Alexandre Pericão Artificial intelligence should no longer be treated as an isolated feature added to a CRM platform. In 2026, the B2B companies generating the greatest value from technology are those connecting data, processes, automation and human intervention within a single commercial system. This is what AI marketing and sales automation means in practice: using artificial intelligence and automation to identify opportunities, prioritise accounts, personalise communications, accelerate response times, support commercial decisions and continuously learn from the results recorded in the CRM.
This evolution includes everything from analytical and recommendation capabilities embedded within CRM platforms to systems capable of interpreting requests, retrieving information and completing tasks across different applications. Explore the main applications of artificial intelligence in Marketing and Sales and discover how AI agents can execute processes connected to CRM, customer service, scheduling and business workflows.
The opportunity is real, but execution remains uneven. McKinsey research indicates that Marketing and Sales are among the business functions where organisations most frequently report revenue increases associated with the use of artificial intelligence. The same research shows, however, that the strongest results tend to emerge in companies that redesign their workflows, rather than simply adding new tools to outdated processes. Read McKinsey’s State of AI report.
This distinction is critical. A CRM with artificial intelligence capabilities does not automatically correct incomplete data, inconsistent qualification criteria, slow response times or a lack of alignment between Marketing and Sales. Technology amplifies the quality of the system in which it operates. When processes are poorly designed, it can also amplify mistakes, duplicated work and inconsistent decisions.
Salesforce’s State of Sales 2026 reinforces the scale of this transformation. According to the report, 87% of sales organisations already use some form of artificial intelligence. High performing teams are 1.7 times more likely to use AI agents for prospecting. Read the Salesforce State of Sales 2026 announcement.
The objective is not to replace sales professionals. It is to remove repetitive work from the process, provide teams with better context and concentrate human effort on the decisions, relationships and negotiations that genuinely move opportunities forward.
How can companies generate more pipeline with marketing automation and AI?
To generate more pipeline, a company must improve four variables at the same time: the number of relevant accounts identified, the speed at which those accounts are approached, the quality of personalisation and the percentage of opportunities that progress between stages. Marketing automation and artificial intelligence can contribute to all four variables when they operate on reliable data and clearly defined commercial processes.
In practice, producing more emails, publishing more content or adding a chatbot to the website is not enough. More activity may simply create more noise. The objective should be to build a system capable of recognising intent signals, connecting those signals to the ideal customer profile, determining the next appropriate action and recording the result in the CRM.
Automation should not be understood solely as a mechanism for sending campaigns. It can support lead qualification, opportunity distribution, meeting preparation, sales follow-up, forecasting, the recovery of inactive opportunities and the measurement of Marketing’s contribution to pipeline.
The following seven plays are organised in a logical sequence. They begin with data quality and integration and end with the creation of a revenue system that continuously learns from results.
1. Unify customer data before automating decisions
The first play may be less visually impressive than deploying an AI agent, but it determines the quality of every other initiative. The CRM should operate as the central source of information about contacts, companies, interactions, opportunities, proposals and customers. When data is dispersed across spreadsheets, email platforms, invoicing systems, forms, calendars and prospecting tools, artificial intelligence receives a fragmented view and produces equally fragmented recommendations.
CRM integration should begin with a map of the existing data sources and a clear definition of which system is responsible for each type of information. A form may create the contact, an ERP may validate customer status and invoicing information, a Marketing platform may record engagement, and the CRM should consolidate the commercial relationship.
It is also necessary to standardise properties such as industry, company size, country, source, product interest, consent, lifecycle stage and sales owner. Without this discipline, lead scoring loses accuracy, segmentation creates the wrong audiences and reports become impossible to compare.
The same problem creates operational risks. Different departments may approach the same person without context, use conflicting versions of the information or interpret the stage of an opportunity in different ways.
Learn how to implement CRM and Marketing Automation as an integrated revenue system.
The problem is not theoretical. In Salesforce’s State of Sales 2026, 46% of sales professionals working with AI agents stated that data quality problems were damaging sales performance. The report also indicates that 51% of sales leaders using AI believe technology silos delay or limit these initiatives. In addition, 42% of sales professionals feel they are working with too many tools. Read the full State of Sales 2026 report.
For many SMEs, the right decision is not to purchase another application. It is to simplify the technology architecture, integrate the systems that are genuinely necessary and eliminate duplicated work.
A unified view also enables Marketing and Sales to work from the same commercial reality. Marketing can understand which campaigns and content assets are associated with actual opportunities. Sales receives context about each account’s previous journey. Management can compare investment, pipeline and revenue without relying on manual reconciliation between contradictory reports.
