SPS Commerce, Inc. (SPSC) Q3 2026 Earnings Report, Transcript and Summary

Good day, and welcome to the SPS Commerce's Third Quarter 2026 Conference Call. [Operator Instructions]. Please note this event is being recorded. I would now like to turn the conference over to Ms. Irmina Blaszczyk, Investor Relations for SPS Commerce. Please go ahead, ma'am.

Good afternoon, everyone, and thank you for joining us on SPS Commerce's Third Quarter 2026 Conference Call. We will make certain statements today, including with respect to our expected financial results, go-to-market strategy and efforts designed to increase our traction and penetration with retailers and other customers. These statements are forward-looking and involve a number of risks and uncertainties that could cause actual results to differ materially. Please note that these forward-looking statements reflect our opinions only as of the date of this call, and we undertake no obligation to publicly update or revise any forward-looking statements whether as a result of new information, future events or otherwise. Please refer to our SEC filings, specifically our Form 10-K, as well as our financial results press release for a more detailed description of the risk factors that may affect our results. These documents are available at our website, spscommerce.com, and at the SEC's website, sec.gov. In addition, we are providing a historical data sheet for easy reference on the Investor Relations section of our website, spscommerce.com. During our call today, we will discuss adjusted EBITDA financial measures and non-GAAP income per share. In our press release and our filings with the SEC, each of which is posted on our website, you will find additional disclosures regarding these non-GAAP financial measures, including reconciliations of these measures with comparable GAAP measures. And with that, I will turn the call over to Chad.

Thanks, Irmina, and good afternoon, everyone. Thank you for joining us today. Supply chain management has reached a critical inflection point across the retail ecosystem. Brands and retailers are adopting AI to transition from reactive operations to proactive execution, expecting clear and quantifiable return on their investment. While AI unlocks powerful operational efficiencies, its true potential is realized when combined with contextual supply chain data. At SPS Commerce, new agentic AI capabilities are combined with the intelligence of a network built on more than 300,000 trading relationships, nearly 3 decades of proprietary knowledge and retailer compliance rules that aren't documented anywhere else. For SPS customers, our agentic AI, called MAX, plays a central role in driving supply chain efficiencies by embedding the SPS network intelligence directly into trading partner workflows and automation systems. As a result, MAX delivers real ROI today and because it runs on network intelligence, that compounds with every new trading connection. It's continually learning to deliver ROI in the future. Building on this foundation, last month, we announced new and expanded capabilities across our product portfolio. These advancements help brands and retailers operate faster, and more proactively with greater visibility across their supply chains to achieve immediate and measurable results. MAX, our embedded agentic AI supply chain offering, is now available to all of our fulfillment customers. Grounded in real-world trading behaviors, MAX guides users through daily supply chain workflows, identifies operational anomalies and prevents costly errors before they occur. Decision Intelligence and expansion of SPS Analytics connect sell-through, inventory and supply chain signals into a unified view organized around retailers' key weekly inventory and forecasting decisions. Visibility Management connects ERP, warehouse management, transportation management and carrier systems into a single real-time purchase order status view. It surfaces potential disruptions before they impact an order or revenue automatically engaging suppliers to resolve them. While these product innovations represent the next step in our vision for the intelligent supply chain network, the true test is the tangible ROI delivered to our customers. RIP-IT Sports, a high-growth brand selling sporting equipment to major