Please describe your proposed solution.
<u>1) Introduction and Problems</u>
Sales is a key part of any business, helping companies grow while building a strong customer base. Both B2B and B2C salespersons follow the same general selling process to connect with prospective clients and build a strong customer base. The sales pipeline and sales funnel are both vital ways to visualize two distinctly different functions within the selling process. The sales pipeline visualizes the buyer/s journey from a salesperson’s perspective, tracking open deals from start to close. Meanwhile, the sales funnel visualizes the buyer’s journey from the prospect’s perspective.
The selling process starts with <u>Lead generation</u>, by approaching potential buyers through different channels (e.g., email, website, social media, campaigns), to attract them to visit the business’s landing page (website) and encourage them to interact with it. With enough interest, these potential buyers will enter the sales process as new leads. <u>Lead qualification</u> assesses whether a potential buyer is a good fit for a business. Finally, the lead is approached by a professional sales agent who will assist the lead in making the purchasing decision via a <u>sales meeting & proposal</u> to <u>close the deal</u>. The following figure shows the reality of the sales funnel. According to Gleanster Research, Bridge Group study, and Forrester Research, only 25% of leads are legitimate and should advance to sales, that means that 75% of your leads aren't being sold the correct thing, or worse, aren't ready to buy when sales agents start pushing offers. Only 17% of leads convert into opportunities and 83% of leads are dead.
Let's quantify that. If you have 100 leads, only 25 of those are right for the product. Of those 25, approximately 21 ( 17% of 25) them will not buy. Which leaves you with a meager 4 of 100 leads. The lead conversion rate is 4%.
The problem to be solved is increasing the lead conversion rate or boosting sales by generating more lead conversion into sales with AI assistance. Aligning Marketing ( Lead generation & qualification) and Sales is the answer.
But somewhere in between, there is a process of understanding the valuable potential buyers (landing page visitors) for the business. Since the selling process has a high cost (both in time and money), it should focus on nurturing only the most engaged and fitted leads to improve and maintain a profitable “Return on Investment” (ROI). This process is called lead scoring.
By scoring leads, businesses can ensure that their marketing and sales efforts are being focused on the right leads. A lot of money can be wasted chasing leads that aren’t going to convert. Predictive lead scoring helps the sales team by providing them with qualified potential buyers (prospects) who are more likely to do business with them. For example, a sales team may choose to focus their efforts on leads with a high score, while leads with a low score may be contacted less frequently or not at all.
Within the lead scoring process, AI ( Machine Learning) can also provide more accurate results than traditional (manual) methods. This is because AI takes into account a more significant number of data points (website behavior, social media engagement, demographic information, and more) and can identify patterns that humans are likely to miss.
Existing Web 2 lead-capturing platforms such as Landingi or ConvertFlow offer convenience and accessibility, but they come with limitations:
Data Ownership and Privacy Concerns:
- Data collection and usage: Platforms often collect user data and may use it for targeted advertising or other purposes without explicit consent. This can raise concerns about privacy and transparency.
- Control and access to data: Users may have limited control over their data and may struggle to access or delete it upon request.
Limited Design Flexibility:
- Pre-built templates: Many platforms offer limited design flexibility, relying on pre-built templates that can feel restrictive and generic. Users may struggle to achieve a truly unique and brand-aligned landing page design.
Cost and Value:
- Subscription fees: While some platforms offer free plans with limited features, most require a monthly or annual subscription fee to unlock essential features and functionality.
- Hidden costs: Additional costs may arise for premium templates, add-ons, and custom features, potentially exceeding budget expectations.
Vendor Lock-in:
- Data portability: Exporting landing page data or switching to another platform can be difficult or impossible, leading to vendor lock-in and dependence on the platform provider.
- Limited control over future changes: Platforms may implement changes or updates that negatively impact existing landing pages, leaving users with little control over their design and functionality.
Security and Reliability:
- Centralized vulnerabilities: The centralized nature of these platforms makes them susceptible to security breaches and data leaks.
<u>2) Proposed solution</u>
We propose and develop an MVP of decentralized lead-capturing service that helps users (businesses, SMEs..) save time & boost sales by generating more lead conversions hence revenue with the same advertising spending. The proposed service can roughly cover two categories in the Sales pipeline: lead generation and lead qualification.
- Lead generation: Users can design or buy landing pages ( from individual designers) without requiring any technical knowledge. It allows users to design their landing pages with simple drag-and-drop tools and templates, without knowing any programming language. After the landing page is created, users can publish it with their domain name and run marketing campaigns on any social platforms ( Facebook, TikTok, Twitter..) to capture their potential buyers.
- Lead qualification: This is a marketing technique that helps decision-makers identify the more profitable potential buyers among the generated leads. As a result, the sales agent will not waste time randomly contacting all prospects and will only concentrate their efforts on more likely converted ones.
<u>a) Architecture</u>
The proposed web3 lead-capturing service is built on Cardano infrastructure, which means that it is not controlled by any single entity. This allows participants as designers to retain more control over their pre-designed landing page templates and earn a fairer share of the revenue. Leads captured data are stored on decentralized storage as IPFS so businesses such as SMEs and individual households have ownership and control over it.
Here's a general overview of the architecture:
1. User Interface:
- Users ( a single person who belongs to a business or an individual) interact with the platform through a web browser.
