Best SEO,

Simply Understandable Data

MOBJAI™ makes sense of your data in a way anyone can understand. By using an Ai first approach and using your own data, Mobjai and Mo provide predictions and Analysis of your business with trusted strategic insights for success. It all starts with your data.

Pre Flight Checklist

What must my data have to be able to make predictions?

  • Data should be in a Comma Separated Value (CSV) file.
  • Predictions require a Minimum of 10 columns to describe the data, with names. Usually the first row in your data, the more columns the better.
  • One of the columns should be a 'Binary Outcome' column, something we want to predict - for example - CLOSED Won Or Closed Lost?
  • Predictions require a Minimum of ±500 Transactional Rows of data each row should have a UNIQUE id
  • Some of the rows will have historical known outcomes - for example - "won" or "Lost"
  • Some of the rows will have unknown outcomes - these are the future transactions we want to predict and should be blank.
UNIQUE id
SalesPerson
Account
Date
Product
Region
Colour
Discount %
Amount
Closed?
1
Jane
Unilever
Jauary
4675
London
Blue
10
£1450
Won
2
Tim
Ford
February
4673
Midlands
Yellow
20
£6792
3
Jane
EE
January
4775
North
Blue
50
£8796
Lost

What does good data look like?

walkthrough data example for MOBJAI

End to end walk through

  • Can we predict if someone will buy our product or service?
  • 3960 Transactional Rows of data used as examples with known outcomes.
  • 1616 Transactional Rows of data with outcomes to predict.
  • 27 Columns describing the transactions.
  • Column AA 'CONVERSION' is the outcome column - 'Converted' or 'Not Converted'.

Sales Opportunity Prediction

Download the example data files and use with your
7 Day MOBJAI™ Trial

SalesLeads2023.csv

3960 Rows, 27 Columns, Known & Unknown Outcomes in Column AA

Sales Use case walk through

In the example above we have a full dataset of 3960 historical and possible transactions with ±2344 rows of data as known outcomes showing whether an customer converted on our website or not at a specific time and under specific circumstances.

The 27 Columns describe the situation for that record or the circumstances that individual was under which may have contributed to why they bought or churned.

There is a very definite outcome column, "Conversion", which gives a binary outcome of 'YES' they did buy or 'NO' They did not buy.

Hidden in all of this data is a prediction. We use a technique called Binary Classification to create multiple Ai models using machine learning.

Binary classification is used to predict one of two possible outcomes. Examples include predicting whether a customer will make a purchase, identifying fraudulent credit card transactions, detecting new planets in deep space signals, or determining the presence of a disease based on medical test results. In the initial data, the outcomes may be represented by labels such as "Yes" and "No" or "Sun" and "Rain".

Can we work out with some accuracy the reasons why people made the decision to buy or not and if so, if we present another customers circumstances, are we able to make a prediction whether they will buy?

So we use this initial 3960 row * 27 Column Data as a Training Data Set where we know the outcomes and the circumstances. Training on this Data Set creates multiple Ai models capable of making a prediction under many different circumstances.

The prediction is a percentage score. 100% means they are definitely going to buy and 0% means they will definitely noy buy. The predictions will be somewhere along this line from 0% to 100%.

It is important to understand Feature selection. This is where we select those columns or features that have an impact on the accuracy of the model and deselect those may have a negative impact.

Features are simply columns in your data that may or may not be contributing factors to the outcome we are looking to predict from your historical records - price, brand, time on website, basically everything that you’ve managed to capture.

Let’s take a look at a scaled down example. You have 6 column headers “time on website”, “emails sent”, “return customer y/n”, “item category”, “deal stage” and “closed won/closed lost”

We would need to deselect “deal stage” because anything that happened past the point of closing can’t impact the outcome.

The outcome has already occurred so we know if “deal stage” is “closed” then the deal did actually close. However, We also know “closed won” is yes.

MOBJAI will look for a shortcut and see all “deal stage” closed deals are also “closed won” yes deals. So will use the impact of the deal stage with a high weighting which affects the model predictions.


The same would be true of columns that are there for your own internal reference like customer ID or transaction record ID - these won’t contribute to the outcome at all so better to remove them.

In the Sales use case example there are no features we need to deselect, everything is relevant to the accuracy of the model.

When we load our full 3960 transaction CSV file, containing both known and unknown outcomes, MOBJAI splits the data into a MISSION dataset of known outcomes and an OBJECTIVE dataset of unknown outcomes.

Once the 'MISSION' training set is loaded, we split the Data into 2 further new Data Sets; Train and Test. We only use ±75% of the data for training so that once we have created the Ai models, we can feed in the 25% of test data to check that we are getting the right predictions because of course we already know the outcome of that 25%. This means we can test multiple models accurately and select the best performing Ai producing accurate results against the data we know outcomes for.

