Sales and Customer ExperienceGrowth Transformation
Winning customers is no longer enough. Profitability is decided across the entire customer relationship
Artificial intelligence is transforming how companies sell, serve and deliver value. But it is also changing what customers need to buy, how much they are willing to pay and what they expect in return.
A company can close its best sales year ever and, at the same time, earn less from each customer than the year before. It added accounts, responded faster, implemented a CRM and deployed an AI assistant that answers around the clock. Every activity metric improved. What doesn’t always show up on the dashboard is that some of those accounts demand more hours than they pay for, that part of the contacts repeat because the problem wasn’t solved the first time, and that some customers now do on their own, with their own AI, the work they used to hire out.
Revenue and profitability aren’t the same thing. Neither are acquiring customers and creating value with them, automating and resolving, or responding fast and delivering a good experience. Artificial intelligence makes that distinction more urgent because it acts on both sides of the relationship: it changes the cost of selling, serving and delivering for the provider, and it changes what the buyer can do without them.
The question is no longer how many customers a company wins, but which customer relationships it can sustain profitably.
Key takeaways
- More customers don’t necessarily mean more value: customers differ in contribution, cost to serve and future value, and some may destroy profitability.
- The profitability of a relationship is decided before signing (qualification, pricing, scope) and after it (delivery, service, renewal).
- Automating conversations isn’t the same as solving problems. A 24/7 channel can increase customer effort if it lacks the ability to resolve.
- The strongest evidence on AI productivity concerns AI that assists people, not systems that fully replace human service.
- AI lowers the cost of executing tasks on both sides. The risk for a provider is to keep charging for what the customer can now do alone, while failing to deliver what the customer can’t or won’t do.
- In Latin America, individual use of AI exceeds the global average while business investment lags, and tolerance for a bad experience is low.
More sales don’t guarantee more value
Revenue, gross margin, contribution, customer profitability and customer lifetime value (CLV) are different measures. Revenue shows how much a customer pays. Gross margin subtracts the direct cost of what is sold. Contribution also subtracts the cost to serve: service hours, customization, scope changes, support, logistics or discounts. Customer profitability also includes the cost of acquiring the customer. And CLV estimates the expected economic value of the whole relationship, considering how long it may last and how it may evolve.
An illustrative example, with no real data, shows why those differences matter. Customer A generates USD 10,000 in revenue and its direct and service costs add up to USD 8,500: it leaves a contribution of USD 1,500. Customer B generates USD 8,000 with costs of USD 4,500: it leaves USD 3,500. The customer with higher revenue contributes less than half as much. If the organization measures commercial success by revenue, it will prioritize the wrong customer.
Hypothetical example for illustration only; it excludes acquisition cost and CLV. Source: Advisent®.
Peter Fader and Sarah Toms start from an observation most companies know but few apply: customers are heterogeneous in their value, and a customer-centric strategy isn’t about treating them all the same, but about allocating acquisition, retention and development resources according to the expected economic value of each relationship.1 It doesn’t mean serving less profitable customers poorly. It means designing experiences the company can sustain economically for each segment.
Academic research confirms that this heterogeneity has consequences. Sunil Gupta, Donald Lehmann and Jennifer Ames Stuart estimated the value of five companies from the value of their customers and found that, in their models, a 1% improvement in retention increased firm value by about 5%, compared with about 1% for an equivalent improvement in margin and 0.1% in acquisition cost.2 These are model-based estimates, with early-2000s data and few companies, but they show why retaining and developing customers can matter more than adding new ones. Denish Shah and co-authors add the uncomfortable nuance: across five companies, between 10% and 35% of customers who bought multiple product or service lines were unprofitable and accounted for 39% to 88% of losses.3 Buying more doesn’t always mean being worth more.
That is why two common shortcuts are worth avoiding. The claim that retaining a customer costs five times less than winning a new one circulates without a study that supports it with that precision, and it depends on the sector, on how costs are allocated and on the customer in question. And the 3-to-1 ratio between customer value and acquisition cost is a rule of thumb from subscription software companies, not a universal standard: the right level depends on margin, payback period, sales cycle and cost of capital. A retained customer who is structurally loss-making isn’t an economic win.