Before moving into more sophisticated use cases, companies should measure the percentage of complete records, the duplication rate, the number of activities recorded automatically, integration coverage and the time spent manually entering information. These indicators reveal whether the organisation has an operational foundation capable of supporting automation with confidence.
Integration should also follow a clear ownership model. The objective is not to synchronise every field across every platform. It is to define which system owns each data point, which data needs to move between systems and which information should remain available only in its original platform.
Explore Liminal’s CRM, ERP and Marketing and Sales integration services.
2. Use AI to score leads, accounts and intent signals
Traditional lead scoring assigns points according to fixed rules, such as downloading an eBook, visiting a pricing page or belonging to a specific industry. This method remains useful, but it has clear limitations. It treats distinct behaviours as if they always had the same meaning and rarely learns from the opportunities that were actually won or lost.
Artificial intelligence enables a more dynamic form of prioritisation by combining fit and intent.
Fit evaluates whether the account matches the ideal customer profile. This may include industry, size, geography, business model, technology used or operational complexity.
Intent evaluates behavioural signals. These may include repeated website visits, interaction with lower funnel content, responses to campaigns, event attendance, demonstration requests or relevant changes within the company.
The practical application is to build a score that indicates not only who appears interested, but who is more likely to become a valuable commercial opportunity. The model can use CRM history to identify patterns shared by won deals, lost opportunities and disqualified leads.
The sales team then receives a prioritised list accompanied by the factors supporting each recommendation. This level of explanation is important because a score without context is likely to be ignored.
Learn how lead scoring works in HubSpot, Zoho CRM and Salesforce.
The model should not consider digital activity alone. A company that repeatedly visits the website may be interested, but it may not have the size, budget or need required to become a viable opportunity. Conversely, a strategic account may display few digital signals and still justify proactive commercial attention.
The score should therefore combine behavioural signals, company characteristics, the existing relationship, potential value and strategic priorities. It should also distinguish between an isolated activity and a pattern of behaviour. Downloading introductory content may represent curiosity. Repeatedly visiting solution, integration, pricing and customer story pages may indicate a more advanced buying intention.
To prevent automation from amplifying historical mistakes, Marketing, Sales and management should review the model. Criteria such as geography, industry or company size may reflect the current strategy, but they can also perpetuate old decisions or exclude emerging segments.
The strongest system combines predictive analysis, business rules and human validation.
The core metrics are conversion rate by score range, the percentage of leads accepted by Sales, pipeline value generated and the average time between the first relevant signal and commercial contact. A model that produces sophisticated scores without improving any of these metrics adds complexity without creating business value.
Platform selection also affects the scoring, automation, reporting and AI capabilities available to the organisation. However, technology should not be selected before the company has clearly defined the process the system needs to support.
Download Liminal’s guide to choosing the right CRM for the company’s processes and objectives.
3. Automate lead response, qualification and distribution
A lead with high purchasing intent loses value when it remains unattended in an inbox, is assigned to the wrong person or receives a generic response several days later. The third play is to design a process that responds to, qualifies and routes each request according to its context, priority and the capacity of the sales team.
The process may begin with a form, chatbot, email, event registration or an integration with an external platform. Artificial intelligence can summarise the information received, enrich the record using authorised public data, identify the likely subject of the request and suggest additional qualification questions.
Automation can then check territory, industry, product, company size, language, existing account ownership and distribution rules. It creates the lead or opportunity, assigns the owner, schedules a task, sends a contextual response and creates an alert when the agreed response time is not met.
This approach improves pipeline because it reduces the number of leads lost between Marketing and Sales. It also prevents every request from being treated in the same way.
A company downloading introductory content may enter a nurturing programme. A strategic account repeatedly visiting solution pages and requesting pricing information may be routed immediately to a senior sales professional. A lead that does not fit the target profile can receive useful information without consuming the sales team’s capacity.
Salesforce found that sales professionals spend almost one working day per week on prospecting and that 48% believe they do not have enough capacity to complete the required volume of cold outreach. Among professionals using AI agents for prospecting, 92% report that the technology benefits this activity.
The advantage is not simply sending more messages. It is freeing up time, maintaining consistent coverage and directing human intervention towards the opportunities that deserve attention.
The automation must also include exception rules. Existing customers, partners, job applications, suppliers and support requests should not enter the same process as new sales opportunities. The system must control duplicates, previous ownership, strategic accounts and contacts already being approached by another team member.
Otherwise, the speed of automation can produce uncoordinated communication and damage the prospect’s experience.