retailers like Dick's Sporting Goods, Walmart and Amazon, is leaning heavily into AI to expedite efficiency and the insight. In its first week live, MAX saved this customer $200,000 by flagging chargebacks and late receivables that the RIP-IT team couldn't see and identified a process error to eliminate future loss. MAX watches every trading partner and explains resolutions in plain language, leveraging retailer-specific compliance patterns that aren't documented elsewhere. MAX recently detected more than 200 shipment status failures between Zevia, a beverage brand, and their 3PL. In those orders, neither Zevia nor the ERP were being updated automatically with the correct workflow status. For a customer of this size, we estimate up to $3 million in receivables could have been delayed had those shipments not been realized and invoiced on time. Something Inked, a creative merchandising and custom branding company that works with major sports and entertainment brands, uses MAX to accelerate invoice searches and exception monitoring. This customer estimates using MAX will save them hundreds of hours per year and thousands of dollars in overhead. According to their project lead, MAX has become a staple in their daily operations. With MAX now live across all fulfillment customers, adoption of our advanced AI features is accelerating. Upcoming Q4 product launches keep us on track to monetize AI-powered solutions next year. SPS Commerce has also embedded agents internally to accelerate fulfillment customer onboarding from contract close to the first live transaction. Since adopting agentic onboarding, hundreds of suppliers have achieved full compliance with new trading partners within 24 hours. Agentic onboarding is also strengthening our competitive position, particularly when new customers plan to integrate their ERP system with the SPS network. In Q3, this accelerated onboarding capability was a deciding factor in securing multiple new customer wins. We are prepared to move as quickly as our customers need us to and delivering faster time to value. To date, we have completed many complex ERP integration and onboarding projects in less than 30 days and we expect to shorten these cycles even further as we advance our agentic capabilities. For example, a manufacturer of safety and fall protection gear needed 21 integrations set up and tested in 2 weeks. Accelerated onboarding with AI agents helped meet the deadline, and our validation agent flagged incorrect carrier codes and shipping serial IDs that would have caused advance shipping notice compliance issues and potential retailer chargebacks. The customer was able to address both errors before production. Similarly, the world's leading dairy company partnered with SPS Commerce following a major acquisition to rapidly onboard over 70 trading partners and avoid delays in order fulfillment of highly perishable products. With agentic onboarding, SPS accelerated testing and business rule configuration for new connections, successfully executing a high-stakes go-live in just weeks. In summary, the progress we have made to date with our agentic transformation initiatives, both across our product portfolio and internally, has exceeded my expectations. I want to recognize all of Team SPS for their agentic-first mindset and commitment to this transformation. Historically, customers have trusted SPS' vast network intelligence to deliver connectivity and compliance with their trading partners. Today, all of our transaction, compliance and proprietary data is powering our agentic AI solutions, enabling MAX to detect process disruptions and anomalies. This strong quantifiable ROI will result in direct monetization of our AI capabilities and drive a higher volume of transactions across our network as customers increasingly rely on SPS solutions as the indispensable intelligence tools they need to manage an efficient supply chain. Notably, we are moving fast while maintaining operational discipline and exceeding our margin targets. With our agentic strategy advancing, we are uniquely positioned to deliver sustainable, profitable growth backed by the combination of AI and SPS' intelligent network, a powerful and durable competitive moat that no one else can match. With that, I'll turn it over to Joe to discuss our financial results.

