Sean McVay

Most enterprise Salesforce integrations fail silently for 6–8 weeks before anyone notices. By the time your RevOps team flags duplicate orders in the ERP, or your sales team realizes inventory lookups are stale, you've already paid the cost in missed quotes, forecast drift, and manual reconciliation. The failure wasn't technical — it was architectural. Here's what breaks: → **Sync vs async confusion** — Teams build synchronous calls to slow ERP endpoints, then wonder why Salesforce page loads take 14 seconds. Async patterns with Platform Events would've solved it, but nobody asked the question upfront. → **No External ID strategy** — Without proper field mapping to external system keys, every data sync becomes a complex transformation problem. Upsert operations fail. Duplicates multiply across Account and Contact objects. → **Flow and Apex collision** — A record-triggered Flow fires alongside an Apex trigger on the same object. No documented order of execution. Race conditions appear in production that never showed up in sandbox. → **Error handling as an afterthought** — The happy path works beautifully in dev. Then production hits: API token expires, target system goes down mid-transaction, partial failures corrupt data. No dead-letter queue. No alerting. No runbook. → **Integration tested with 10 records, deployed at 100,000** — Unit tests in a dev sandbox tell you nothing about governor limit exposure at volume. The integration works until it doesn't. The pattern is consistent: organizations treat integration as a technical task and assign it to developers without an integration architect in the room. The cost compounds annually. Workarounds multiply. Technical debt hardens. We've seen this across manufacturing, SaaS, and financial services orgs. The ones that avoid it start with an integration blueprint — not a single line of code — and design for failure modes, not just features. We broke this into a full evaluation framework internally. If your Salesforce-to-ERP integration is already live and you're not confident in the error handling layer, that's the audit worth running now.

1

I have been following the Salesforce Agentic Enterprise License Agreement (AELA) debate, and I think something important is getting lost. A unique case of push and pull happened between Salesforce and Gartner last week. Gartner warned that Salesforce's AELA might convert from unlimited access to defined quantity contracts at renewal. In fact, Hannah Decker from Gartner said customers should clarify how long the arrangement lasts and negotiate caps on renewal increases. Salesforce pushed back. Bill Patterson, their EVP, said the claim is inaccurate. Renewals remain flexible. Because AI compute costs may shift, they tailor terms so customers get maximum value from usage. Here is what I think. Both are right. Salesforce is making a bet on customer stickiness. Miguel Milano, their CRO, said something revealing in December. If a customer deploys so much that the deal is not profitable for Salesforce, that customer becomes the happiest. "And then I have another 20 years to monetize that customer." This is simply enterprise software economics when you play the long game.  👉🏼 Give unlimited access initially.   👉🏼 Let customers build dependencies.  👉🏼 Renegotiate when switching costs are high. Gartner is not wrong to flag this. But I do not think Salesforce is running a bait-and-switch either. Decker made a point that matters more than renewal terms. "Value is not necessarily being measured effectively at this point." That is the actual problem. Organizations sign up for unlimited AI without clear ways to measure what they get. At renewal, they have usage data but no value data. If Agentforce handles 84% of support cases, what is that worth? If Data Cloud processes 32 trillion records and surfaces insights that drive revenue, how do you price that? AELA bundles Agentforce, Data 360, and MuleSoft into a fixed fee for two or three years. Unlimited usage with predictable costs. Users love that. But Gartner is right about exit terms. When that period ends and your consumption exploded, what does renewal look like? Here is what I tell clients. 👉🏼 First, use unlimited access strategically. Deploy AI where you can measure outcomes, not everywhere just because you can. 👉🏼 Second, track usage obsessively. Know where AI delivers value versus where it just consumes resources. 👉🏼 Third, negotiate exit protections now. Get renewal caps in writing. 👉🏼 Fourth, use the flexibility Salesforce offers. They said renewals remain flexible. Hold them to that. Clear value delivery gives you leverage at renewal. I think Salesforce is being smart, not deceptive. If you use AELA to experiment, measure outcomes, and build real AI processes, both sides win. Salesforce monetizes a sticky customer. You renew because ROI is clear. If you consume blindly, Gartner's warning plays out. Renewal becomes painful. The pricing model is not the problem. Unclear value is. Which side of the debate are you on? #AgenticAI #AIStrategy MuleSoft Community

80

8 Comments

We’ve seen too many Salesforce projects start in the wrong place. Too many discovery sessions focus on “what features do we need?” instead of “what journey do we want to support?” When you start with features, you end up with a system that mirrors your internal departments. It works fine… until a customer moves from one team to another and the experience falls apart. When we start with the customer journey, something different happens: 1. The process design naturally cuts across silos 2. The data model supports end-to-end visibility 3. Metrics reflect outcomes, not just activity counts The truth is, Salesforce already has the flexibility to follow your customer’s path. It just needs to be set up with that path as the blueprint. That’s why we tell every client: if you can’t clearly describe your customer journey from first contact to long-term loyalty, you’re not ready to implement #Salesforce. #crm #customerjourney #customerexperience

13

1 Comment