The relevant metrics include time to first response, the percentage of leads contacted within the agreed timeframe, meeting booking rate, qualification rate, distribution by sales owner and opportunities lost due to a lack of follow-up. Without these metrics, automation may appear efficient while concealing serious operational failures.
To understand how one platform can centralise forms, contacts, Marketing, Sales, workflows and reporting, companies can examine a practical CRM ecosystem.
Explore HubSpot’s integrated approach to CRM, Marketing, Sales and Customer Service.
4. Personalise nurturing and prospecting using real CRM context
Personalisation no longer means inserting the company name into the subject line of an email. In B2B, a relevant message should reflect the industry, the person’s role, the likely challenge, the maturity of the account, previous interactions and the stage of the decision process.
Artificial intelligence can prepare variations quickly, but the context should come from the CRM and from previously approved information sources.
A nurturing sequence may adapt the content according to the subject researched, level of engagement and existing relationship. In outbound prospecting, an AI agent may research the account, summarise relevant developments, identify potential priorities and prepare an initial draft.
The sales professional validates the hypothesis, adjusts the tone and decides whether the message should be sent. This combination preserves control and quality while reducing the time dedicated to research and preparation.
HubSpot reported in 2026 that 69.2% of Marketing professionals believe they are capable of using customer data to create personalised experiences. The same report indicates that 93.2% believe personalised or segmented experiences contributed to generating more leads and purchases. Read HubSpot’s 2026 State of Marketing report.
Technical capability does not, however, guarantee differentiation. As more companies use artificial intelligence to produce communications, generic content becomes easier to recognise and ignore.
The use of AI only creates a genuine advantage when it helps the company understand each contact more accurately, adapt the value proposition and select the right moment to communicate.
Discover how CRM, automation, analytics and AI are enabling hyper-personalised growth.
The practical rule is straightforward. Artificial intelligence should increase relevance, not merely production.
Each workflow needs a clear reason to exist, a defined audience, a specific value proposition and exit criteria. When a lead responds, books a meeting, changes stage or stops showing interest, the automation must adapt. The CRM should also prevent contradictory, duplicated or excessive communication between teams.
An effective sequence should not be a rigid chain of messages. It should contain decisions based on behaviour.
A person researching technical information may receive an implementation guide or relevant use case. An account interacting with content about costs may receive a return on investment analysis. A lead that has already spoken with a sales professional should leave generic campaigns and begin receiving communication aligned with the opportunity.
In Account Based Marketing, this contextualisation becomes even more important. The company no longer communicates with isolated leads. It coordinates content, messages and commercial activity for different stakeholders within the same account.
Measuring this play requires more than tracking email open rates. Companies should analyse positive responses, meetings booked, opportunities created, progression speed, opt-out rates and influenced revenue.
Conversion optimisation depends on the relationship between communication and commercial progress, not on the total volume of messages produced.
5. Transform meetings and calls into actionable data
A significant proportion of the most valuable information about an opportunity remains trapped in meetings, calls and individual notes. When this knowledge does not enter the CRM, the team loses context, managers cannot assess risk accurately and forecasting depends on incomplete perceptions.
Conversation intelligence tools can transcribe interactions, produce summaries and identify objections, competitors, commitments, next steps, dates and stakeholders. Automation can then update fields, create tasks and suggest content or actions.
However, an automatically generated summary should not replace the sales professional’s responsibility for the opportunity. Its purpose is to reduce administrative work and make commercial records more consistent.
This use case improves conversion in three ways.
First, it ensures that the next steps are clearly documented and connected to a date.
Second, it allows the organisation to identify patterns among opportunities that progress and those that remain blocked.
Third, it creates material for coaching based on actual conversations rather than occasional evaluations.
Salesforce’s State of Sales 2026 indicates that high performing teams are 1.4 times more likely to use AI agents for sales coaching.
The analysis of sales conversations can also reveal recurring objections. When several opportunities are blocked by concerns about integration, pricing, security or implementation timescales, the problem may not be limited to the sales professional’s performance.
There may be a weakness in the value proposition, documentation, Marketing content or even the product itself. Information collected through the CRM should support decisions across Marketing, Sales, Product and management.
The company must define what data may be collected, who may access it, how long it will be retained and how the applicable privacy and consent requirements will be respected. Trust disappears quickly when technology is introduced without transparency.
Automation should support the commercial process. It should not create unnecessary surveillance.
Relevant metrics include the percentage of meetings recorded, opportunities with a defined next step, time between the meeting and follow-up, the most frequent objections, progression by sales professional and the effect of coaching on close rates.