Profitability is decided before and after signing
In many service companies, the economics of an account are set before the work begins. Qualification determines whether the customer has the problem, the budget and the maturity to benefit from what is sold. Pricing and discounts set the available margin. The written scope (or its absence) defines how much extra work will be absorbed without charge. Sales promises set expectations that operations will have to meet. A salesperson who closes quickly with an aggressive discount and a vague scope can improve the quarter and worsen the next two years’ profitability.
After signing, the sale ends for the salesperson and the experience is only beginning for the organization. The handoff from sales to onboarding, delivery, service and renewal is where gaps tend to appear: information that doesn’t travel, commitments the delivery team didn’t know about, customers who have to explain twice what they had already explained. When sales, marketing and service work as separate systems, no one is responsible for the economics of the full relationship. That is the problem we examine in our article on RevOps and revenue systems.
How many of our current accounts would we sell again under the same price, scope and service conditions?
Customer experience also has a cost structure
It helps to separate terms often used as synonyms. Customer experience is the accumulated perception of all of a customer’s interactions with an organization. Customer service is the set of interactions aimed at resolving questions, needs or problems. Customer success, typical of recurring models, seeks to help the customer achieve results and sustain usage. And cost to serve is what it costs to deliver all of that to each customer. An excellent experience the company can’t afford isn’t an advantage: it’s a promise that sooner or later degrades.
In 2010, Matthew Dixon, Karen Freeman and Nicholas Toman challenged the idea that loyalty is built by constantly exceeding expectations. Based on research with more than 75,000 customers who had dealt with contact centers or self-service channels, they concluded that, in service situations, reducing the effort needed to solve a problem was more closely related to loyalty than trying to delight.4 The data describe well where effort originates: 57% of callers had first tried to resolve their issue on the website, and 56% had to re-explain their problem. The later book by Dixon, Toman and Rick DeLisi develops that idea and the Customer Effort Score.5
Two caveats keep that research from becoming dogma. Its conclusions concern service interactions, not the whole experience: in product design or in emotionally significant moments, surprising the customer can make sense. And effort isn’t the only variable that explains loyalty. What it does make clear is that effort has a double cost: it damages the relationship and makes operations more expensive, because every repeat contact, every channel switch and every escalation consumes resources.
Every contact that doesn’t resolve is both an operating cost and a relationship cost.
Responding isn’t resolving
A chatbot can answer in seconds and still increase customer effort: if it doesn’t understand the problem, if it forces the customer to repeat information, if it can’t access the system where the solution lives, if it blocks access to a person or if it answers something irrelevant. Round-the-clock availability solves the wait, not the problem. Measuring an assistant by the number of conversations or by the share of queries that never reach a person confuses deflection with resolution: a customer who abandons the chat wasn’t necessarily helped.
Customers are saying this quite clearly. In a Gartner survey of 3,566 B2B and B2C customers conducted in February and March 2026, half said their interactions are easier when companies use generative AI, and 87% considered it essential that companies using it offer the option to reach a human agent.10 The same research found customers were roughly three times more likely to use third-party generative AI tools than a company’s chatbot to resolve a service issue.11 The question for any company is uncomfortable: if customers prefer to ask their own assistant, what value does ours add?
There is academic evidence that implementation matters. Cammy Crolic and co-authors showed, with field data and four experiments, that when customers arrive angry, a humanlike chatbot lowers their satisfaction, their evaluation of the company and their purchase intention; the effect doesn’t appear with customers who aren’t angry.9 And there are visible corrections: in 2025, Klarna, which had promoted replacing agents with AI, announced it would hire people again so customers could always talk to someone, and its CEO acknowledged that the cost-focused approach had produced lower quality.23 It wasn’t an abandonment of AI, but an adjustment of the model.
