Today, as the AI wave swept the world, an AI sales company called Clay has successfully stood out in the traditional SaaS market with its unique “predictive selling” model. In just a few years, Clay’s revenue has increased sixfold, and its valuation has soared to $1.5 billion. How do they do it? This article will take you on an in-depth look at the secrets of Clay’s success and explore the technological innovation and business model behind it.
At the 3rd Sequoia Capital AI Summit not long ago, there was an impressive sentence:
The accumulation rate of AI results will determine the upper limit of your company’s value growth.
This is particularly evident in the field of AI sales. In the past two years, by helping companies collect sales leads, a number of AI sales companies have risen rapidly, the most representative of which is Clay.
To put it simply, what Clay does is to integrate the data of 100 data vendors, such as Hubspot, LinkedIn, Maps, and CRM, combined with the Agent’s research capabilities, to crawl, judge and compare, summarize and process the information on the web page, and complete the basic information retrieval work similar to that of SDR (Sales Development Representative) employees.
At the same time, they have also further strengthened the platform’s capabilities through AI, such as building highly targeted lead lists for users, and automatically generating personalized emails, blog posts, etc.
Different from traditional SaaS products, Clay chooses a more flexible “points payment system” in terms of pricing model. This has also greatly increased the willingness of enterprises to pay and accelerated the growth of the company’s business.
After achieving 10x growth in 2023, Clay’s revenue will once again grow by more than 6 times in 2024. With the rapid growth of revenue, the company’s valuation has also risen.
In January this year, Clay completed a $40 million Series B financing, valuing the company at $1.25 billion. In May, in a new round of old stock transfers, Sequoia’s bid for the company increased to $1.5 billion.
Let’s take a look at this “10,000 people” AI sales company.
01 Multiple data integration capabilities to predict high-value leads
The strongest sales are not eloquent or year-round, but good at fakes.
In the past, marketing relied on employees to manually sort out the data of various channels, which not only slowed down the response, but also the marketing methods on the market were similar. In order to compete for limited customer resources, enterprises began to fight “price wars” and join sales input.
However, the CRM system commonly used by enterprises is complex to operate, and the field design is disconnected from actual business needs, which is both troublesome and impractical.
After 10 years of interaction design, why did I transfer to product manager?
After the real job transfer, I found that many jobs were still beyond my imagination. The work of a product manager is indeed more complicated. Theoretically, the work of a product manager includes all aspects of the product, from market research, user research, data analysis…
View details >
The emergence of AI has brought a turnaround to the sales industry. In the entire sales workflow, pre-sales GTM (product commercialization) and customer service are the most suitable for AI to play a role.
The much-needed automation features in the pre-sales phase focus on lead generation and outbound calls.
After discovering these pain points, an AI sales company called Clay took action. They have increased customer response efficiency by 2-3 times, 30% of enterprises use their AI to handle 500,000 tasks a day, and some agencies with annual revenue of millions of dollars.
The key to this 40-person team, which has only been established for 3 years, is its strong multi-source heterogeneous data integration capabilities.
Clay has developed the AI research agent “Claygent”, which can be seen as an AI+SDR Agent. It allows users to create customized data sources and rich workflows tailored to their specific needs, allowing businesses to find lead information on the web.
It builds a “three-step” workflow around the data intelligence and automation engine: retrieving and retrieving data, verifying and giving sources, and outputting search results in a specified format.
The first step is to retrieve and obtain the data.
“Claygent” integrates information from 75 data providers, including job sites, public databases, and news media, such as Crunchbase, a customer management system, etc., to aggregate information into a unified platform for users to access at one time.
“Claygent” doesn’t even let go of the customer’s social feed. On platforms like LinkedIn, HubSpot, and Clearbit, Claygent automatically detects potential customers when they have job changes, financing changes, job postings, etc., and categorizes them by location, company size, experience, social network connections, etc., helping sales teams strategize accurately.
In terms of data integration, the core of Clay’s technology is the “GPT-4+ binary search method”.
Before starting the search, Claygent does not directly crawl the entire website, but consults GPT-4 to determine which parts are most likely to contain the required information, and then uses a dichotomous search method to search.
The so-called dichotomous search method is to select one part of the website to check first, and if the target data is not found, move on to another part. This approach narrows down the search by gradually narrowing down the search to pinpoint the desired information.
This search method not only accurately collects the information users need, but also significantly improves the efficiency of scraping.
The second step is to verify and give the source.
“Claygent” also combines cross-validation with multiple data sources, such as comparing Crunchbase and PitchBook data, to ensure the reliability of the output results. Users can also ask Claygent to give the source of their data.
