Most CPOs I talk to in 2026 are treating the AI shift as a soft pivot: evolving the product line, adding agentic capabilities, running a hybrid pricing model, letting old and new coexist. It is not a soft pivot. It is a cannibalization decision, and pretending otherwise is the most expensive mistake a product executive can make right now.
The two systems problem
Picture a 30M ARR SaaS business at 110% NRR that launches an agent-native version priced per outcome. Three quarters in, you have 1M in agent ARR and an exhausted team. The legacy still drives 96% of revenue, so the CRO keeps pushing seat renewals, comp is still tied to ACV, and CS is staffed for ticket volume that scales with seats. The agent product needs a different buyer, a different sales cycle, and variable revenue the CFO has never modeled. Engineering is asked to maintain the legacy and ship the new product at once, and does half of each, badly.
This ends one of two ways. The legacy wins, the new product gets quietly killed, and a startup displaces you from below. Or the legacy loses, the company eats a 30-50% revenue drop in one bad year, and the CPO gets fired for cannibalizing the business. The third way requires deciding to do it.
The seven-decision sequence
Once the call is sunset, and most of the time it is, the decisions come in this order.
Decide sunset or refresh. Set the sunset date 18-24 months out: less than 18 makes migration a fire drill, more than 24 dissolves the team's urgency. Publish it internally first. Reorg into a successor team and a smaller maintenance team that keeps the lights on but does not ship features or own roadmap. Rewrite sales comp before anything else moves: new legacy deals earn 50% of historical comp, successor deals earn 150% of equivalent ACV, legacy renewal accelerators disappear, migration accelerators appear. Pre-sell the board on the gross margin trough, walking them through the curve before it hits the financials. Communicate to customers in cohort waves, largest accounts by individual CPO and CRO call first, mid-market by structured email and webinar next, long tail by public announcement last. Track migration as one number, percent of legacy revenue migrated, in every weekly review.
The coalition and what I got wrong
The seven decisions move because a coalition forms behind them: CEO, CPO, CFO, CRO, CCO. Each has a reason to fight and a reason to support. The CRO fights hardest, so give them the outcome-pricing upside, often 2-3x the SaaS price for the same customer, and make them the hero of the new motion. Win yourself last: the legacy was the right answer for its time, the successor for the next, and loyalty to a product is not a virtue when the world has changed. The coalition takes 90 days, mostly one-on-one conversations.
I have run versions of this at Salesforce, at SOCi, and now at Smartcat. Three things I got wrong before: I waited too long to publish the sunset date internally, which preserved the legacy team's ambiguity and slowed everything. I let the CRO water down the comp plan, and three quarters of slow successor growth traced to that one decision. And I tried to make the post-sunset reorg humane by staying vague, when vagueness is not kindness. The cleanest transition I have seen ran 21 months, dipped from 30M to 24M at the trough, and recovered to 58M six months after sunset. The dirtiest ran 36 months, ended flat, and lost three executives.
Run the exercise this week: legacy line versus the successor you would build today, compare the buyers, and if the buyer is the same you have a sunset. Draft the seven decisions on one page and show it to your CFO.