When we introduce a new, fresh, unseen set of 'OBJECTIVE' data to the best performing model, with just the circumstances but not the outcome, we can predict the likelihood of whether that customer will buy or not.

The whole point of creating predictions is to focus your resources of time, effort and money.

Sample Output from Mo

“Oh my goodness, I am absolutely thrilled to review your business data! I can see some exciting opportunities to grow revenue and save money, based on the business drivers you provided.

Firstly, let’s focus on the time spent on your website. This is an incredibly important driver of online sales, and it’s fantastic to see that it makes up 30% of your current business. However, we need to ensure that this time is being spent efficiently and effectively, leading to higher conversion rates and ultimately, more revenue.

One recommendation I have is to invest in optimizing your website’s user experience. This could involve improving page load times, simplifying navigation, and ensuring that product information is easy to find and understand. By doing so, we can encourage visitors to spend more time on your website and increase their likelihood of making a purchase.

Secondly, let’s talk about response to email number 5. While this driver only makes up 7% of your business, it is still a crucial touchpoint with your customers. We need to ensure that your email communication is on point, personalized, and relevant to each customer’s needs and preferences.

One recommendation I have is to implement a customer relationship management (CRM) system that allows you to track customer interactions and personalize your communication accordingly. This can include targeted email campaigns, personalized product recommendations, and loyalty programs. By improving your email communication, we can increase customer loyalty and ultimately, drive more revenue.

Finally, let’s look at the impact of sending a question to support, which makes up 3% of your business. While this may seem like a small percentage, it’s important to remember that customer support interactions can have a significant impact on customer satisfaction and loyalty.

One recommendation I have is to invest in improving your customer support channels, such as implementing live chat or chatbots on your website. By providing quick and efficient support, we can increase customer satisfaction and ultimately, reduce the number of returns and refunds, saving you money in the long run.

Overall, I am absolutely thrilled with the potential opportunities for growth and savings based on the business drivers you provided. By optimizing your website’s user experience, improving your email communication, and investing in customer support, we can drive revenue and save money while enhancing the overall customer experience.”

More questions to ask

Sales opportunity scoring is a method used to prioritize and evaluate potential sales leads based on their likelihood of closing. This can help a business by:

  1. Focusing sales efforts on the most promising leads, which can increase the chances of closing deals and achieving sales targets.

  2. Identifying which leads are most likely to convert into customers, allowing the business to allocate resources more efficiently.

  3. Tracking the performance of different sales strategies, so the business can identify what is working and what is not.

  4. Providing insights into which leads are most valuable to the business, so the business can target similar leads in the future.

  5. Managing the sales pipeline more effectively, which can help the business forecast revenue and identify potential issues before they become a problem.

Customer retention scoring is a method used to evaluate and prioritize existing customers based on their likelihood of continuing to do business with a company. It can help a business by:

  1. Prioritizing customers based on their likelihood of remaining loyal, which allows the business to focus on retaining its most valuable customers.

  2. Identifying which customers are most likely to leave, which allows the business to take proactive measures to keep them.

  3. Tracking the performance of different retention strategies, so the business can identify what is working and what is not.

  4. Providing insights into which customers are most valuable to the business, so the business can target similar customers in the future.

  5. Managing customer relationships more effectively, which can help the business forecast revenue and identify potential issues before they become a problem.

  6. Cost-effective as retaining an existing customer is cheaper than acquiring a new one.

  7. Helps in building a strong brand reputation and customer satisfaction.

  8. Helps in identifying customers that have a high lifetime value and potential to become brand advocates.

Sales lead scoring is a method used to evaluate and prioritize potential sales leads based on their likelihood of becoming a customer. It can help a business by:

  1. Prioritizing leads based on their likelihood of converting into customers, which allows sales teams to focus their efforts on the most promising leads.

  2. Identifying which leads are most likely to become customers, which helps the business allocate resources more efficiently.

  3. Tracking the performance of different sales strategies, so the business can identify what is working and what is not.

  4. Providing insights into which leads are most valuable to the business, so the business can target similar leads in the future.

  5. Managing the sales pipeline more effectively, which can help the business forecast revenue and identify potential issues before they become a problem.

  6. Lead Scoring also helps the sales team to stay organized and have a clear understanding of which leads to focus on, and which ones are less likely to convert.

Cash collection scoring is a method used to evaluate and prioritize customers based on their likelihood of paying their outstanding invoices on time. It can help a business by:

  1. Prioritizing which customers to follow-up on first, based on their likelihood of paying on time.

  2. Identifying which customers are at risk of not paying their invoices, which allows the business to take proactive measures to improve their chances of receiving payment.