Thank you, Chad, and welcome, everyone. We reported a strong third quarter of 2026 as core business growth exceeded expectations while we delivered significant margin expansion. Revenue was $197.6 million, a 4% increase over Q3 of last year. Recurring revenue grew 4% year-over-year. Two factors impacting the growth rate in the quarter. One is the divestiture of the 3P Revenue Recovery business on June 30, 2026, and the second factor is the incremental impact from our 1P Amazon Take Rate business. Together, the divestiture and the Amazon Take Rate business negatively impacted revenue growth by 500 basis points in the quarter. In reference to the divestiture of the 3P Revenue Recovery business, we said in the Q2 call that we expected divestiture to reduce second half 2026 revenue by approximately $10.5 million. We continue to believe the divestiture sharpens our focus on the strategic opportunity with 1P suppliers. The 1P Amazon opportunity continues to be important to our revenue recovery growth strategy for the long term with a $750 million addressable market. In reference to the impact from the Amazon Take Rate business for the quarter, it was driven by two factors. First, we saw higher-than-anticipated conversion of take rate revenue to subscription revenue. While this has created a headwind to revenue in the second half of 2026, we expect it to benefit the business in the midterm and align with our long-term strategy for revenue recovery. Second, take rate revenue was lower than anticipated, resulting from recent Amazon policy changes that impacted our 1P take rate customers. As a reminder, under the take rate model, revenue is recognized as a percentage of the actual funds successfully recovered from Amazon suppliers. In recovery, yields are tied to the retailer policies. In the quarter, Amazon has implemented certain policy changes, which have been unfavorable to recovery yields. Our current go-to-market approach is focused on subscription model, the take rate model remains common in Amazon revenue recovery market, and allows us to capture a larger share of enterprise customers. Looking at our customer growth, the total number of recurring revenue customers in Q3 increased sequentially to 46,700 and an average revenue per customer was $16,200. The divestiture reduced our total customer count by approximately 7,300 3P customers, who were smaller than our average customer size, which positively impacted ARPU in Q3. Adjusted EBITDA increased 17% to $70.9 million or 36% of revenue, outperforming our long-term margin target of 35%. This reflects SPS' strong operational execution as we continue to unlock efficiencies across the organization. Looking ahead, we plan to provide an updated adjusted EBITDA target on our fourth quarter conference call. We expect continued margin expansion to be driven by operating leverage and increasing use of onboarding agents, which today is reflected in our gross margin expansion. Adjusted gross margin, which excludes stock-based compensation expense, was 73% in Q3 and improved 170 basis points year-over-year. Adjusted sales and marketing was 20%, G&A was 13% and R&D was 7% in Q3 and all declined year-over-year as a percentage of revenue. Turning to liquidity and cash flows. We ended the quarter with total cash, cash equivalents and short-term investments of $184 million. Free cash flow for the quarter was $58.5 million, bringing our trailing 12-month free cash flow to $202.6 million, up 37% year-over-year. In Q3 2026, we deployed 84% of free cash flow to repurchase $48.8 million of SPS shares. Now turning to guidance. For the fourth quarter of 2026, we expect revenue to be in the range of $199.3 million to $201.3 million, representing approximately 4% growth at the midpoint, which includes a negative impact from the Amazon Revenue Recovery take rate business and the divested business. Excluding these impacts, our underlying fourth quarter revenue growth is expected to be between 8% and 9%. We expect adjusted EBITDA to be in the range of $72.0 million to $74.0 million. We expect fully diluted earnings per share to be in the range of $0.74 to $0.79 with fully diluted weighted average shares outstanding of approximately 36.0 million shares. We expect non-GAAP diluted income per share to be in the range of $1.31 to $1.36, with stock-based compensation expense of approximately $17.0 million, depreciation expense of approximately $6.1 million and amortization expense of approximately $12.3 million. For the full year 2026, we expect revenue in the range of $786.9 million to $788.9 million, representing approximately 5% growth over 2025 at the midpoint of the guided range. Excluding the previous mentioned impacts, our underlying full year 2026 revenue is projected to grow high single digits. We expect adjusted EBITDA to be in the range of $267.4 million to $269.4 million, reflecting an adjusted EBITDA margin of 34% at the midpoint, an increase of approximately 330 basis points compared to full year 2025. We expect fully diluted earnings per share to be in the range of $2.29 to $2.35, with fully weighted average shares outstanding of approximately 36.6 million shares. We expect non-GAAP diluted income per share to be in the range of $5.05 to $5.11 and with stock-based compensation expense of approximately $69.5 million, depreciation expense of approximately $24.1 million and amortization expense for the year of approximately $40.8 million. For the remainder of the year, on a quarterly basis, investors should model approximately 30% effective tax rate calculated on GAAP pretax net earnings. In summary, our financial results clearly demonstrate the strong operating leverage embedded in our business model. Expanding our adjusted EBITDA margin to 36% underscores both our commitment to disciplined cost management and the tangible efficiency gains we're unlocking through agentic customer onboarding. With that, I'd like to open the call to questions.