- They can explore landing page templates, create & edit landing pages, and purchase them.
2. Decentralized:
- The platform uses blockchain (Cardano & decentralized storage as IPFS) to store data about templates & captured leads ownership, designer royalties, and user consuming history.
- This ensures that the data is secure and tamper-proof.
3. Smart Contracts:
- Smart contracts are used to automate the payment of royalties to designers.
- When a user purchases a landing page template or consumes the service to capture leads, a smart contract automatically distributes a portion of the revenue to the designers.
4. Storage:
- Landing page templates and data of potential customers are stored on a decentralized storage network, such as IPFS.
- This ensures that the landing page is accessible to everyone, even if the platform goes offline.
5. Tokenization:
- Landing page template NFTs can be purchased and traded using cryptocurrency tokens.
- This creates a new market for designing and allows fans to support their favorite designer directly.
6. Oracles:
- Oracles are used to provide external data to smart contracts.
- For example, oracles can be used to verify the authenticity of landing page template NFTs or to track the price of cryptocurrencies.
<u>b) Business model</u>
Web3 lead-capturing services can generate revenue in several ways, including:
- Template sales: Designers can sell their templates as NFTs, through traditional subscriptions, or pay as the user goes.
- In-app purchases: Users can purchase features such as exclusive templates or designs.
- Token sales: Platforms can sell tokens to raise capital or to give users access to exclusive features.
Benefits:
- Fairer compensation for designers: Web3 lead-capturing platforms can help designers earn a fairer share of the revenue from their designed templates.
- More control for designers: Designers have more control over their templates and can choose how it is distributed and monetized.
- Direct connection between designers and businesses: Businesses can directly support their favorite designers by purchasing their templates or NFTs.
- New market for landing page design: Web3 designing platforms create a new market for design, which can help designers reach a wider audience and generate more revenue.
- Decentralization: Decentralization helps to reduce the risk of censorship and ensures that designers and businesses have more control over their data.
<u>c) AI-powered lead scoring</u>
The main idea is to assign scores to all leads based on how their characteristics match with the pre-established profile of a converted buyer. The leads that score above a specific threshold are considered ideal buyers. The bottleneck of such an approach is the determination of profile-relevant attributes with their respective weights. A decentralized AI-powered landing page generator is an attractive topic but we will propose it in the next round. In this proposal, we will build a decentralized landing page generator and focus on AI-powered lead scoring.
In Traditional (i.e., Manual) lead scoring, the task of finding the most relevant customer traits and points assignation is handled either by a marketing expert or senior sales department manager. The manager’s experience can help to determine the most optimal key attributes among thousands of possible features. Traditionally, scores are calculated based on how well the lead fits the company’s buyer portrait (demographic details) and their commitment (behavioral details). The above figure presents the architecture of a traditional lead-scoring system. Due to human nature, the sales manager cannot always define precisely what features are more important, his behavior can be biased or based on prior prejudices, he also tends to trust acquired knowledge and rarely updates the selection process.
<u>Our proposed AI approach: Predictive Lead Scoring with AI assistance</u>
The concept of predictive lead scoring is based on a statistical approach called
propensity modeling. This technique attempts to predict the chances that a potential buyer (visitor) will perform certain actions (purchase, reservation, etc.). By combining machine learning and data mining, it tries to forecast the behavior of the targeted audiences and the likelihood of successive conversions. Machine Learning (ML) algorithms can detect automatically the relations and hidden rules in historical sales data to pick relevant attributes and discover useful patterns that indicate a lead’s propensity to conversion. The constructed model is trained and evaluated to give the lowest possible false-positive predictions. If an error/anomaly is perceived in a prediction, it can be annotated and re-added to the training data which allows the model to adjust itself and stay relevant, especially in growing businesses. The below figure presents the architecture of a Lead Scoring System with AI assistance.
In simple, ML can assist with lead scoring in two ways:
- It can be used to improve the accuracy of the scores generated by AI.
- It can “teach” AI how to score leads.
To improve the accuracy of AI-generated scores, machine learning can be used to identify patterns in data that are not easily detected by humans.
- For example, a lead who regularly visits your website/landing page during work hours and reads articles about your product is likely to be further along in the buying process than a lead who only visited your site once. AI can take into account these types of behaviors to provide a more accurate score.
At the start, we use a public dataset, that has been manually scored by humans, and widely used in the lead's prediction process. The AI system will then use this data to learn how to generate scores that are similar to those created by humans. As more data is fed into the system, the AI will continue to learn and improve its ability to generate accurate scores. Second, dataset understanding is an important step in selecting relevant data points. The dataset contains several variables that cover the following aspects:
- The outcome of the Lead (Converted or not).
- Visitor actions on the website (e.g., pages visited, time spent).
- Pieces of information are collected through website forms (e.g., contact, newsletter).
- Lead source (search engine, referrer, direct).
The Dataset contains 9240 data points with 37 features. These features are properly stored for each prospect to describe its characteristics. Some features have numerical values (6 features), such as the website’s visit duration and visit frequency, others are categorical such as search keywords, lead source, and contact preferences.
Lastly, accurate prediction is our priority, so we select the machine-learning algorithm entitled gradient-boosted trees modeling based on its impact on sales performance.