USING ARTIFICIAL INTELLIGENCE IN B2B SALES: A PRIMER

OPPORTUNITY 4: SMARTER PRIORITIZATION OF LEADSCase study A software company was looking for a solution to effectively prioritise their leadsand direct their sales effortsLeadsquared provided a solution to score leads based on engagement andbehaviourThe company saw an increase in email engagement of 10-12% and an increasein customer conversion of 5-7%How did they do it?Engagement scoreLeads arecaptured inthe systemWeights are givento various factors,which are used todetermine the finalscore, based onclient’s needSource: Leadsquared; MXV interviewsAlgorithmgeneratesBased on number ofwebsite visits,frequency ofcommunication etc.in the last 30 daysLead scoreBased on leadbehaviour(communicationetc.) over the last100 daysMXV ConsultingPrediction score The two scores arefactored in to arrive atthe final score As more deals getclosed, the system isable to moreaccurately predict ascore based onfirmographic insights18

140%PREDICTED GROWTH IN ADOPTION OFINTELLIGENT SYSTEMS BY SALES TEAMSDURING 2017-2046%Source: Salesforce; ForresterOF COMPANIES SAY THAT MARKETING ANDSALES ARE THE AREAS WHERE THEY AREINVESTING IN AI THE MOSTMXV Consulting19

THREE AI TECHNIQUES TO IMPROVE SALES TEAM PRODUCTIVITYNUDGESCOACHINGPioneered by Richard Thaler and Cass Sunsteinin 2008, ‘Nudges’ are positive reinforcementsor suggestions that can influence themotivation and behaviour of individualsWhether to convert deals or retain customers,knowing what to say and how to say it iscrucialAI can enhance traditional SFA systems.Moreover, it can help determine whichreinforcements are benefitting individualmembers of the team, and personalize thenudgesTechnology can recognize sentiment inspeech, and guide reps to speak in a moreeffective mannerIt can also analyse calls after they are done toidentify best practices in deal closureSource: Literature review; MXV interviews and analysisMXV ConsultingCLIENT INTELLIGENCEHaving superior knowledge about a potentialclient can vastly improve conversion rates.Knowing when to pitch, what to pitch and themessaging to be usedUsing NLP and Machine Learning, reps cannow have real time information on potentialclients’ behaviour and history, and getsuggestions for various aspects of pitching20

Case StudyWHEN AN ALGORITHM BECOMES A MANAGERAt Uber, an algorithm can perform manytasks a manager usually doesGUIDE DRIVERS by tellingTRACK STATISTICS of driverssuch as cancellation rates, tripscompleted, ratings etcPROVIDE MOTIVATIONALNUDGES to drivers, such asthem where they can earnhigher fares and where demandis more‘Great Work!’ and ‘You’re now in thetop 10% of drivers’Questions that remain unanswered:Who is held responsiblefor faulty informationprovided to employees?Source: The New York TimesCan employees effectively voicegrievances when everything is dataand analytics based?MXV ConsultingCan management get awaywith errors or wrongdoings byblaming a glitch?21

MANY MANAGERIAL FUNCTIONS IN SALES ARE GETTING AUTOMATEDAllowing managersto focus onCertain tasks can be automatedLead qualificationForecasting quarterly revenueBuilding and growingrelationships with new &existing clientsDetermining effective pricingCreating strong salesprogrammesMIS reports and certain others augmentedConceptualizing &implementing creativesales and marketingstrategiesIdentifying emergingmarkets & opportunitiesClient intelligenceCoachingMotivatingSource: Literature review; MXV interviewsMXV Consulting22

OPPORTUNITY 5: BETTER SALES PITCHESCase studyAxtria worked with a pharma company to guide its sales team on when to meetclinics/physicians, what products to pitch and in which order using machine learningCRM data was used to obtaininformation about physicians/clinics(responsiveness, transactionbehaviour), and reps (deals made,efficiencyThird party vendors provided data onclinics/physicians (timings, productsprescribed, patients served) as wellas patients and insurance (treatmentregiment, diagnostic process)The data was run through theML algorithm that could identifypatterns of behaviour inphysicians/clinics being metProvides guidanceto reps onWHEN?WHAT?IN WHICH?to meetproducts to pitchsequence to pitchSource: Axtria; MXV interviewsMXV Consulting23

