Boosting Profits with AI Scoring in Sales & Marketing
When it comes to commercial activities, there is a vicious feud between marketing and sales departments in most businesses.
Marketing teams are tirelessly working to convert as many leads as possible through various ad campaigns, events, social engagement, SEO, SEM and many more tools. The teams end up with a database of new leads who have shown interest in the product and might have some willingness to make a purchase, aka leads.
In some companies the marketing executives also handle lead nurturing where they engage with the leads through different interactions to better excite their leads into purchasing higher volumes or more frequently. Usually, this process is called pushing leads down the funnel which is conveniently presented in the figure below
Once the lead is ready to buy, sales executives start rushing to close the leads and make sure they make the transaction. As the business grows, and the marketing team gets more proficient with their execution, naturally the number of ready-to-buy leads increase in quantity; and when that happens, salespeople can’t of course go through each and every customer without some form of assistance Therefore, sales executives either resort to picking at random or relying more on their network of clients rather than the leads sent by the marketing team.
The inevitable happens, conflict arises, and arises fast. The sales department disputes that the marketing people are not generating high quality leads, or building attractive product mixes, and Marketing teams usually believe that sales are not as effective as they think.
The fact is that there is always many lost opportunities that were not attended to properly, who were just a phone call, a push notification or an email away from transacting.
This is where the true impact of lead scoring comes into play. Lead Scoring is arguably the most effective tools when it comes to Marketing and Sales and increasing the conversion rate for any business. Lead Scoring is giving each lead or potential customer in your database a point-value, to determine their readiness and willingness of buying your product.
Presenting the Evidence
Traditionally, lead scoring is done through obsolete scan and point methods, where a domain expert looks at any database of potential buyers and prioritizes the few whom he or she thinks are most likely to buy within that specific industry.
The lead scoring algorithm is exposed to the historical data of customers and recognizes patterns in demographics, behaviors, messaging, channels, seasonality, and other dimensions, and Auto-prioritizes the database, on a score given to each individual lead on their readiness and willingness to buy. Through the process, the Machine Learning (ML) algorithm classifies the leads/customers within the database into multiple behavioral segments, for business managers to better understand their audience.
Picture: From Synapse Analytics Konan* MLOPs Platform
As shown the above platform screenshot, the customers of this particular business were segmented into 10 main segments, based on the behavior of these clients to the product mix that the client is willing to push on them.
“The Champions” are of course the best ones who are Willing and Capable to buy the product at hand. While the “At Risk” segment, are about to leave for another vendor or company and require immediate attention.
All the leads in each segment are given a behavioral score based on which the algorithm allocates the lead to the relevant class as is illustrated and highlighted in the screenshot of the platform below.
Picture: From Synapse Analytics Konan MLOPs Platform
The data input to such a model can be:
· Demographic Data (Occupation, Age, Gender, Place of Living, Job Title, etc.)
· Behavioral data (Interacting with Email, Forms, Web-properties, Webinars, etc.)
· Segmentation Data (part of a certain cohort, Past customer, Channel from which they were acquired, etc.)
Sales teams will look at the auto-sorted list of leads who are most ready to buy, for example having a score >0.7 (70%). While marketers can identify the leads who need more nurturing, those who fall between 0.3 & 0.6, and start their engagement to make them more ready through different reach-out strategies and ideas.
This ensures that sales are not losing time with leads who will not buy, and marketers can keep engaging with those who will buy in the future so as not to lose them for the competition through regular offers and communication.
Furthermore, your marketing team can understand what the best channels are to target each segment for a particular offer or product mix. The team can further predict the outcome of their campaigns based on the ML modeling their customer based and predicting their engagement and conversion. Not to mention, that your commercial team can better allocate their budgets, to minimize the cost of acquisition of leads, and better retain the existing customers. The previous can be clearly seen in the below illustration of the platform.
Picture: From Synapse Analytics Konan MLOPs Platform
Data Scientists, along with marketers and sales managers can build ML Lead Scoring algorithms that can and will dramatically increase your lead to customer conversion rates by offering your customers better experience, focusing your marketing initiatives and activities, and ensuring you have more productive commercial activities.
According to Stu Schmidt, Vice President of Solution Sales at Cisco Webex, only a 10% increase in lead quality, can add up to 40% more sales productivity.
Since businesses are always looking for impact driven initiatives, let’s go through a scenario, to see the impact of ML based lead scoring.
Let’s assume your company, let’s call it A & Co., has 10 Sales reps. Since we are considering a data science use-case in Egypt, let’s assume that the average salary is EGP 240,000 ($15,200 USD) per sales rep per year.
We can imagine that your marketing team converts a 1,000 leads per month. We know that, on average, there are ~60- 70% bad leads in a given leads list. It’s a bit tricky to calculate the opportunity cost of such a statistic, but we can estimate the cost of bad leads by multiplying the EGP 240,000 * 10* 70% = EGP 1,680,000 / Year. This is the cost wasted on leads that never close into buying customers.
Further, let’s assume a lead to sale conversion rate of 3%, and the average sales value is about EPG 200,000 (~$1,270). If the sales team close 30 deals per month, your sales team would be bringing a 6,000,000.00 EGP/Month (~$382,160), this would account for a revenue of EGP 72,000,000 (~$4,600,000.00).
To simplify the case, let’s assume that the conversion rate of 3% is consistent every month for 12-months.
The company then decides to hire the best data scientists in Egypt and asks them to help the marketing and sales teams into building a baseline ML based lead scoring model to increase the conversion rate. Since, you are the early phases of adoption, we can predict that the initial model would increase current conversion by 5%.
This would increase the 3% conversion rate to 3.15%, closing 31.5 deals per month, increasing the monthly revenues from EGP 6,000,000.00 to EGP 6,300,000.00.
Taking a yearly view, your machine learning model would translate into EGP 75,600,000.00 (~$4,800,000.00 USD) yearly revenue, accounting for a total increase of EGP 3,600,000.00 in the first year.
One key piece of information about ML models is that they get better the more they are used in operations and are exposed to better data. Using a MLOPS (Machine Learning Operations) platforms such as Konan will ensure that your models in production are getting better, and more accurate with everyday use.
That said, the previous is a worst-case scenario, if for example, the model got better with time and the second year the conversion rate increased by 10%, closing 33 deals per month. Then the monthly revenue would increase from EGP 6,000,000.00 to EGP 6,600,000.00; and you can see how this trend would continue every time the model gets better.
Let’s try and summarize the cast below
Therefore, just a 5% increase (which is just the baseline) can lead to an EGP 3,600,000.00 difference in yearly revenue.
If you are interested in a lead scoring impact calculator for your business, click this link.
So, with an initial base ML model to score the leads, A & Co. is now closing ~1.5 more deals per month, which is reflected directly in your topline. Moreover, as the model is getting better, the machine is learning and getting more accurate, reaching a 3.85% conversion rate by end of year, increasing the total sales by 20%.
As shown, lead scoring can be a super transformative investment to any business who is looking to grow and have more effective commercial operations, while making sure to interact with relevant customers and retain existing ones.
If you are looking for the tools, techniques, and talent to kickstart your Lead Scoring initiatives within your company, Synapse Analytics has an expert team of data scientists, machine learning engineers, and software engineers that work to create our AI platform KONAN that can completely streamline and uplift your business. Feel free to contact us anytime to get more information on how to kickstart and implement KONAN within your business.
*KONAN: Konan is a Machine Learning Operations platform that helps our clients to operationalize their machine learning models, to accelerate machine learning impact, while bridging the gap between data scientists, IT/application managers and business users.