[Operator Instructions]. And our first question for today will come from Scott Berg with Needham.

I got a couple of things here. Let's start off with the fourth quarter guidance. It looks like the actual guidance is a little bit below what was implied from the third quarter. And given your commentary on the changes at Amazon, I guess the first question is, is that the driver for all the changes with the implied fourth quarter revenue guidance? And I guess second question within the Amazon impact is -- relates to -- is this kind of a permanent change or impact to -- I assume that is that revenue recovery business. Should we think about annualizing that difference in terms of the revenue recovery there? Or does this get mitigated maybe as you move more of the customers from that transactional arrangement to more of a subscription agreement?

Yes, Scott, I'll take the first part of that. So on the revision of the guidance, Scott, 100% of that was related to the Amazon policy changes and the acceleration of moving customers from take rate to subscription. If we didn't have the impact of those changes, we would have raised guidance on the year. So that was the driver.

Yes. And I'll give you a color on the overall strategy. I mean we remain confident in revenue recovery as a product category. If you exclude the Amazon take rate portion, our revenue recovery is the fastest growing portion of our overall business. We have seen these policy change headwinds with Amazon. However, Amazon is a major portion of this ecosystem. We're still bringing on new customers, especially with multiple retailer support. And Amazon is a key component of that together typically with Walmart and Target as well. And in terms of the revenue model, we have been successful in converting some of the take rate customers. You may recall that Carbon6 was take rate only. And now some of those have converted to subscription. And in fact, that moved a little quicker in the quarter than we had expected, which long term is a positive for us. What I will say, though, is the market still is take rate, right? So there still is some take rate out there. Competitors are offering take rate. So it's not like we expect the take rate business to go to 0 over time. But I strongly believe that as we add more features consistent with our product strategy, especially on chargeback avoidance, not just on the recovery side, that's going to certainly make the product more compelling overall and customers probably more inclined to pay for that type of capability on a subscription basis.

Got it. Helpful. And then for a follow-up question, Chad, can you give us some color on what the net new customer environment was like? If we go back to the third quarter call, I know you all talked about some improving trends around enablement campaigns and I think new customers coming from some of your ERP channel partners. But I just want to understand if that environment has improved kind of as you had expected 90 days ago. And will that ultimately drive customer count ex the Carbon6 divestiture to still be in line with your expectations to be about flattish for the year from the beginning?

Yes. Sure, Scott. So on the retail programs, the retail activity in the quarter was quite strong as we expected. The activity continues here in Q4, and we're even carrying some momentum into 2027. What I would say on the results in the quarter is that the monetization of those retail programs based on the retailers they were did skew a bit more towards ARPU contribution than customer count contribution. On the ERP stuff, a little bit slower in the mid-market ERP area. I think there's room as companies sort through their AI and ERP decisions for that to pick up in the future. And overall, probably close to that sort of net neutral customer count that we had described earlier.

The next question will come from Chris Quintero with Morgan Stanley.

This is [ Beatriz ] on for Chris Quintero. Congrats on the impressive margin progress. Can you unpack where you're seeing the greatest operating leverage, whether that's AI-driven efficiencies and onboarding and customer support or broader automation or expense discipline? And how much runway remains for the further expansion?

Yes. Thanks for the question. I think there's a couple of areas that are really driving that margin improvement overall. And the first, I would say on the gross margin side, I think that's been the biggest driver of our overall improvement. And there's a couple of things going on there. Most of the success there year-to-date is not related to the agentic onboarding because that's just been more of a recent rollout that's still ramping. I think that will drive a lot more of the margin improvement over the next 12 to 18 months, which gives me a lot of confidence that there's a lot of room to run on the gross margin line. Most of the year-to-date benefits have come from the process improvements we've kind of talked about that we made over the last 12 to 15 months. The team has been really laser-focused on these processes even before the agentic onboarding. We've really been able to build out the team in a way that can scale in a much more efficient manner. So that would be the first area where we're seeing improvements on the gross margin side, driving bottom line improvements. The second area I would say would be on the G&A side. We've done -- over the last couple of years, I've talked about this last time, which is we had made some investments in the back office, upgrade some of our systems, some of our infrastructure, and we're starting to see the real economies of scale of those investments. So I would say historically, I thought G&A was running a little high based on those investments, and you're starting to see that come down. And that's a combination of getting the economies of scale on those investments, but also some of the internal use cases we're starting to see on the AI front within the organization. So it's a combination of those two things are probably the biggest driver. And to your question about how we see it going from here, we have a high level of confidence, like I said, that we can continue to drive margin improvements and a lot of the future will be driven by some of the agentic things we're working on. We've already talked about the onboarding, but we're also doing things on the go-to-market side. We're building things across the G&A team as well. So feel really good about what we've accomplished to date, but also see a lot more upside going forward.