In Latin America, the margin for error is smaller. According to Zendesk’s CX Trends 2025 report, 84% of Latin American consumers would switch brands after a single bad experience, compared with 58% in Europe and 51% in North America; at the same time, 73% said they don’t mind whether an AI or a person serves them, as long as the problem is solved.24 These figures come from a service software vendor and measure intentions, not behavior, but they point the same way: customers don’t demand a human channel on principle, they demand resolution. Then there is the channel. In a study commissioned by Meta from Kantar across 22 markets, including Argentina, Brazil, Colombia and Mexico, 73% of adults surveyed preferred to contact businesses through messaging and 75% said they trust a business more if they can message it.25 In the region, sales, service and after-sales often happen in the same WhatsApp conversation. An assistant that fails there doesn’t fail at support: it interrupts the whole commercial relationship.
| Model | What it does | Where it tends to work | Main risk |
|---|---|---|---|
| AI only | Answers and, if integrated, executes | Simple, frequent, low-risk queries | Deflection without resolution, frustration |
| Humans only | Full human service | Complex, emotional or high-value situations | High cost and limited capacity |
| AI assisting people | Suggests answers, summarizes, retrieves information | Teams with high turnover or varied cases | Dependence without judgment |
| AI with human escalation | Solves the simple and escalates the rest with context | High volume with mixed cases | Escalations that force customers to repeat themselves |
| Autonomous agents with oversight | Execute tasks end to end under human control | Standardized processes with integrated systems | Consequential errors and diffuse accountability |
The metrics that distinguish responding from resolving exist: first contact resolution, repeat contact rate, successful self-service resolution, human escalation rate, time to resolution, customer effort, error rate and cost per resolved case, not per conversation handled. From there they can be linked to conversion, retention and contribution.
Automating more doesn’t mean earning more
The most cited study on generative AI productivity in customer service is, in fact, a study of people assisted by AI. Erik Brynjolfsson, Danielle Li and Lindsey Raymond analyzed the staggered rollout of a conversational assistant that suggested responses to 5,172 customer support agents at one company. Productivity, measured as issues resolved per hour, rose by about 15% on average, with larger gains in speed and quality among less experienced agents and small effects among the most experienced, who even showed slight declines in quality.8 The study also recorded more polite customers and fewer requests to speak with a manager. What it doesn’t show is that an autonomous chatbot produces those results. The difference between AI that assists a person and AI that replaces them is central to interpreting any productivity promise.
Productivity, moreover, doesn’t automatically turn into profitability. In another Gartner survey of 1,303 leaders across functions, only 24% of service and support leaders had been able to demonstrate positive financial returns across their AI use cases, even though that function was, among the ten analyzed, the one devoting the largest share of its budget to AI.11 Not having demonstrated a return isn’t the same as not having one, but the figure shows how hard it is to connect operating efficiency with economic results. McKinsey found, in a survey of 440 customer care executives, that the top-performing group had AI deployed at scale far more often and reported significant improvements in customer experience more frequently than the rest.12 It is an association reported by the executives themselves, not proof of causality. And Salesforce reports that 70% of organizations with AI service agents saw measurable value within 60 days, a figure from a company that sells those agents and that doesn’t define what it means by value.13
In Latin America, the gap between adoption and results is visible. In PwC’s latest global survey of 4,454 CEOs, 56% of Latin American respondents had seen neither higher revenue nor lower costs from AI, the same share as globally.28 AI can improve productivity, shift costs from one area to another or simply generate more interactions. Which of the three happens depends on process redesign, not on the tool.
Buyers no longer depend on salespeople to get informed
In 2011, Matthew Dixon and Brent Adamson published research on thousands of salespeople that identified five sales profiles and found that, in complex sales, the profile they called the Challenger accounted for the largest share of top performers.6 Its distinctive trait was the ability to teach the customer something new about their business, tailor the message to their context and take control of the conversation. The conclusion that was most unsettling at the time was that personal relationships, on their own, didn’t explain performance. The value delivered in the conversation did. The model is especially relevant to complex sales, not to every transaction.
