It’s not often you have a conversation that makes you question whether the advice you’ve been giving for 15 years still applies. But that’s exactly what happened to me last month. In fact, it happened twice, with two different organisations, both exploring models that challenged some fairly established channel thinking.
For years, we’ve encouraged partners to move towards recurring revenue, build their services capability, and help customers adopt more of the technology they sell. There have been good reasons for all three. But these conversations left me wondering… what if that’s not true anymore? What if what got us “here”, won’t get us “there”?
1. When Adoption Becomes a Liability
The first conversation was with a software vendor that was now considering moving to consumption pricing rather than a fixed monthly fee. That surprised me a bit. Like much of the industry, I’ve spent years promoting predictable revenue, ongoing customer relationships, and a business that doesn’t have to start from zero every month. This contradicted the fixed monthly pricing we’d spent years encouraging.
When I asked why, they explained that they had incorporated AI into their subscription offering, and a fixed monthly fee was becoming harder to sustain. Here’s why.
In managed services, successful adoption creates opportunities to sell additional users, applications, or services. Costs can increase too, but a partner who has invested in automation and standardisation can support more activity without adding a corresponding amount of expense. Get the model right, and greater adoption improves profitability.
With AI, there’s another cost to consider. Many providers charge according to the information their models process and generate, measured in tokens. The more work the AI does, the higher the underlying bill becomes.
Imagine an MSP introducing an AI assistant for a fixed monthly fee. The customer finds it useful, tells their colleagues, and soon everyone is using it throughout the day. Exactly the adoption we would normally encourage. Except the MSP is now paying more to deliver a service for which it receives exactly the same revenue.
You could end up doing an excellent job of customer success and a poor job of making money.
Recurring revenue still makes sense. Consumption charges recur too. But a base fee with an included allowance, plus charges for additional usage, may be more sensible than unlimited access and a business model that depends on customers not using what they’ve bought.
2. When Services Lose Their Edge
The second conversation was with an ERP vendor who said they wanted partners who could just sell, rather than deliver services. Again, we’d spent years saying there’s little money in product sales; it’s all in services. In fact, I even wrote an article about building recurring AI services last year.
Yet here was another company bucking a well-established trend. This was their reasoning.
This organisation developed ERP technology for small and medium sized companies. Recently they had started introducing a direct offering for smaller customers, bundling software with a basic implementation service that was delivered by AI agents.
It was a standardised package with limited customisation, but the implementation time could now be reduced to 15 days, compared with the six to 12 months associated with conventional ERP projects.
Consequently, they were looking less for partners who delivered implementation services, and more for partners who could uncover and close new sales opportunities. For them, the definition of adding value wasn’t about services… it was about the sales effort.
Now if you’re a partner, you might be reading that and thinking “so the vendor keeps the services revenue, and I get a referral fee?” And that’s a fair concern.
But I think “finding a customer and clipping the ticket” understates what a good partner actually contributes. Having access to a customer is useful. Understanding that customer well enough to identify a problem they haven’t clearly articulated, ask the right questions, and help them decide what to do about it, is considerably more valuable.
Which means the economics need to reflect that. If vendors want partners to focus on selling rather than delivering, they need to reward selling in the same way they used to reward delivery. Otherwise partners will simply put their effort into selling another vendor’s offering.
3. When Success Shrinks the Renewal
These conversations also got me thinking about the LAER model (Land, Adopt, Expand, Renew). We win the customer (Land), help them use the solution successfully (Adopt), grow their spending (Expand), and retain their business (Renew).
It was the “Expand” stage that I started to reflect on.
Additional users have traditionally been a useful source of that growth. But what happens when AI helps the customer handle more work with fewer people?
For example, a customer contact centre might be able to manage more enquiries without requiring additional staff (and therefore not having to buy additional software licences). If AI takes over enough of the workload, the customer might even reduce its seat count.
The customer may have every reason to renew. They just need fewer licences when they do. From their perspective, that’s success. From the perspective of a salesperson whose target depends on licence growth, it’s a rather inconvenient definition of success.
Put that alongside the first example. Greater AI usage can increase delivery costs, while fewer human users can reduce seat revenue. Unless the commercial model captures the value elsewhere, a partner can help the customer achieve more and watch its own numbers go backwards.
4. When Certifications Miss the Point
Which brings me to partner programs. We’re comfortable counting certified engineers and measuring technical competencies. Those measures help establish whether a partner can deliver our solution.
But as AI takes on more routine implementation work, how much weight should those capabilities carry?
A customer may happily let an AI agent configure an application once they’ve decided to buy it. Getting them to that decision is another matter, especially when they’re replacing a system their business depends on. The partner who understands their concerns, has the relationship, and earns their trust still has something valuable to offer.
An exam result is easier to record than customer confidence. But ease of measurement shouldn’t determine importance. Technical certifications tell us something about whether a partner can implement our product. They tell us much less about whether customers will buy it on that partner’s advice.
Summary
These conversations haven’t convinced me to abandon 15 years of channel advice. But they have convinced me that vendors should revisit the assumptions behind their strategies and programs.
Will greater AI usage improve partner margins? Where will growth come from if customers need fewer licences? And are we rewarding the contribution we need now, or the one we found easiest to measure five years ago?
Our technology is changing what customers need from partners. It would be unfortunate if our partner programs were the last to notice.