Got it. And then can you just expand on what the specific Amazon policy changes were? Are these new policy changes from what impacted you in the past?

Yes. These policy changes related to the situation when one of our suppliers makes a short shipment to Amazon, also a process around our ability to redispute around policy, redispute a submission.

And both of them were new in the quarter.

Next question will come from Dylan Becker with William Blair.

This is Jackson Bogli on for Dylan Becker. I was just wondering if you could give us a sense of how customers are responding to MAX now that it's generally available in this quarter. And maybe how you see the impact on day-to-day workflows for customers translating into deeper engagement with the SPS platform over time?

Yes. Absolutely, Jackson. So we're seeing really positive feedback from the customers that are in MAX Chat. In fact, we're seeing quite a bit of return usage. So the product will prompt them to go in, begin usage, really high return usage in the product. And then as we're moving forward in our development plans for MAX agents, which will be the primary monetization mechanism going forward, we are already in co-development with a group of about 20 customers in the MAX agent and we expect the early adopter program for that to be expanding to about 200 customers as we turn from kind of Q4 to Q1. And a large number of those customers are heavy MAX Chat users who are very anxious to get access to the new MAX agent capability that's coming.

Great. That's super helpful. And then maybe one for Joe or Chad, as you look at over the next few years, what gives you the most confidence in the durability of the growth algorithm? And maybe within that, just any comments on capital allocation across buybacks, M&A and organic investment also that Joe touched on earlier.

Yes. Definitely highly confident in the growth algorithm going forward. I mean if you look at our results here, the core really did go strongly and actually exceeded our expectations. Also consistent with how we've been describing the growth algorithm where we expect to drive more on the ARPU side, certainly, we can modestly grow customer count, but we expect to drive most of it through the ARPU side. And the big drivers of that ARPU growth going forward are going to be the continued expansion with existing customers, primarily driven by adding more connections on the network, the upselling and cross-selling of analytics and revenue recovery. And now with the emergence of these AI capabilities on top of that, that will give us another upselling SKU there. And what I will add is the steady momentum of us to be able to serve larger and larger customers, right? So as we've done a number of things to both our product portfolio and our service delivery model, we are increasingly being able to serve larger and larger customers, which has a positive impact on ARPU.

Yes. And then on the capital allocation question, right now, we've got our buyback in place. We're buying back in the market every day, and we think that's the best use of capital. Now obviously, over the next 12 to 18 months as things change, we always keep our options open. But right now, that's going to be the main focus is buying our stock back with our capital.

Next question will come from George Kurosawa with Citi.

Maybe just to follow up on that question about growth durability. I know I'm a quarter early here, but as we think about 2027, we'll all be sharpening our pencils on modeling and you laid out the 8% to 9% exit rate on the core business. Is that maybe the right way to think about the approach for how you guys are thinking about next year or just maybe puts and takes on what might drive this business to sustain growth next year?

Yes. Thanks for the question, George. I think without getting -- we don't want to get too far ahead of 2027. So obviously, I'm not going to provide any guidance, but we believe that the momentum we're seeing on the core will continue as we see right now. We feel really confident about that going into 2027. I think there's a couple of things going on from a growth algorithm. And Chad covered it a little bit earlier, but I think the other thing that's going on is that our GRR continues to improve on a year-over-year basis, right? A lot of things we're investing in as it relates to agentic onboarding, shortening that time it takes to get on board, the things we're doing with MAX and MAX Chat, I think that's going to drive a lot of stickiness in the product. I think it's going to allow us to increase price over time. And so I think there's a couple of things going on in the core that gives us confidence in kind of the growth algorithm that Chad was talking about and the way we see 2027. And obviously, on the Q4 call, we'll provide a little bit more color on 2027.

Okay. That's great color. And then I did want to follow up on this agentic onboarding. Gross margins seem like saw some healthy expansion in the quarter. To what extent has agentic onboarding impacted Q3? I know you said it's relatively early still. And then if you just think about the number of maybe onboarding projects that have gone through this process. How has that started to trend? How do you think about in terms of the share of onboarding that can be done agentically as we think through next year?