That conclusion is even more relevant today. A G2 study of 1,076 B2B software buyers, conducted in March 2026, found that 51% start their research with an AI chatbot more often than with a search engine, and that 69% ended up choosing a different vendor than planned because of those assistants’ recommendations.20 6sense’s 2025 report, with about 4,000 buyers, reports that 94% use large language models at some point in the buying process and reach first vendor contact about 61% of the way through it.21 Both come from companies with a commercial interest in the subject and focus on technology, but they agree that buyers arrive at the conversation with more information and a shortlist already formed.
In Latin America, the starting point is even more pronounced. According to the Microsoft AI Economy Institute, in the second half of 2025 the share of the working-age population using generative AI exceeded the global average of 16.3% in Uruguay, Colombia, Chile, Argentina, Mexico and Brazil, with values between 17% and 23%.26 Companies, by contrast, lag behind: according to the 2025 Latin American Artificial Intelligence Index, the region invests about four times less in AI than its economic weight would suggest.27 It is an asymmetry with commercial consequences: Latin American customers may come to the conversation with more AI than the provider serving them. How that decision forms outside the channels a company can observe is something we explore in our article on branding and growth.
What does our sales conversation offer that the customer can’t get by asking an AI?
When customers can do for themselves what they used to hire out
AI doesn’t only change how a provider sells. It also changes the boundary between what customers do internally and what they buy. In October 2026, Reuters reported that AT&T general counsel David McAtee had stopped paying outside law firms for many of the tasks traditionally performed by junior lawyers.15 According to the company, LegalEdge, an in-house unit combining lawyers, legal operations and AI created in January 2026, handled nearly 100 matters in its first six months, and AT&T announced a collaboration with OpenAI to develop AI-powered legal workflows in which legal decisions remain with lawyers.16 Coverage describes outside advisers as becoming focused on complex strategy and specialist expertise. No savings figures are public.
The case doesn’t prove that companies are no longer hiring law firms. It shows that, for a large company with internal capacity, the make-or-buy boundary has moved for routine tasks. Something similar appears in marketing: in Gartner’s 2025 survey of 402 CMOs, 22% said generative AI had allowed them to reduce reliance on external agencies for creativity and strategy.22
The counter-evidence is just as important. In 2025, demand for hours in the US legal market grew 2.5%, rates rose a record 7.3% and about 90% of legal spending was still billed by the hour.17 BCG reported that AI- and technology-related services already represent more than 40% of its revenue.19 And among small and mid-sized US law firms that use AI widely, 26% raised prices.18 AI isn’t reducing total spending on professional services, at least not yet. It is redistributing it.
The data point that best describes that redistribution comes from clients. In Thomson Reuters’ Future of Professionals 2026 report, with about 1,800 professionals in legal, tax, audit and compliance across 62 countries, 78% of corporate clients considered it very important or essential that their providers deliver AI-enabled quality improvements; only 6% said most of their providers already do, and 32% expect to reconsider provider relationships in the next twelve months.14 It is an intention, not an actual switch. But it suggests the risk for a provider isn’t being replaced by AI, but falling behind a client who already uses it.
For Latin America, the question has an additional dimension, because the region also exports professional services. In 2022, Argentina exported USD 2.167 billion in legal and accounting services, the largest category within professional, technical and business services.30 In Brazil, 77% of lawyers said they used AI frequently in 2026, up from 55% a year earlier, although only 34% of their organizations have a dedicated AI budget.31 If clients abroad bring routine tasks in-house, an export model based on hours of execution will be more exposed than one based on judgment and specialization. There is no evidence that this is already happening at scale; it is a strategic question, not a forecast.
To structure that decision we propose The AI-Era Make-or-Buy Matrix by Advisent®, a conceptual tool (not a validated academic model) that crosses two variables: the company’s internal capability, enhanced by AI, and the level of complexity, risk and accountability the task demands.
Advisent® conceptual framework, not a validated model. Other variables (urgency, regulation, data access, opportunity cost, scale) can move a task to another quadrant. Source: Advisent®.
Price isn’t the only variable. A company may have the technical capability to do something and still prefer to hire it out because it doesn’t want to take on the risk, because it needs someone to answer for the result, because its team’s opportunity cost is high or because regulation requires a professional signature. The reverse can also happen: tasks that used to require a provider move in-house because AI lowered the barrier to entry and the company has data and context the provider lacks.