OPPORTUNITY 6: SOFTWARE NUDGES TO DRIVE SALESPERSONSCase studyIf you increaseyour daily callsby 1, you can bein the top 3 inyour territorySales effort andperformance data oneach individual sales repis stored in a CRMThese were yoursales last month –do you want toremain at thislevel?The system sends behavioural nudges to repsbased on learnings. Nudges are customized foreach sales rep with either positive or negativereinforcement. This could be in the form ofactions, goals they can achieve or areas ofimprovementAI engine usesCRM data to learnabout eachindividualThrough machine learning,the AI understands whichnudges have the most impacton KPIs, and adjusts them foreach rep accordinglyWorxogo used behavioural nudges to enable 75% of the sales team hit targets andcreate a 20% increase in ‘focus product’ sales for a client in B2B domainSource: Worxogo; MXV interviewsMXV Consulting24

OPPORTUNITY 7: Conversation INTELLIGENCE“Talk less aboutprice anddiscounting” “Talk more aroundclues to customerqualification”Intelligence platform recordsand transcribes sales calls Software identifies the key phrasesthat good reps use and unhelpfulmessages that low performing repscommunicateThis information gets more accurateover time Managers can coach sales teamsbetter based on this informationThey may also get real time alertswhen a sales rep is using languagethat is detrimental, enabling earlyinterventionCompanies deploying this technology report a 3X increase in revenue and win rate,and a 50% reduction in ramp timeSource: Gong; Literature reviewMXV Consulting25

OPPORTUNITY 7:7: CONVERSATIONINTELLIGENCEOPPORTUNITYConversation INTELLIGENCESpeechAnalysisAnalysis AndAnd In-CallIn-Call GuidanceGuidance IncreasesIncreases WinWin RatesRatesSpeechAI system processesphone conversation dataManagers can view andanalyse conversations anddevelop coaching insightsUses NLP to understandsentiment and purpose ofspeech on both endsContinuously learn frombehaviour patterns andimprove over Calls analysed on variousparametersProvide real-timeguidance to helpemployees engagebetter with clientsCompanies using this technology claim improvements in close rates of up to 15%,up to 30% acceleration in time to close and increase in revenue of up to 10%Source: Cogito; Literature reviewMXV Consulting26

8BILLIONPER YEAR SAVED IN GLOBAL BUSINESSCOSTS BY 2022 DUE TO CHATBOTS10 -20%Source: IBM; Literature reviewIMPROVEMENT IN FORECASTINGACCURACYMXV Consulting27

AI IS SUPPLEMENTING ANALYTICS FOR ACCOUNT MANAGEMENTPRICINGAI can be used for dynamic pricing of products and services foreach micro-segmentPREDICTIONBy using existing data, ML systems can predict future outcomes of salesopportunities, such as win rates and deals at risk, for an organization, leadingto easier and more accurate decision making processesFORECASTINGBy analysing sales reps KPIs and behaviour while selling, algorithms cannow accurately forecast revenue for each sales person to providequarterly benchmarks for teams to work towardsCROSS SELLINGThe skill to correctly predict which products a customer is most likely to buy along withanother can be achieved by finding patterns in customer and product dataCUSTOMER ENGAGEMENTOne of the most underutilised tools of AI in B2B sales today,engagement intelligence can improve customer lifetime value byproviding support and personalization to each clientSource: Literature review; MXV analysisMXV Consulting28

CAN MACHINES PREDICT PURCHASE BEHAVIOUR BETTERTHAN THE CUSTOMER?Amazon has developed an algorithm that can predict what customerswant before they orderCustomer data is collected,such as purchase historyand website activity, as wellas telephonic inquiries andresponse to marketingBased on this data,Amazon can predict whatan individual person orpeople in an area will buyat certain points in timeAmazon can ship theseitems to a hub nearbybefore an order is placedWhen an order is finally placed, the order will reachmuch faster than if the demand were not predictedSource: Literature reviewMXV Consulting29