The value appears when the information collected changes decisions and behaviour.
6. Predict pipeline risk and recommend the next best action
A sales forecast based only on the opportunity stage tends to be both optimistic and late. Two opportunities in the same stage may have very different probabilities of closing.
One may involve several decision makers, a recent meeting and an approved proposal. The other may have had no activity for several weeks and may depend on a single contact without decision making authority.
Artificial intelligence can analyse historical performance, time spent in each stage, frequency of interaction, contact seniority, conversation sentiment, changes in value, commitments, competitor presence and next steps to identify risk.
Instead of presenting only an abstract probability, the system should explain the signals and recommend an action. This may involve engaging another stakeholder, clarifying a requirement, booking a meeting, revising the proposal or closing an opportunity that is no longer progressing.
This play increases sales efficiency because it reduces time spent on deals with a low probability of success and helps managers intervene earlier. It also improves forecasting, resource allocation and the quality of pipeline review meetings.
A recommended next action can be particularly useful when the number of opportunities exceeds the team’s follow-up capacity. The system can highlight deals with no recent activity, opportunities that have remained in a stage for longer than usual, proposals without a response and accounts where a key decision maker has not yet been involved.
Prioritisation no longer depends solely on the sales professional’s memory or the order in which deals appear in the pipeline.
The risk is treating a prediction as a certainty. Models trained on historical data may fail when the offer, market, team or commercial process changes.
For this reason, recommendations should be treated as decision support. The manager remains responsible for testing assumptions, challenging incomplete information and distinguishing a genuine signal from a correlation with no operational value.
The relevant metrics include forecast accuracy, the difference between the predicted and actual close date, the percentage of pipeline without activity, average time per stage, postponement rate, win rate and pipeline coverage.
A strong forecasting system does not hide uncertainty. It makes uncertainty visible and helps the team act on it.
Forecasting must also be connected to an analytics structure capable of examining campaigns, sources, segments, teams, products, sales cycles and revenue. Without this analytical layer, the company sees numbers but does not understand the factors behind them.
Explore the CRM metrics and dashboards required to analyse sales pipeline in 2026.
7. Create a closed loop between Marketing, Sales and revenue
The final play connects all the previous ones. Many companies manage campaigns, leads and opportunities as separate processes.
Marketing measures contacts and engagement. Sales measures meetings and closed deals. Management receives reports that do not clearly explain which activities created revenue or where commercial potential was lost.
A closed loop begins by defining shared stages from the first interaction through to the sale, renewal or expansion. Each stage should have entry criteria, exit criteria, a responsible owner and a service level agreement.
Campaign and source information must follow the lead through to the opportunity. Disqualification and loss reasons should be structured. Value, product, segment, cycle length and margin should be recorded in a way that supports consistent analysis.
Artificial intelligence can identify patterns that manual analysis would struggle to detect. It may reveal campaigns that generate many leads but little pipeline, segments with longer sales cycles but greater value, content associated with won opportunities, points of abandonment and differences between teams.
Automation can then adjust audiences, cadences, distribution rules, messages and investment. At this point, Marketing Automation platforms stop operating as campaign execution tools and become part of a commercial learning system.
McKinsey argues that the opportunity is not limited to automating Marketing tasks. It lies in redesigning a continuous growth engine that connects insight, creation, personalisation, orchestration and execution. Read McKinsey’s analysis of continuous growth with AI.
This perspective is especially relevant for SMEs and mid-market organisations, where limited resources require clear priorities.
It is not necessary to automate everything. Companies should automate the points where time, data or opportunities are being lost, and then measure the effect on pipeline.
The closed loop must also return information to Marketing. When an opportunity is lost, the recorded reason should contribute to better segmentation, content, positioning and qualification criteria.
When a deal is won, the company should understand which interactions, messages, channels and content assets were present throughout the journey. Without this connection, Marketing continues to optimise for leads and Sales continues to operate without knowing which types of demand deserve further investment.
The principal indicators are pipeline created by source, cost per opportunity, conversion rate between stages, sales cycle length, average deal value, win rate, influenced revenue, productivity by sales professional and return on Marketing investment.
The analysis should answer three questions: where is pipeline being created, where is it being lost and which change is most likely to improve the outcome?
Marketing and Sales cannot scale as separate departments. Alignment should include shared definitions of a qualified lead, handover rules, required information, response times, follow-up expectations, recycling processes and structured feedback about lead quality.
Explore seven practical ways to scale Marketing and Sales efficiently.
How can AI improve conversion and sales efficiency?