Yes. So I think about the gross margin impact specifically from agentic onboarding, frankly, is kind of minimal. As Joe mentioned earlier, most of that's been driven over work the team has done on the onboarding process in the previous 18 months. However, the agentic onboarding has certainly helped us in the quarter with respect to new customer acquisition. We had several competitive opportunities where the customer was integrating their ERP system with their EDI provider. Those customers specifically chose SPS Commerce based on the speed at which we could onboard them to our network. We think not only will the speed of that onboarding then obviously be more efficient and drive continued gross margin improvements, but truly be a competitive differentiator. We'll probably unlock a little bit more demand in the market as well because if you're not facing a several month implementation, you're probably more likely to go forward with the project. Where we are today is that we've had hundreds of customers that are kind of on our more simple web form onboarding be done in a number of hours with agents. And then I'd say select handful of more complex ERP integrated projects be done with the agent-assisted onboarding and agent-assisted validation, but it has a really big momentum and we expect more and more projects to be done this way in Q4 and in 2027.

And George, I'll add one more piece of color there. We probably got another 12 months of road map on agentic onboarding. Like Chad said, we've started with kind of the easier customers, the web fulfillment customers. And as we move across the different ERP integrations, we're knocking those off one at a time. So what you're going to see over the next 12 months is as we roll out more and more agentic onboarding across different customers, you'll see that margin improvement over the next 12 to 18 months that will set us up for even longer-term improvement.

The next question will come from Matt VanVliet with Cantor.

I guess I wanted to dig in a little bit more on the Amazon changes, I guess, how are the economics for the customer contracts differing when they switch to subscription? How should we think about that playing out over a couple of year period? And then any other color on like what percentage of those customers have actually shifted to the subscription model so far?

Yes. So typically, a customer will have kind of a proven history of recoveries at a certain level and also see the value of the incremental product features that we're adding to help them with the analytics around chargebacks in general, other analytics around the whole recovery process. And as they see that value, then there's kind of a conversion point based on the volume of historical recoveries into a new subscription amount. We are leading with subscription in our go-to-market. We still do take new business as a take rate when competitive requirements demand that. And there's still some room as the majority of Carbon6 customers we acquired in this area were on take rate. It's still the minority of those customers that have converted to subscription, but that conversion is underway.

Okay. And then curious on any commentary you have around the overall competitive landscape and how your retailer enablement efforts are going? Any traction to speak of there and just sort of how you're viewing that business longer term?

Yes. So the retail programs continue to be a great lead source for new supplier customers into our business. Q3, in particular, we had some very large programs, as I mentioned previously, that interaction tended to be more monetized on the ARPU side. There were new customers that came in through those programs as well, but a little bit more skewed to the ARPU side. And demand for that offering where we're digitally enabling all the suppliers for these big retailers and distributors, we continue to be unique in offering that to the market. And as we continue to explain all the benefits of that in the market, seeing some upticks in demand for that solution.

Next question will come from Lachlan Brown with Rothschild & Company.

I just want to unpack that you mentioned that you would have raised the full year revenue guidance had there not been the impact of the Amazon 1P policy and revenue changes. So could you just dive into what went better than expected in the core fulfillment business in the quarter?

Yes, I think a couple of things went better than expected. I think one, Chad talked about this. These retail campaigns that were really strong coming into the quarter. Those really converted into higher ARPU. So I think our ARPU was stronger than we expected. The other area is the GRR. It was really solid in the quarter. It continues to increase on a year-over-year basis, and that came ahead of plan. And then we continue to see momentum in the subscription side of the Revenue Recovery business. And so I think the combination of those 3 things really kind of exceeded our expectations in the quarter.

And can you also update us on the campaign pipeline visibility over the next couple of quarters? I guess specifically how is the current mix tracking between your customer onboarding versus expanding connections on those existing accounts. And I guess, last year, you noted that holiday season dynamics caused some enablement campaign delays. So I just want to just double check that you're not seeing the same thing here again this year.