What still deserves to be paid for
When AI lowers the cost of producing a first draft, a preliminary analysis or a comparison, the value of some services shifts. What customers still need, and are willing to pay for, tends to concentrate on what they can’t or won’t do alone: diagnosing the right problem, interpreting information in context, exercising judgment, integrating solutions with their systems and organization, implementing, managing risk and answering for the result. Information has become abundant; judgment and accountability have not.
That is Advisent®’s hypothesis, and it has limits. There are services where execution remains the main value, because scale, reliability or convenience matter more than judgment. There are customers who pay so they don’t have to deal with something, even if they could. There are markets where AI increases demand for specialists, because it creates new problems (data governance, security, regulation, integration) that didn’t exist before. And there are services where AI makes it possible to charge more, because it visibly improves quality or speed. The shift in value isn’t uniform; it is a map each company has to draw for its own portfolio.
Personalization illustrates that balance well. Mark Abraham and David Edelman frame it as an organizational capability, not a marketing tactic, and organize it around five customer promises (empower me, know me, reach me, show me, delight me) that require data, technology, processes, governance and trust.7 According to the BCG personalization index that accompanies their work, leading companies grow about ten percentage points faster per year than laggards. But personalization has a cost, and not every customer or interaction justifies it. The question isn’t how much to personalize, but where personalization creates enough value to pay for its complexity.
AI can lower the cost of delivering a service. It can also lower what customers are willing to pay for what they can now do themselves.
Products and services: two economics of experience
The distinction between products and services doesn’t coincide with the distinction between B2B and B2C: there are products and services in both markets. What changes is where the cost to serve forms and where profitability can be lost. In products, much of it plays out in comparison, availability, logistics, returns and repeat purchase. In services, in scope, team utilization, unbilled changes and renewal. In Latin America, where large companies can be up to 33 times more productive than micro-enterprises according to ECLAC, the cost to serve weighs very differently depending on the size of the company providing the service.29
| Dimension | Product companies | Service companies | Where AI acts |
|---|---|---|---|
| Discovery | Comparing attributes, prices and reviews | Reputation, specialization and referrals | Buyers compare and shortlist with their own AI |
| Conversion | Price, availability and ease of purchase | Qualification, proposal, pricing and scope | Faster proposals; risk of discounts without judgment |
| Delivery | Logistics, stock and lead times | Onboarding, team utilization and quality | Team productivity and task automation |
| Cost to serve | Returns, support and shipping | Scope changes, unbilled hours, coordination | Can fall, or just shift to another area |
| Service | Frequent, standardized queries | Ambiguous, high-context queries | Self-service for the simple; assistance for the complex |
| Retention | Repeat purchase and subscription | Renewal, expansion and cross-selling | Early detection of risk and opportunity |
| Willingness to pay | Direct comparison with alternatives | Perceived value of judgment and outcomes | Falls for tasks customers can do themselves |
Customer Growth Economics: a way to decide where to invest
To connect acquisition, experience, cost to serve, retention and value, we propose Customer Growth Economics by Advisent®: a conceptual matrix that classifies customers or segments by two measurable variables. The first is their current contribution, after subtracting the cost to serve them. The second is their expected future value: likelihood of renewal, expansion potential and ability to generate referrals. It isn’t an academically validated model; it is a tool for sales, operations and finance to discuss the same data.
Advisent® conceptual matrix, not a validated model. It requires estimating cost to serve and future value with explicit criteria; a low-contribution customer may have indirect value (referrals, learning, access to a market). Source: Advisent®.
The matrix doesn’t suggest dropping customers. It suggests that each quadrant deserves a different design of experience and investment. For customers worth developing, more sales and customer success attention. For those worth protecting, continuity and priority. For high-contribution, low-potential customers, efficient service that doesn’t erode margin. And for low-contribution, low-potential customers, an honest review of pricing, scope or channel: often the problem isn’t the customer, but a service model no one designed for them.