Artificial intelligence improves conversion when it helps the team make better decisions at each point in the commercial process. This means selecting the right accounts, responding faster, adapting the message, preparing meetings more effectively, detecting risk and recommending the next action.
It improves efficiency when it automates data collection, research, activity logging, distribution, task creation, summaries and reporting.
However, efficiency without effectiveness can be damaging.
A company may automate a weak sequence and send more irrelevant messages. It may accelerate lead distribution without defining qualification criteria. It may create sophisticated forecasts based on an inaccurately maintained pipeline. It may also produce dozens of reports without identifying the activities that genuinely influence revenue.
The correct order is strategy, process, data, integration, automation and artificial intelligence.
Reversing this sequence turns technology into an expensive layer placed over existing problems.
Adoption should not be treated as a training session delivered at the end of a project. Microsoft’s Work Trend Index 2026 found that organisational factors such as culture, management support and talent practices explain more than twice the reported impact of AI when compared with isolated individual effort. Read Microsoft’s Work Trend Index 2026.
Implementation therefore requires clear ownership, usage rules, monitoring, performance reviews and continuous improvement.
Users should understand what the automation does, which data it uses, when they should accept or reject a recommendation and how errors should be reported. Without this context, the team may ignore the functionality or place excessive trust in it.
Neither behaviour produces a mature commercial operation.
Which partners can help scale Marketing and Sales with AI and CRM?
Partners that can genuinely help scale Marketing and Sales through AI and CRM do not begin with the tool. They begin with growth objectives, the commercial process, available data and the bottlenecks preventing teams from producing consistent results.
They should be capable of evaluating CRM, automation, integrations, reporting, artificial intelligence, adoption and governance as components of the same system.
They should also maintain sufficient independence to recommend the solution that best fits the organisation, rather than forcing every problem into one platform.
In some cases, the priority will be correcting the existing CRM. In others, it will be integrating the ERP, reorganising the funnel, implementing lead scoring, automating nurturing, creating dashboards or preparing AI agents with controlled access to the correct data.
A specialist partner should also distinguish configuration from transformation.
Configuring properties, pipelines, workflows and dashboards is necessary, but it does not guarantee results. The company’s business objectives must be translated into processes. Responsibilities need to be defined. Scenarios must be tested. Data needs to be prepared. Users require training and adoption must be monitored.
The quality of the partner should be measured by its ability to convert technology into processes that teams actually use and results that can be observed in pipeline performance.
This evaluation is particularly important when comparing platforms such as HubSpot and Zoho, as implementation success depends as much on the partner’s capabilities as it does on the features of the software.
Platform selection should consider process complexity, number of users, integration needs, reporting model, automation requirements, internal capabilities and the total cost of future development.
Not every company needs the platform with the broadest feature set. Every company does, however, need a system capable of supporting its operating model.
For a more detailed assessment of the platforms, explore Liminal’s dedicated pages for HubSpot and Zoho.
Grow B2B pipeline with Liminal
Liminal helps B2B companies transform CRM, Marketing Automation, data and artificial intelligence into an integrated growth engine.
The work begins with strategy and process design because a technically correct implementation can still fail when it does not reflect commercial reality, is not adopted by the team or does not produce information that management can use.
As a consultancy specialising in MarTech, CRM, automation, integrations and analytics, Liminal provides an impartial assessment of the platforms and architectures that best fit each organisation.
The intervention may include CRM selection and implementation, integration with existing systems, redesign of Marketing and Sales processes, the creation of automation workflows, qualification models, dashboards, AI use cases and adoption plans.
Explore Liminal’s approach to CRM and Marketing Automation implementation.
This approach can be applied to new implementations and to technology ecosystems that already exist but are not delivering the expected return.
Rather than automatically replacing the technology, Liminal identifies failures in data, processes, integration, usage and measurement. Priorities are then defined according to their expected impact on pipeline, conversion and commercial efficiency.
Results depend on the ability to connect strategy, technology and execution. In the GoContact project, the use of CRM and Marketing Automation contributed to doubling the number of customers generated by Marketing and reducing cost per lead by 50%.
The objective is not to add more technology to the stack. It is to create a simpler, measurable and scalable operation in which Marketing generates higher quality demand, Sales concentrates its effort on the right opportunities and management understands what is creating pipeline and revenue.
In 2026, the advantage does not necessarily belong to the company using the greatest number of AI tools. It belongs to the company capable of connecting strategy, data, processes and people within a system that continuously learns and improves.
This is precisely where Liminal’s integrated experience across Marketing, Sales and technology becomes decisive.
Talk to Liminal about building a more scalable CRM, automation and AI ecosystem.