Yes. So the overall pipeline and activity on programs right now is strong. It will be strong in Q4, and we expect to carry some of that into '27. We are not anticipating any delays. But some of those programs here for Q4 have not actually kicked off yet. So they could be subject to delay, but we're not anticipating that. And I'd say the contribution is sometimes difficult until we get the program underway. I would just remind everyone kind of with our growth algorithm, we do expect a lower growth rate on the customer additions and a higher growth rate on the ARPU. And as the programs came in within Q3, they came in exactly consistent with that overall growth algorithm.

The next question will come from Mark Schappel with Loop Capital Markets.

Chad, could you talk a little bit about what you're seeing in early demand for Decision Intelligence and maybe how it's priced relative to the former analytics offering?

Yes, Mark. Great. Yes. So what we've done here is, we mentioned this in the last call, we completely replatformed the technology of our analytics product and as part of that went through a renaming exercise. So you'll hear us referring that to Decision Intelligence going forward that will also follow our segment reporting in the financials. A couple of things that have been real positives on this. One, the customer sentiment and the supporting retention metrics for the customers that have gone on to this new platform are more favorable than the old platform. So we're seeing positive momentum there. And then additionally, we're more competitive in new business with this new platform. And then I'd say, overall, this new platform is really necessary to set us up to execute on our road map for Decision Intelligence, which includes a number of AI capabilities, which we expect to be able to launch there in the first half of 2027. So overall, the new platform, who knows if the new name is helping. I think the new platform certainly is helping, but you'll hear us refer to that as Decision Intelligence going forward.

Great. And then Eduardo has been on board now for about a year or so, as Chief Commercial Officer in that role. What are some of the more meaningful changes you're making to sales coverage or incentives or even account ownership or what are you planning going forward there?

Yes, absolutely. So on the retail side, one of the changes we've made is really segmenting that sales force into existing customer account managers and other sales executives focused on new business. This positive momentum in the retail activity right now, I would say, that's one of many factors. But between Eduardo and Tony, who leads our retail business, they've done a nice job with those changes and those have been positive. As we look at our customer treatment strategy on the supplier side, we've really aligned customer success and sales much more closely. I think that has been beneficial to this ARPU expansion as we're driving more expansion with the customers, driving cross-selling. And as Joe mentioned, the improved retention metrics, I think, are also coming from those improved customer treatment strategies. The team has really done a nice job on the customer treatment strategy work, and that's paying off for the metrics now.

The next question will come from Parker Lane with Stifel.

This is Jack McShane on for Parker. I wanted to go back to the topic of growth now hinging a bit more on ARPU and just kind of dig deeper, like is this a temporary mix or will cross-sell and attach be the material key drivers of the business in the near -- into the midterm in '27 and beyond? And how do you weigh the opportunities in cross-sell, whether it be new connections, ancillary products like analytics and MAX?

Yes. So there's really three factors driving that ARPU. Certainly the largest one is expansion of the existing customers. It's definitely a land and expand model. We tend to get customers with one trading partner when we land them in one of those retail programs and then we expand them from there. Obviously, analytics and revenue recovery are addressable to the medium and large customers that we have. And then we'll have the MAX agent SKU on board in 2027 to be another product there. And as I mentioned earlier, as we move more and more of our capabilities with the ERP integrations, we're able to serve medium and large customers, that has a positive impact on ARPU as well. So we continue to drive new customers through the retail programs through the channel programs. But I think kind of keep in mind, a lot of this customer count every single quarter is being driven by very low revenue customers, both in the ads and both in the churn. But all of those customers that we start off through a retail program as a one TP, we have the potential to expand a portion of those up to be some of our highest value customers from there.

And Jack, just to add one, [ ranking ] them. I think the expansion of trading partners is still the biggest driver of ARPU. And then after that is the cross-sell of analytics and revenue recovery. And then next year, obviously, we'll have a little bit more to sell with the MAX agents. And so that would feed into that cross-sell opportunity.

Great. And then I'd love to hear your philosophy moving into next year as you begin monetizing MAX, how do you plan to strike the balance of broad adoption within your base and monetization that so many of your software peers have been attempting to figure out over the past year or so? And where is that AI budget coming from within your base?

Yes. Terrific question. We will favor broad adoption. I mean, we believe that this provides value and the more customers have access to it and use it, the value becomes more apparent for them, and that will allow us to take price over the relationship with the customer. So we will be charging for the MAX agents in 2027. But to your point, our programs are going to be focused on customer adoption in 2027 versus trying to maximize price before the customer really sees the value.

