On October 30, 2026, MoEngage chief executive Raviteja Dodda described the company’s new Merlin Global Agent as one a marketer can hand a goal such as growing repeat purchases, after which the agent “flags a metric drop, offers to diagnose it, and suggests the next campaign.” Three days earlier, Sprinklr opened a beta of six connected marketing agents, including an Ideation Agent that can recommend adjustments to content already planned or scheduled as market conditions change. When the agent starts from the goal, the goal becomes the document a marketing team most needs to get right.
Basil C. Puglisi has argued for years that the measure comes before the money. His Factics method, introduced in 2012, pairs a fact with a tactic and a measurable outcome defined before money moves, and his October 2026 essay One Hero Ad Is Not a Test: What Meta’s AI Ad System Means for Small Budgets closed on the rule that the governing limit is the evidence threshold written before the spend. Goal-driven agents make that order harder to skip. When software proposes the campaign, the people who own the strategy still own the goal and the baseline, along with the decision about whether a result counts.
MoEngage and Sprinklr move the agent’s starting point to the goal
MoEngage presented its release at NEXT Autumn’26 as a shift in how campaigns begin. The company wrote that most campaigns still start the same way: “pick a segment, build a message, schedule it, and hope the average customer in that segment cares.” Its answer combines the Merlin agent harness with AI decisioning from Aampe, which MoEngage acquired earlier this year. In the company’s words, the workflow moves from “segment, build, schedule” to “strategize and assign,” and the marketer sets the strategy and guardrails while “the AI decides content, channel, and timing for each person.”
MoEngage keeps people in the approval chain at several points. In the shared workspace, “the marketer approves at each handoff and stays in control,” and the new SMS tools produce drafts only, so a human stays in the loop. The business context the decisioning engine reads includes the goal to optimize for, the signals that count as positive or negative, and what the AI must not do, such as frequency limits. That list is a written brief, and its first line is the goal.
Sprinklr describes a similar division of labor. Its release says the agents are designed to help teams move from live signals to recommended action faster “while keeping people in control of strategy, creativity, approvals, and brand decisions.” The Budget Optimization Agent uses campaign objectives and performance to recommend how budget is allocated before launch and adjusted while campaigns are running. Both vendors sell these products and both releases describe intended use, so a client should treat the descriptions as claims to test in a pilot.
A Sprinklr-backed study shows how few teams write the goal down first
Sprinklr published a measurement study one day after its agent launch. The report, “Show Your Work,” was produced by Metric Sherpa in partnership with Sprinklr and surveyed 807 US marketing, service and customer experience practitioners in September. Among respondents whose organizations use AI in regular live operation, 92% said it met or exceeded expectations, and only 27% said they could quantify a result precisely using records or reports.
The goal figures matter most for agent buyers. Nearly one-quarter of respondents said their organizations started deploying an AI tool without first defining what they wanted it to achieve. Only 34% define clear goals and success metrics before deployment, and 29% set baselines. Only 17% assign a named owner for AI business outcomes, and respondents with a named owner reported precise quantification nearly twice as often as those without one, 46% compared with 23%. The results are self-reported and were not independently audited, and the study comes from a vendor with a commercial interest in measurement. The pattern still deserves attention, because many teams report success without the goal or the baseline that would let them prove it.
An agent that starts from the goal needs a written goal before it runs
A team that hands a vague goal to a proactive agent gives up its strategy at the first step. “Grow repeat purchases” can justify a discount campaign or a win-back sequence, and a higher send frequency is a third option. Each choice has a different cost to margin and to the customer relationship. The agent will pick among them using whatever the brief says. If the brief says little, the agent’s choices become the strategy by default.
Puglisi treats the written goal as a checkpoint under his Checkpoint-Based Governance, because it’s the moment where a human decides what the AI is allowed to pursue. The goal statement for an agent should name the outcome metric and its baseline, the range the team expects, the actions the agent may propose and the ones it may take on its own, and the limits it can’t cross, such as discount depth or messages per customer per week. A named outcome owner signs it before the agent is switched on. Every campaign the agent proposes then goes back to that owner for a recorded accept, modify or reject.