  3. Tracking the performance of different collection strategies, so the business can identify what is working and what is not.

  4. Providing insights into which customers are most likely to pay on time, so the business can target similar customers in the future.

  5. Managing cash flow more effectively, which can help the business forecast revenue and identify potential issues before they become a problem.

  6. Improving the overall efficiency of the Accounts Receivable process by focusing on the customers who are most likely to pay.

  7. Helping to reduce bad debt and late payments, which can have a significant impact on the financial health of the business.

  8. Improving relationship with customers by effectively communicating and reminding them of their outstanding payments.

Marketing spend optimization using Machine Learning is a method of using data analysis and machine learning algorithms to optimize marketing budgets and campaigns. It can help a business by:

  1. Identifying which marketing channels and tactics are most effective, which allows the business to allocate budget more efficiently.

  2. Optimizing marketing campaigns in real-time by adjusting the budget and targeting based on results.

  3. Improving the ROI of marketing spend by identifying the most profitable target audiences and channels.

  4. Providing insights into customer behavior and preferences, which can help the business improve its overall marketing strategy.

  5. Identifying patterns, trends and hidden insights in customer data which can be used to create more effective marketing campaigns

  6. Automating the optimization process and reducing the human error which can lead to more accurate predictions and decisions.

  7. Reducing the time and costs of running manual optimization tests, allowing the business to experiment with more options.

  8. Continuously learning and adapting, which can help the business stay ahead of the competition.

Student churn identification and intervention using Machine Learning is a method of using data analysis and machine learning algorithms to identify students who are at risk of leaving a school, college or university and to intervene to prevent it. It can help a business by:

  1. Identifying students who are at risk of dropping out early, which allows the institution to take proactive measures to retain them.

  2. Identifying patterns, trends and hidden insights in student data which can be used to understand the reasons why students are leaving.

  3. Automating the identification process and reducing the human error which can lead to more accurate predictions and decisions.

  4. Reducing the time and costs of running manual analysis, allowing the institution to identify more students who are at risk of leaving.

  5. Improving retention rates, which can have a significant impact on the financial health of the institution.

  6. Identifying students who may be in need of additional support, such as academic or financial assistance, which can help to improve their chances of success.

  7. Improving the overall student experience by addressing the issues that are causing students to leave.

  8. Continuously learning and adapting, which can help the institution stay ahead of the competition.

Identifying sports injury recovery optimization using Machine Learning is a method of using data analysis and machine learning algorithms to optimize the recovery process of athletes who are injured. It can help a business by:

  1. Identifying patterns, trends and hidden insights in injury data which can be used to understand the factors that contribute to injury and recovery.

  2. Automating the identification process and reducing the human error which can lead to more accurate predictions and decisions.

  3. Optimizing the recovery process by identifying the best treatments and rehabilitation strategies for each individual athlete.

  4. Reducing the time and costs of running manual analysis, allowing the team or institution to identify more athletes who are at risk of injury.

  5. Improving recovery rates, which can have a significant impact on the performance of the team or institution.

  6. Identifying athletes who may be at risk of re-injury, which can help to prevent future injuries.

  7. Improving the overall athlete experience by addressing the issues that are causing injuries and prolonging recovery.

  8. Continuously learning and adapting, which can help the team or institution stay ahead of the competition.

Predictive maintenance optimization using machine learning is a technique that uses data analytics and algorithms to predict when equipment or machinery is likely to fail, so maintenance can be scheduled proactively. This helps organizations to optimize their maintenance activities and improve equipment performance and reliability.

By using machine learning algorithms to analyze data from equipment and other assets, organizations can gain insights into usage patterns, performance issues, and other factors that can impact equipment longevity. This information can then be used to develop predictive models that identify when equipment is most likely to fail.

The benefits of predictive maintenance optimization using machine learning include:

  1. Cost savings: By predicting equipment failures before they occur, organizations can reduce the costs associated with unplanned downtime and emergency repairs.

  2. Improved efficiency: Predictive maintenance enables organizations to schedule maintenance work proactively, reducing the need for unscheduled downtime and improving overall efficiency.

  3. Better asset management: By using machine learning algorithms to analyze data from equipment and other assets, organizations can gain insights into their performance and usage patterns, enabling more effective asset management.

  4. Increased safety: Predictive maintenance can help identify potential safety hazards before they occur, reducing the risk of accidents and improving overall workplace safety.

Overall, predictive maintenance optimization using machine learning can help organizations achieve significant benefits, including cost savings, improved efficiency, and increased safety.

What Can I use Ai for?