What is really being renegotiated
For years, commercial growth could be measured with reasonable accuracy by the number of customers and sales volume. That measure falls short when the cost to serve varies so much across customers, when automation can multiply interactions without solving problems and when buyers can do on their own part of what they used to hire out. Profitability is no longer decided only at the close: it is decided in qualification, in pricing, in delivery, in every service contact and in renewal.
Technology can reduce the cost of creating and delivering certain services. But it can also reduce what buyers are willing to pay for tasks they now perform themselves. That is why the advantage doesn’t simply depend on adopting AI, but on redefining what is sold, to whom, at what price, with what level of service, what is automated and what stays human.
Are we building profitable customer relationships, or just processing more transactions?
The value that deserves to keep being bought is the value customers can’t, won’t or shouldn’t produce alone.
Frequently asked questions about sales and customer experience
What is the difference between customer experience and customer service?
Customer experience is a person’s accumulated perception of all their interactions with an organization, from discovery to renewal. Customer service is one part of that experience: the interactions aimed at resolving questions, needs or problems. A company can offer fast service and still deliver a poor experience if the problem isn’t solved or the sales promise isn’t kept.
What is customer profitability?
Customer profitability is what a company earns from a customer after subtracting the direct cost of what it sells, the cost to serve them (service, customization, scope changes, support, logistics, discounts) and the cost of acquiring them. Two customers with the same revenue can have very different profitability, and a customer with higher revenue can leave less contribution.
What is cost to serve?
Cost to serve is the set of resources a company devotes to serving and keeping a customer beyond the cost of the product or service itself: service hours, repeat contacts, coordination, unbilled changes, returns, shipping or customization. It is one of the variables that most differentiates profitability across customers, especially in service companies.
What does customer lifetime value measure, and what doesn’t it measure?
Customer lifetime value (CLV) estimates the expected economic value of the whole relationship with a customer, considering how long it may last and how much margin it may generate. There is no single formula: different models use different assumptions about retention, margin and discounting. On its own, it also doesn’t capture a customer’s indirect value, such as referrals or access to a market.
Do AI chatbots improve customer experience?
It depends on whether they resolve or only respond. In a 2026 Gartner survey, half of customers said generative AI makes their interactions easier, but 87% considered it essential to be able to reach a person. Academic research shows that, with angry customers, a humanlike chatbot can lower satisfaction. Experience improves when the assistant can actually solve problems and hand off to people without friction.
How do you measure the return on AI in customer service?
By measuring resolution, not volume: first contact resolution, repeat contacts, successful self-service resolution, escalation rate, time to resolution, customer effort, error rate and cost per resolved case. Those indicators then need to be linked to conversion, retention and contribution, including the total cost of the technology. Deflecting a query isn’t the same as resolving it.
Will AI replace salespeople and consultants?
Available evidence doesn’t show widespread replacement. It shows a change in what customers value: with abundant information and AI assistants, buyers arrive better prepared and expect salespeople or consultants to provide diagnosis, judgment, context, implementation and accountability for results. What is most at risk is routine execution work customers can do on their own.
When should a company do something in-house with AI, and when should it hire a provider?
It depends mainly on two variables: the company’s internal capability, enhanced by AI, and the complexity, risk and accountability the task demands. Routine, low-risk tasks tend to move in-house when capability exists; high-risk or highly specialized tasks usually call for an external partner or a hybrid model. Urgency, regulation, data access and opportunity cost also matter. Price isn’t the only variable.
Sources and further reading
Academic foundations
- Fader, P., & Toms, S. E. (2018). The customer centricity playbook: Implement a winning strategy driven by customer lifetime value. Wharton School Press.
- Gupta, S., Lehmann, D. R., & Stuart, J. A. (2004). Valuing customers. Journal of Marketing Research, 41(1), 7–18. doi.org/10.1509/jmkr.41.1.7.25084
- Shah, D., Kumar, V., Qu, Y., & Chen, S. (2012). Unprofitable cross-buying: Evidence from consumer and business markets. Journal of Marketing, 76(3), 78–95. doi.org/10.1509/jm.10.0445
- Dixon, M., Freeman, K., & Toman, N. (2010). Stop trying to delight your customers. Harvard Business Review (July–August).
