• 2026
  • November - 14 articles
  • October - 30 articles
  • September - 24 articles
  • 2025
  • April - 19 articles
  • March - 14 articles
  • February - 16 articles
  • January - 15 articles
  • 2024
  • December - 15 articles
  • November - 15 articles
  • October - 20 articles
  • September - 17 articles
  • August - 20 articles
  • July - 18 articles
  • June - 20 articles
  • May - 22 articles
  • April - 12 articles
  • March - 14 articles
  • February - 13 articles
  • January - 11 articles
  • 2023
  • December - 12 articles
  • November - 12 articles
  • October - 16 articles
  • September - 11 articles
  • August - 13 articles
  • July - 13 articles
  • June - 13 articles
  • May - 12 articles
  • April - 11 articles
  • March - 15 articles
  • February - 12 articles
  • January - 13 articles
  • 2022
  • December - 14 articles
  • November - 12 articles
  • October - 11 articles
  • September - 12 articles
  • August - 13 articles
  • July - 13 articles
  • June - 13 articles
  • May - 12 articles
  • April - 12 articles
  • March - 14 articles
  • February - 12 articles
  • January - 13 articles
  • 2021
  • December - 15 articles
  • November - 12 articles
  • October - 14 articles
  • September - 11 articles
  • August - 15 articles
  • July - 12 articles
  • June - 14 articles
  • May - 12 articles
  • April - 14 articles
  • March - 15 articles
  • February - 11 articles
  • January - 11 articles
  • 2020
  • December - 14 articles
  • November - 11 articles
  • October - 13 articles
  • September - 11 articles
  • August - 9 articles
  • July - 11 articles
  • June - 16 articles
  • May - 13 articles
  • April - 13 articles
  • March - 17 articles
  • February - 10 articles
  • January - 12 articles
  • 2019
  • December - 12 articles
  • November - 11 articles
  • October - 12 articles
  • September - 12 articles
  • August - 14 articles
  • July - 11 articles
  • June - 12 articles
  • May - 14 articles
  • April - 12 articles
  • March - 14 articles
  • February - 14 articles
  • January - 17 articles
  • 2018
  • December - 14 articles
  • November - 13 articles
  • October - 17 articles
  • September - 14 articles
  • August - 14 articles
  • July - 19 articles
  • June - 17 articles
  • May - 18 articles
  • April - 20 articles
  • March - 18 articles
  • February - 18 articles
  • January - 19 articles
  • 2017
  • December - 19 articles
  • November - 16 articles
  • October - 19 articles
  • September - 21 articles
  • August - 22 articles
  • July - 17 articles
  • June - 19 articles
  • May - 20 articles
  • April - 18 articles
  • March - 20 articles
  • February - 13 articles
  • January - 6 articles
  • 2016
  • December - 10 articles
  • November - 9 articles
  • October - 8 articles
  • September - 10 articles
  • August - 10 articles
  • July - 8 articles
  • June - 11 articles
  • May - 8 articles
  • April - 11 articles
  • March - 11 articles
  • February - 11 articles
  • January - 9 articles
  • 2015
  • December - 13 articles
  • November - 13 articles
  • October - 14 articles
  • September - 13 articles
  • August - 11 articles
  • July - 12 articles
  • June - 14 articles
  • May - 11 articles
  • April - 12 articles
  • March - 12 articles
  • February - 12 articles
  • January - 9 articles
  • 2014
  • December - 10 articles
  • November - 9 articles
  • October - 13 articles
  • September - 12 articles
  • August - 13 articles
  • July - 14 articles
  • June - 10 articles
  • May - 14 articles
  • April - 15 articles
  • March - 17 articles
  • February - 14 articles
  • January - 18 articles
  • 2013
  • December - 20 articles
  • November - 18 articles
  • October - 21 articles
  • September - 19 articles
  • August - 21 articles
  • July - 22 articles
  • June - 20 articles
  • May - 23 articles
  • April - 26 articles
  • March - 24 articles
  • February - 29 articles
  • January - 24 articles
  • 2012
  • December - 22 articles
  • November - 24 articles
  • October - 27 articles
  • September - 27 articles
  • August - 25 articles
  • July - 22 articles
  • June - 20 articles
  • May - 28 articles
  • April - 24 articles
  • March - 28 articles
  • February - 24 articles
  • January - 24 articles
  • 2011
  • December - 24 articles
  • November - 18 articles
  • October - 21 articles
  • September - 21 articles
  • August - 21 articles
  • July - 20 articles
  • June - 23 articles
  • May - 27 articles
  • April - 22 articles
  • March - 22 articles
  • February - 16 articles
  • January - 20 articles
  • 2010
  • December - 21 articles
  • November - 18 articles
  • October - 20 articles
  • September - 13 articles
  • August - 11 articles
  • July - 9 articles
  • June - 8 articles
  • May - 9 articles
  • April - 11 articles
  • March - 12 articles
  • February - 10 articles
  • January - 10 articles
  • 2009
  • December - 11 articles
  • November - 9 articles
  • October - 11 articles
  • September - 10 articles
  • August - 10 articles
  • July - 10 articles
  • June - 10 articles
  • May - 11 articles
  • April - 13 articles
  • March - 13 articles
  • February - 7 articles
  • January - 10 articles
  • 2008
  • December - 12 articles
  • November - 8 articles
  • October - 16 articles
  • September - 11 articles
  • August - 13 articles
  • July - 13 articles
  • June - 14 articles
  • May - 13 articles
  • April - 13 articles
  • March - 9 articles
  • February - 14 articles
  • January - 11 articles
  • 2007
  • December - 11 articles
  • November - 12 articles
  • October - 12 articles
  • September - 4 articles
  • August - 4 articles
  • July - 4 articles
  • June - 2 articles
  • May - 6 articles
  • April - 5 articles
  • March - 1 article
  • Monday, November 16, 2026