Your next question will come from Clark Wright with D.A. Davidson.

I wanted to touch on the agentic capabilities that you guys are bringing to market. What makes SPS Commerce uniquely positioned to provide the agentic capabilities to organizations versus what can be achieved with off-the-shelf AI offerings?

Yes. So the data really is what makes the difference here. So if you look at the underlying data that is driving our agentic use cases, first, there's all the data that the customer has in our network. And keep in mind, this is usually a superset of their supply chain data. Not all this data is making it back to their ERP systems or their warehouse management systems or transportation management systems. So there's probably the most robust data set of supply chain data a customer has is going to be on the SPS network. Second is we're seeing major trending across all the major buying organizations that run across our network. So obviously, we're not going to share customers' competitors' data, but what we can do through our agents is see certain trending with certain trading partners, certain patterns that may exist and then what's really critical to all this, and honestly, I think underappreciated out there, is the proprietary databases that we have around retailer trading expectations for their supply chain partners. So we -- and many of these are undocumented as well. So we have these robust rule data sets that dictate exactly how the supply chain is expected to be performed for each of the retailers on our network and the agents have access to all of that. So when we combine the customers' data, the trending data on the network, and these very unique supply chain rule sets we have about each major retailer, that combination is really unlocking some very powerful use cases. And if you look at just more generic AI tools out there, they would not have that same level of access or understanding.

Appreciate that commentary. And then I wanted to potentially ask you just about the policy changes. Is there anything that the company can do proactively to minimize the financial impact of new policy changes like what we have seen in recent years?

Yes. I think this strategy of adding more features in our product strategy to be about a total chargeback management and chargeback prevention solution and have less of the value proposition be just about the recoveries combined with the overall migration to subscription revenue that we discussed, I think those would be a couple of things that are mitigating factors. Overall, I will just say that we do have competitors in the market that sell Amazon on a take rate basis. When we offer that, especially in larger enterprise deals, it is helpful in winning new customers. In the quarter, we utilized our overall multi-retailer approach to win some very large enterprise retailers on to -- or suppliers on to revenue recovery. So I do think we'll need to have some type of offering to take rate at least for the next couple of years based on competitive dynamics.

[Operator Instructions] Our next question will come from Joe Vruwink with Baird.

I think of the network, it was set up to support omnichannel fulfillment. We now have the rise of personal agents factoring into the future of commerce. How do you think of that as a driver? And I guess there's multiple ways to take a customer enablement activity and actually needing to upgrade your supply chain for this as a prospect. And what does SPS need to do on the back end to participate effectively if this is going to be a new channel?

Yes. Well, first, congrats to your Brewers, Joe. Yes, great question. I think the commerce agents are going to drive kind of a new level of expectation on the supply chain as a result of the personalization and choices that are going to come from that. I think that will, just like we saw with omnichannel, increase the requirements around collaboration, the timing of collaboration as we move more and more toward these kind of instantaneous delivery expectations. It really puts a more need for a network like SPS to facilitate these supply chain transactions. I will also add that as commerce agents start to move back in the supply chain to merchandising agents and vendor management agents, these are the things that we're looking at in the midterm and I think position SPS very well to further assist retailers in how they're managing their supply chain as agentic technology moves from just the consumer back to more of the supply chain and buying operations in retailers and distributors.

Okay. That's great. And then maybe, Joe, just to be clear on some of the math. So I think if I just take the divestment evenly in 3Q and 4Q that's like a 3-point revenue impact. You're talking about 5 points in total to the 1P changes, 2 points or like $4 million. And I guess within that, how much is truly a headwind given the policy changes versus just a mechanical impact where transitioning to a sub that's worth slightly less for a period of time?

Yes. A majority of it is from the policy changes, Joe, that would be a majority of that impact. And then the conversion from take rate to subscription is a smaller portion of that.

I'm showing no further questions in the queue. Ladies and gentlemen, this concludes today's conference call. Thank you for your participation, and have a wonderful day. You may now disconnect.