OPPORTUNITY 8: EASIER AND MORE ACCURATE FORECASTINGCase studyRelatas claims to have worked with an IT services company and used AI to increasetheir forecast accuracy to 90% and reduce deals at risk by 52%How they did it?Sales forecastingbased on eachrep’s performanceInsights on dealsthat are at risk andwhich more likelyto closeThe system analysesinteractions in the datafor each rep like dealsize, time to close etc.Data from CRM,financial transactions,emails, chats andcalls fetched by theplatform (structured &unstructured data)Source: Relatas; MXV interviewsMachine Learning isused to train the data. Asnew inputs are added tothe platform, accuracy ofinsights improvesMXV ConsultingHighlights intent topurchase and nextbest stepsDrafts personalisedemails for contactswho are falling outof touch30

OPPORTUNITY 8: EASIER AND MORE ACCURATE FORECASTINGCase studyIllustrative usesSalesforce’s Einstein Prediction Builder is an AI tool for predictionwithout having to code. Einstein can predict several outcomes usingMachine Learning technology, helping businesses stay ahead of issuesForecast revenueCustomer churnExample – To Predict Which Accounts Are Likely To Pay Late12First, create a formula that defines what a‘Late Payment’ is. This is what will bepredictedSalesforce Einstein can predict whichclients are more likely to not pay on time43The system can now predict which clientsin the future will pay late, including newclients. The prediction updatesautomatically as new data is enteredSource: SalesforceNext, take all the past invoice dataavailable, both clients who have paid andnot paid. This data will be used for Einsteinto learn the factors that seem to affect alate paymentMXV ConsultingBased on the invoice data, Einstein cananalyse which factors affect late paymentsthe most, i.e. method of payment, industry,location etc.31

OPPORTUNITY 9: IMPROVED DECISION MAKINGCase studyHow did they do it?Artivatic worked with an office insurerto automate their underwriting anddecision making process1Corporates sent information to the Insurer(KYC, Financials etc.)2System searched through public data (social,public data, interactions) for more data points3The ML engine trained using existing data toprovide accurate claim conditions andpremiums4The self learning AI used this data todetermine the specifics of the new policy foreach company the insurer is working with5Automatically approved policies after learningfrom past actions and decisions takenOutcome85%Accuracy of policy personalizationand predictionReduced decision time from 4-7days earlier to40%2-3hoursIncrease in conversion rateSource: Artivatic; MXV interviews System monitored the policies for automatedclaims processing6 Health activities of the companies trackedover time to update policy premiumMXV Consulting32

OPPORTUNITY 10: DYNAMIC PRICINGCase studyExamplesEnergy management &automation companyA Fortune 500 electric company was facingissues with their manual pricing system. Theyreceived 40,000 requests for negotiated priceseach month, and their quote turnaround timewas too slowAfter switching to an AI powered pricingmechanism, the company managed to createover 10,000 pricing segments for differentcustomers and products. The segmentsresponded to factors such as project type,geography, product mix etc.The result was a decrease in job quoteturnaround time from several days to lessthan 4 hours. Additionally, the manufacturerrealized a 100% ROI on the solution in lessthan one year.Source: Literature reviewBuilding products manufacturerInconsistent pricing prevented thismanufacturing company from earning fairmargins. Sales reps were pricing products at alow rate to protect against customer pushbackThe company decided to implement an AIdriven pricing solution for a part of theirproduct line, while maintaining manual pricinganalysis for the rest as a comparisonThe financial results showed that the productspriced using AI had margins that were 130bps higher than those priced traditionally(2.3% compared to 1%). Moreover, this wasachieved while maintaining historical volumesMXV ConsultingIndustrial supplies companyA European Maintenance, Repair andOverhaul (MRO) distributor needed help withprofitability and pricing compliance in severalEuropean marketsThe company realised that technology on itsown was not enough to maintain a sustainedgrowth. Leadership, management andengagement were also required to effectivelyleverage the AI pricing solutionAlong with strengthening the competency ofemployees, the company saw a 200 bpsincrease in margins in the first three months,with 95% compliance achieved33

OPPORTUNITY 11: ENHANCED ENGAGEMENT FOR ACCOUNTSCommon issues that customers havewith current resolution systemsLong wait times to speak to executivesLengthy conversation required to reach aproblem / solutionRepetition of new information when talkingto new executivesHow Chatbots can resolve these issuesInstant replies when asked aquestionEngagement chatbotsuse NLP to understandsentiment and intent ofcustomers, and ML tolearn their behaviourpatterns to optimizeresponse time andinformationUnderstands intent and has learntrelevant problems to identify issuesfasterLearns each customer’s data forincreased personalization andaccount knowledge with each andevery interactionAvailable 24x7 and on multiplechannels (social, website, mobileapplication)Unavailable to speak at all hoursExampleApplied its chatbot to an insurance company, to answer commonqueries regarding policy questions and premiums. As a result,customer engagement scores increased by 64%Source: Avaamo; Literature reviewMXV Consulting34

OPPORTUNITY 12: MORE ACCURATE CROSS SELLINGCross sell analysis methodsCustomer SegmentationCustomers are segmented based onshared attributes. These segmentsare then analysed by algorithms tocross sell products that are relevant toeach categoryThese include average spend,location, industry, annual revenue,products purchased etc.Market Basket AnalysisClusteringUsing this method, companies candetermine which products are more likelyto be bought if a given mix of productshave already been bought in the pastWhile segments are created based onpre-existing attributes, clusters findsimilarities between customers who havepurchased a certain item.Using an algorithm, a system can predictthe mostly likely product that a givencompany will buy, based on companiesthat buy the same products as themIf product A has been bought by clientswho have lower revenue and based inDelhi, the system will pick this up andsuggest to pitch this product to other lowrevenue Delhi based companiesFor all 3 methods, machine learning is used to improve accuracy of cross selling prediction and groupingsExampleZilliant provided cross selling insights for amotion and fluid control technologymanufacturer using Machine LearningSource: Zilliant; Literature reviewMXV ConsultingOutcome21%Increase in customer revenueafter the first two months35

IN BRIEF: 12 WAYS AI CAN TURBO CHARGE YOUR SALESMore Accurate Cross Selling afterEnhanced Lead Generationsegmenting and clustering client datausing NLP systems12Dynamic PricingMechanisms which use real1112Dynamic LeadQualification through intentbased classificationtime market dataEnhanced Engagement ForExisting Accounts usingAI INB2BSALES10chatbotsAutomated And ImprovedDecision Making that improves349Higher Engagement withProspects by employing bots andreinforcement learningSmarter Prioritization ofLeads based on prospectbehaviour and activityover timeEasier And More AccurateForecasting via each sales rep’s58pipeline evaluationBetter Sales Pitches by identifying76patterns of client behaviourConversational Intelligence toIndividualised Nudges to incentiviseguide and coach sales teamssales reps based on personal motivationMXV Consulting36

FOR THOSE WHO SUCCEED, THE IMPACT CAN BE VERY LARGEOn artificial intelligence interventions3DramaticimprovementUp to50%2Marginalimprovement(by 10%)22% 11%510%Scenario of Indian salesteams without SFA22%510%After sales force automationDramatic increase(by up to 50%)1Dramaticimprovement(by up to 20%)510%Salesperson’s faceto face selling time Sales force automation (SFA)frees up time for actual sellingby automating lot of manualtasks and sales processes.And the data captured formsthe base for measuring effortsand outcomes to drive overallsales productivity Also, a SFA tool forms thebase for AI driveninterventions, the impact ofwhich can be significant onsales conversion rateSales conversion rateSource: Literature review; MXV case database and estimatesMXV Consulting37

ENDNOTE: A CALL TO ARMS20% and 10% - two numbers to consider. A maximum of 20% of all B2Bsalespersons time is spent in front of customers and barely 10% of leads convertinto sales. This is not a matter of efficiency alone. Administrative work, follow-upsand prospecting take a lot of productive time away – while poor lead quality,immature qualification processes and lack of engagement result in large drop offs.The job of a salesperson is challenging in the best of times.This whitepaper outlines 12 principal opportunities to strengthen sales using AI.There will be others. But these 12 act as a solid starting point. We can increase thequality and quantity of leads, improve the process of conversion and enhance thenature of customer engagement – all of these have tangible business results – andthey can be initiated today.Yes, there are definitely some obstacles and hesitancies with regards to adopting AIon a broad scale. For one, the data infrastructure of many Indian companies is notstrong enough to effectively implement AI technologies. Either inadequate data is

B2B sales funnel Use cases (Sample) Solution providers SAMPLE APPLICATIONS IN THE B2B SALES PROCESS MXV Consulting Trawl through websites and social media, convert unstructured data into structured data and apply algorithms to find the right fit Finding contacts of key personnel within a company Engage website leads