    GS interviews PayNearMe's Steve Kramer on the hidden costs of failed payments

    A failed payment can lead to a second attempt, a support call or a manual review. Even when the payment eventually goes through, the work it took to collect it can add to the cost of acceptance. In this Q&A with The Green Sheet, Steve Kramer, vice president of product at PayNearMe, explains how businesses can spot recurring problems, measure the cost of resolving them and clarify who does what when a payment fails. He also discusses how AI agents can help customers complete routine payments and when disputes or exceptions need a person’s judgment.

    Green Sheet: 1. How can businesses distinguish an isolated payment failure from a deeper flaw in system design?

    Steve Kramer: Businesses should start by looking at what happened before and after the failure. A single decline tells very little about why the payment failed. Was the customer able to try another payment method? Did they understand what to do next? Did they complete the payment, or did they have to call for help?

    Patterns across failed payment attempts can reveal a larger problem. If customers repeatedly abandon the same step, contact support about the same issue or struggle to recover from a failed attempt, the business should examine the payment journey. The payment platform may be processing transactions as designed, even when the customer experience forces customers and staff to do unnecessary work.

    It's important to compare failures by payment method, channel and point in the journey, then connect those findings to repeat attempts, support contacts and eventual completion. Payment Experience Management means looking across that entire journey to understand whether the business is helping customers get to a successful outcome.

    GS: 2. Beyond lost revenue, which costs of failed or abandoned payments are most often overlooked, and how should businesses measure them?

    SK: The most overlooked costs are often the work that follows a failed payment attempt: a support call, a manual account review, an exception that operations needs to resolve or another collections contact. A payment that eventually succeeds can still be expensive to collect.

    Businesses should connect payment events with the servicing and operational work they generate. Measure how often a failed attempt leads to a support call, how much time staff spend resolving the issue and how many additional contacts customers need before completing payment. Include relevant return and dispute costs, while avoiding double counting work that appears in multiple systems.

    That creates a clearer view of the total cost of acceptance. Transaction fees are one component, but they don't capture the full cost of getting paid. A lower transaction fee may offer little benefit if the experience creates more manual work or leaves more payments unfinished.

    GS: 3. How could an AI agent help a customer move from a payment question to a completed payment?

    SK: Most payment-related calls are the same few routine questions: what do I owe, when is it due, how do I pay. None of that needs a person. An AI agent can handle these interactions, giving customers a faster, more natural way to get answers, make payments or get help without waiting on hold. To do that effectively, the agent needs access to the information and payment capabilities required to take the next step while following the business's rules.

    Consider the tolling industry. A driver gets a toll invoice in the mail and calls with a question about the amount due or how to pay. With the appropriate integrations and permissions, an agent could pull up the invoice by license plate or invoice number, confirm it is talking to the right person, explain the charges and help them pay on the call or send a secure payment link to their phone.

    A dispute requiring investigation or a decision outside the agent's authority should go to a person.

    Businesses can make it easier for customers to move from understanding what they owe to completing a payment. Lending illustrates why that matters: a question about a loan amount due can become another call or a delayed payment if the customer cannot resolve it in the same interaction.