In addition, each business can set its own qualification criteria for potential customers. For example, recently released AI products or features can be “+10 points”, recent sales team expansion can be “+15 points”, etc. Once everything is set up, “Claygent” tests each unique lead to determine how likely they are to make a purchase.
The third step is to output the search results in the specified format.
“Claygent” can output search results in the specified format. For example, text, numbers, URLs, and other custom formats.
In the “active contact” link after “discovering leads”, Clay also provides a simple automated outreach function. For example, email content generation, but Clay can’t automatically send emails like Outreach and Apollo.
However, Clay can integrate dedicated sales engagement tools like reply to enforce follow-up rules and generate personalized emails/SMS across platforms. With Reply, Clay users can create sequences for leads that leverage email, phone, LinkedIn, and WhatsApp for outreach.
Founder Kareem Amin was a Microsoft engineer and focused on lowering the barrier to programming with Nicolae Rusan in the early days. In 2021, as customer needs shifted from functional satisfaction to value co-creation, Clay strategically shifted to sales automation to help enterprises become data-driven
02 Word-of-mouth communication drives growth, and points are paid – usage is value
The key to Clay’s early success lies in pricing strategies and content marketing.
In the sales industry, when “AI solutions” gradually become standard, homogeneous product demonstrations, similar subscription packages and overwhelming list of functions are also accelerating customer choice fatigue.
Clay chose to use “anti-routine” promotion, which does not force credit card binding, and does not have a complex paywall, and directly opens the product to a full-function trial.
This means that users can freely explore, trial and error, and personally experience the improvement of AI in sales data processing.
This very sincere style of play not only quickly caught the attention of customers, but also inadvertently ignited the engine of word-of-mouth communication. Clay’s community has over 11,000 people participating in discussions, and there are over 100 business cases built on top of Clay.
Prior to this, Clay had also tried a variety of growth methods, such as email marketing, paid advertising, SEO, and even billboards in San Francisco, with the main goal of attracting enterprise customers who “already have a demand but don’t know which tool to choose”, but the effect of “word-of-mouth communication” is still the most obvious.
Now, Clay has established a set of feedback loops for content production:
Clay offline events are held around the world, and the content produced from the event is organized into LinkedIn posts, then integrated into blog posts, and finally compiled into guides. In this way, business decision-makers can also intuitively see how their peers use AI to process massive amounts of data.
The reason why customers are willing to share is because of the product characteristics of Clay. These clients are mostly agents who want to establish a professional presence through LinkedIn. Of course, the Clay team also actively assists customers in creating content, such as introducing new features and collaborating on articles.
In the later stage, in terms of SEO optimization, Clay can also be regarded as understanding the user psychology. Clay’s target users are “data-sensitive businesses eager to improve sales efficiency.” Therefore, their SEO optimization focuses on “AI”, “automation”, and “customer data” elements, such as “sales automation tools”, “AI data analysis”, and “customer lead mining”.
In terms of pricing, Clay also cleverly did not adopt Figma’s seat-by-seat pricing model, but chose a method that is more in line with the interests of customers – the Credits system.
Clay’s core products are data querying, lead mining, and intelligent matching, and the value depends on usage, not the number of users, so ToB Saas is not applicable to seat-based pricing.
On Clay’s pricing page, customers can purchase a certain number of credits in advance, and Clay consumes these credits according to the customer’s actual data query, processing, matching, and other operations.
The traditional charging method is like having people with different meals pay the same buffet fee – businesses pay for features they don’t use. Clay is billed by the number of data operations and publishes a pricing transparency report annually. This method is like a mobile phone data package, pay as much as you use, and flexibly adapt to the scale of the enterprise.
03 Summary
In the long run, the story of AI sales has just begun.
The reason is simple, the workflow of the sales link is long, and the value of providing end-to-end products and services is large enough. At the same time, the results delivered by sales efforts often have a clear goal, and it is easier to measure value with a measurable service effect.
At the same time, the cost of AI sales is much lower than that of human labor. Alex Rampell, partner at a16z, once used Zendesk, an AI customer support software company, as an example:
If a company uses Zendesk to handle 2,000 tickets, 1,000 Zendesk agents can replace 1,000 agents. In this way, it costs $1.4 million a year to use AI salespeople, while 1,000 human salespeople cost $75 million.
Calculated, the cost of a ticket (sales lead) with a human salesperson (customer service) is $37.50, while the cost of an AI salesperson is only $0.69.
The results are measurable, coupled with the huge advantages of AI on the cost side, although AI sales products are in the early stages of development, they have shown good value, and their value will also extend from the simplification of the company’s current sales activities to the reconstruction of enterprise sales and workflows.
