.

Marchex is much cheaper than their perhaps more successful comparables.  It’s also cheap compared to the money invested in their remaining conversational analytics business.  Three recent acquisitions, Sonar in Dec 2019 for $9.5 million, Call Analytics for $35.0 million in Nov 2018, and Telemetrics for $10.1 million also in Nov. 2018, total $54.6 million.  The R&D spend in 2018, 2019, and 2020 ran about $15 million, $20 million, and $20 million for a total of $55 million.  The heavy spending includes integration of the acquisitions, development of AI applications, and development of Marchex Stream.  Marchex Stream is consolidating all their data onto a central server using Snowflake to enable AI applications.  Tying the data together from multiple customer sites all over the country enables AI analysis and real-time implementation.  “The technology is powered by Marchex’s growing base of more than one billion minutes of analyzed consumer-to-business conversations and employs deep learning techniques including convolutional neural network developments pioneered by the Marchex MIND Lab team”.

Recent financial performance of the remaining conversational analytics business looks approximately like this:

2020:  $46 million run rate with $10 million ebitda loss. Down despite buying low revenue Sonar in Dec 2019 for $9.5 million (Even Q1 before coronavirus was down)

2019:  $52.3 million but this is up in part from Call Analytics and Telemetrics acquisitions in Nov 2018 for $35.0 million and $10.1 million (with $3 million more earnout)

2018:  $35.4 million

2017:  $29.4 million

Management attributes slower growth in 2019 and 2020 to the use of resources to integrate the acquisitions and to a reduction in customer call volumes due to Covid.  Marchex is paid for performing analytics on phone calls at a rate of roughly 5 cents a minute with roughly 70% gross margins.  Roll out of the newly developed AI Sales Edge applications were delayed as customers focused on just surviving the pandemic.  In particular, their senior living business was “just destroyed” and roll-outs of software to dentists offices were simply stopped as the dentists laid off the implementation staff.  Trevor (investor relations lead) notes that the next couple of quarterly reports will certainly be slow, especially Q4 2020 which is seasonally slow.  He does note that customers are finally now at least talking to them about rolling out Sales Edge pilots when Covid is over.  Contracts with larger customers are extending to two or three years instead of one with lower minimums.  Trevor believes the longer contracts reflect the increasing importance of conversational analytics to their customers.  Some Sales Edge contracts are becoming a flat monthly rate per location rather than tied to call volumes.

The history of the company has been mostly negative going public in a 2005 IPO at $6.50 a share and reaching $20 a share before beginning a long torturous slide.  The company has paid dividends, including special dividends of excess cash, and bought back shares multiple times, but still the returns have been poor.  The company founder, Russell Horowitz still serves as co-ceo along with long-time executive Arends.  Horowitz owns approximately 15% of the company but through class A super-voting shares that give him control.  Horowitz and Arends each have cash salaries of $300K which is fairly modest for their Seattle location.  They’re suffering along with shareholders as their options and incentives have turned out to be mostly worthless.  The largest shareholder, Jonathan Brolin, comes from a private equity background and runs Edenbrook Capital like a private equity fund investing in public companies.  He explains his patient investment philosophy in this presentation that explains that losses for extended times are acceptable to him in order to build long-term shareholder value.  He’s supportive of management’s efforts to remake the company as detailed in this letter.  Marchex is Edenbrook’s largest position after buying the additional 5 million shares in the tender offer at $2.15 and more shares at $2.50 from put option exercises. 

Horowitz and Arends have let losses drag on for some time to pursue the emerging conversational analytics business including the new sales engagement tools.  Trevor at investor relations acknowledges that the company's performance has not been good.  “We’re a $2 stock”, he notes, but he corrected me when I suggested if the company didn’t turn around the financials within three years, the company might be sold.  “I don’t think Horowitz will let it go on even that long”.   Marchex is an attractive acquisition target with the largest customer base and the largest recorded call data set in the industry.  Large data sets are required for effective AI development, and data is much better if drawn from the particular industry. Data sets of conversations from automotive dealerships is needed to develop AI applications for automotive dealerships. Marchex has a significant advantage in developing the new AI-based sales engagement tools for their existing customers. If Marchex fails, it likely wouldn’t be hard to find a buyer, and Horowitz’s 15% interest is worth far more than his $300K cash salary.  Success with their Sales Edge AI offering is definitely not assured, but investors are unlikely to lose much money from current prices even if it’s not successful.

Scuttlebut

A corporate development guy at a startup conversational analytics firm that mostly serves call centers provided some comments.  He noted that call center software had become stagnant for a few years.  The development of natural language processing changed that, however.  The software can now create transcripts and notes automatically for a $20 an hour call center employee that gets yelled at a lot.   AI is a "big leap" helping agents with an insurance claim as it comes in live. 

On the other hand, an executive at one of the leading competitors to Marchex was more skeptical of AI.  All our customers are asking for AI.  However, they don’t even know what it is; they just know they want it.  Actually creating an AI algorithm that gives insights that ultimately results in higher sales conversions is elusive.  He suggested we ask company’s touting AI for specific examples where it was successful.  Marchex has clients clamoring for AI also, but watch to see if the new Sales Edge applications actually improve financials for those clients.  Customers want to try new AI stuff, but there's a risk it doesn't help their bottom line. The thesis for this investment is that the new AI applications will result in higher value software. If it doesn't, Marchex's advantage in selling it to their existing customers won't be worth anything.

Finally, a Marchex senior employee spoke highly of Horowitz.  He noted it was a diverse group at the company and that the positive culture and the strategy the company is pursuing are the reasons he’s at Marchex.

In Summary,  Conversational analytics has arisen from call analytics due to technological advances in natural language processing and artificial intelligence.  Marchex has the largest customer base in call analytics and the largest data sets from which to derive AI insights.  They have a competitive advantage offering these higher value applications to their large existing customer base.  Marchex has spent a lot of R&D and acquisition money to pursue the opportunity, and the stock trades at a fraction of this spend and a fraction of the valuations of competitor's private-market financing transactions.  The decline of the now-divested Local Leads business coupled with all this investment has made this company’s financials disastrous for a long time.  Losses will continue for at least a while longer.  Either the investments begin to pay off with potentially excellent stock price reaction or the company will be sold within two or three years.

Catalyst

Divestment of Local Leads to create a pure-play conversational analytics software company

Emerging higher value sales engagement tools come to market and create growth

Potential sale of the company if growth doesn't arise