    GS: 4. For businesses with high volumes of payment-related inquiries, such as tolling operators, how should they evaluate whether an AI agent is improving payment resolution and reducing servicing costs?

    SK: Businesses should measure whether the customer's issue was resolved and what it cost to get there. Call volume and average handling time can help explain efficiency, but neither tells you whether the customer completed the task.

    Tolling operators should track payment completion after an interaction, repeat contacts about the same invoice, transfers to staff and the time staff spend resolving those cases. It is also important to review the accuracy of the information provided and whether transfers happened at the appropriate point. A shorter call is not an improvement if the driver has to call back to resolve the same issue.

    One important measure is how effectively the agent handles frequent, repeatable inquiries that would otherwise require staff time. Automating those calls lets experienced staff spend their time on disputes and the cases that need human judgment, and lets operators keep up with growing volume with the team they have.

    Establish a baseline and compare similar types of inquiries before and after implementation. Track cost per resolved inquiry alongside the total cost of acceptance.

    These comparisons help operators determine whether the agent has eliminated work or simply shifted it elsewhere. Operators can use the findings to decide where to expand automation.

    GS: 5. When a payment fails, how should responsibilities be divided among the business, processor and financial institution to resolve it without sending customers in circles?

    SK: The business should own the customer's path to resolution, even when another party needs to investigate the underlying issue. Customers should not have to understand the payments ecosystem to know what to do next.

    The processor should provide the transaction information and technical support available to it. The financial institution should address issues within its control, such as an account restriction or an authorization decision. The business should use that information to give the customer an appropriate next step, whether that is trying another payment method, correcting information or contacting their financial institution about a specific issue.

    Businesses, processors and financial institutions should understand their responsibilities before payment failures occur. With the right information and permissions, an AI agent can help guide customers through the resolution process, but the business must still own the customer experience. If a case needs human review, the handoff should include the information already gathered so the customer does not have to start over.

    GS: 6. What should financial institutions and fintechs fix first to improve payment reliability, and which metrics show that collection rates and support costs are improving?

    SK: Start with the recurring problem that causes the most incomplete payments and unnecessary work. That might be a confusing error message, a difficult account-access process or a payment flow that offers no useful next step after a decline. Use evidence about where customers get stuck to decide which problem to fix first.

    Then measure the outcome across the journey. Look at completed payments, successful recovery after an unsuccessful attempt, time to payment and the share of payments completed through self-service. Pair those measures with repeat contacts, manual exceptions and servicing cost per completed payment.

    For collections, compare similar account groups over consistent periods so changes in the account mix do not distort the results. An increase in calls placed is an activity measure; an improvement in collections requires evidence that more of the amount due was paid. The goal is a more reliable path to payment that also reduces the work required to support it.

    GS: 7. How can agentic AI help businesses reduce the hidden costs associated with failed payments, manual servicing and customer support?

    SK: Agentic AI can reduce costs beyond transaction fees by handling repeatable work that would otherwise require staff time. The agent can answer routine payment questions, help customers make or schedule payments and give customers a direct way to pay during outreach.

    How much work the agent eliminates depends on its access to the underlying payment process. If a conversation ends with another handoff or leaves the customer unsure what to do, the business may still incur the same servicing costs. Access to payment data and business rules helps the agent take an appropriate next step within the interaction.

    Businesses should verify the benefit through fewer repeat contacts, less manual intervention and more completed payments. They should also account for the cost of operating and overseeing the agent. The useful measure is the net change in the cost to serve and collect, alongside the quality of the customer experience.

    GS: 8. As AI agents take on more responsibility for payment-related interactions, where should businesses draw the line between automation and human involvement?

    SK: Businesses should draw the line around the agent's authority and ability to resolve the issue reliably. Routine tasks with clear rules are good opportunities for automation. The agent should transfer cases to a person when customers raise disputes, exceptions require human judgment or the agent cannot verify the information needed to proceed.

    Businesses should define the agent's permissions and escalation criteria within the workflow, then regularly review how the agent performs.

    Transparency matters as well. PayNearMe's AI Servicing and Collections Agent, for example, identifies itself as a virtual assistant at the start of a call and confirms the customer's identity before sharing account or payment details. Businesses should use automation to resolve routine needs efficiently and keep people available for situations that require human judgment.

    Notice to readers: These are archived articles. Contact information, links and other details may be out of date. We regret any inconvenience.

    skyscraper ad