How Puglisi would set up a goal-driven marketing agent for a client
The first step is a one-page goal brief for each agent, written before the pilot starts. It records the baseline for the outcome metric over a recent comparable period, the expected range after launch and the date of the first review. The Metric Sherpa study found that only 29% of respondents set baselines, so this step already places a client among the minority of surveyed organizations that do.
The second step separates proposals from actions. During the pilot, the agent may diagnose a drop and propose a campaign, while launches, budget shifts and changes to scheduled content wait for the owner’s decision. The owner records why each proposal was accepted, modified or rejected, and the team widens the agent’s authority only after that record shows a pattern the owner trusts.
The third step ties the review to the brief. At the review date, the owner compares the outcome metric against the baseline and the expected range, and decides whether the agent scales or stops, with changes in between. A result that clears the range on a metric nobody chose in advance doesn’t count as a win for the goal in the brief. The Metric Sherpa study found that 85% of respondents reported gains in at least one metric category their organization did not identify as tracked before launch and that habit is what this step is meant to stop.
A client can track the setup with two numbers. The first is the share of active agents with a signed goal brief that includes a baseline, and the target is all of them. The second is the accept, modify and reject split on agent proposals, reviewed monthly. A reject rate that stays high usually means the goal or the limits need rewriting, and a persistently high modify rate suggests the owner is still doing the strategy work by hand. The owner records the reasons before changing either.
The signed goal brief keeps the strategy with the people who answer for it
These agents are in beta or early release, and their results in client accounts are still unknown. The direction is clear enough to plan for now. MoEngage and Sprinklr both describe agents that read a goal, watch the metrics and suggest what to do next. A marketing team that writes the goal and the limits before the agent starts, with a baseline to measure against, keeps the strategy with a person who answers for it. An agent can draft that page, and the owner’s signature is what makes it the strategy. The first review date in the brief is the point at which the owner decides whether the agent’s choices met the goal.
Sources
- Dodda, R. (2026, October 30). Act. Decide. Introducing the next frontier of agentic engagement. MoEngage. https://www.moengage.com/blog/next-frontier-of-agentic-engagement-next-autmn-2026/
- Sprinklr. (2026, October 27). Sprinklr launches Marketing AI Agents beta to help enterprise marketers move from insight to action [Press release]. https://www.sprinklr.com/newsroom/sprinklr-launches-marketing-ai-agents-beta-to-help-enterprise-marketers-move-from-insight-to-action/
- Sprinklr. (2026, October 28). Sprinklr research: 92% of organizations claim AI delivered improved customer experience, only 27% can prove it [Press release]. https://www.sprinklr.com/newsroom/sprinklr-research-92-of-organizations-claim-ai-delivered-improved-customer-experience-only-27-can-prove-it/
- Puglisi, B. C. (2026, October 5). One hero ad is not a test: What Meta’s AI ad system means for small budgets. basilpuglisi.com. https://basilpuglisi.com/meta-ai-ads-small-budgets/
Questions readers ask
What did MoEngage announce at NEXT Autumn 2026?
MoEngage announced the Merlin Global Agent, the Merlin Agent Harness and AI decisioning from Aampe built into its platform. A marketer can give the Global Agent a goal such as growing repeat purchases, and it flags a metric drop, offers to diagnose it and suggests the next campaign.
What are Sprinklr Marketing AI Agents?
Sprinklr opened a beta on October 27, 2026, for six connected agents covering ideation, creative, campaign execution, compliance, community engagement and budget optimization. Sprinklr says people stay in control of strategy, creativity, approvals and brand decisions.
How many companies define goals before deploying AI in marketing and customer experience?
In the Metric Sherpa report produced with Sprinklr, only 34% of respondents said their organization defines clear goals and success metrics before deployment, and 29% set baselines. Nearly one-quarter started deploying an AI tool without first defining what they wanted it to achieve.
What should a goal brief for a marketing agent include?
Basil C. Puglisi recommends a one-page brief with the outcome metric, its baseline, the expected range, the actions the agent may propose and take, the limits it cannot cross and the first review date. A named outcome owner signs it before the agent runs.
How can a team tell whether a goal-driven agent is working?
Track the share of active agents with a signed goal brief that includes a baseline, and the monthly accept, modify and reject split on the agent’s proposals. At each review, compare the outcome metric against the baseline and the expected range written before launch.
#AIg
Leave a Reply