- Dixon, M., Toman, N., & DeLisi, R. (2013). The effortless experience: Conquering the new battleground for customer loyalty. Portfolio/Penguin.
- Dixon, M., & Adamson, B. (2011). The challenger sale: Taking control of the customer conversation. Portfolio/Penguin.
- Abraham, M., & Edelman, D. C. (2024). Personalized: Customer strategy in the age of AI. Harvard Business Review Press.
- Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at work. The Quarterly Journal of Economics, 140(2), 889–942. doi.org/10.1093/qje/qjae044
- Crolic, C., Thomaz, F., Hadi, R., & Stephen, A. T. (2022). Blame the bot: Anthropomorphism and anger in customer–chatbot interactions. Journal of Marketing, 86(1), 132–148. doi.org/10.1177/00222429211045687
Business and institutional research
- Gartner. (2026, August 4). Gartner survey finds 87% of customers say companies using GenAI for customer service must provide access to a human agent. gartner.com
- Gartner. (2026, July 8). Gartner survey finds customers are three times more likely to use third-party GenAI than company-provided chatbots for customer service. gartner.com
- Kleinstein, B., Buesing, E., Amar, J., & Haug, M. (2026, February 23). Building trust: How customer care leaders pull ahead with AI. McKinsey & Company. mckinsey.com
- Salesforce. (2026). State of service: AI agents edition. salesforce.com
- Thomson Reuters. (2026, June 22). Future of professionals 2026. thomsonreuters.com
- Randazzo, S. (2026, October 8). How AT&T is using AI to reduce its reliance on law firms. Reuters. reuters.com via tradingview.com
- Johnson, M. (2026, October 9). AT&T uses AI to reduce reliance on outside law firms. Lawyer Monthly. lawyer-monthly.com
- Thomson Reuters Institute & Georgetown Law Center on Ethics and the Legal Profession. (2026). 2026 state of the US legal market. thomsonreuters.com
- Clio. (2025). Legal trends report 2025. clio.com
- Boston Consulting Group. (2026, April 23). BCG revenue: 22nd consecutive year of growth. bcg.com
- G2. (2026, April 15). New G2 research: Half of B2B software buyers now start their research with AI chatbots. prnewswire.com
- 6sense. (2025). 2025 buyer experience report. 6sense.com
- Gartner. (2025, May 12). Gartner 2025 CMO spend survey reveals marketing budgets have flatlined at seven percent of overall company revenue. gartner.com
- Bloomberg. (2025, May 8). Klarna slows AI-driven job cuts with call for real people. bloomberg.com
Latin America
- Zendesk. (2025). CX trends 2025. Regional data reported by Expansión (2025, January 9). expansion.mx
- Meta & Kantar. (2025). State of business messaging. whatsappbusiness.com
- Microsoft AI Economy Institute. (2026). Global AI adoption in 2025. Data reported by Forbes Argentina (2026, February 26). forbesargentina.com
- CENIA & ECLAC. (2025). Latin American Artificial Intelligence Index (ILIA) 2025. Findings reported by DPL News. dplnews.com
- PwC. (2026). 29th Global CEO Survey. pwc.com
- ECLAC. (n.d.). Acerca de las microempresas y pymes. cepal.org
- Knowledge Economy Observatory. (2023). Informe n.º 27. Argentine Ministry of Economy. argentina.gob.ar
- OAB SP. (2026, March 26). Três em cada quatro advogados usam inteligência artificial no trabalho. oabsp.org.br
The AI-Era Make-or-Buy Matrix by Advisent® and Customer Growth Economics by Advisent® are proprietary conceptual developments. The concepts of customer centricity and customer value heterogeneity (Fader and Toms), customer effort (Dixon, Toman and DeLisi), the Challenger model (Dixon and Adamson) and the five promises of personalization (Abraham and Edelman) belong to their authors. Research from Salesforce, Zendesk, Meta, G2 and 6sense comes from companies with a commercial interest in the topics they study.
Last updated:
