Google is adding five words to the information attached to an ad: “How this ad was made.” TechCrunch reports that consumers can open My Ad Center from ads across Search, YouTube, and Discover to see whether the creative was produced or edited with AI. When advertisers use Google’s generative tools, the disclosure is enabled automatically. When they use other tools, the advertiser supplies it.

It is a small product change with an interesting consequence. Context about origin now travels with the creative object into the audience experience.

I keep coming back to a question that has little to do with ad disclosure. If context can travel with an ad, can business value travel with an agent’s output?

It would overstate a modest UX feature to claim Google has revealed the future of SaaS renewal. It has, however, made a useful design pattern visible. Information that once lived behind the scenes can become part of the object a person actually sees, uses, and evaluates.

The dashboard is losing its narrator

For years, SaaS usage had a human constituency. When a manager logged in, built a report, or shipped a campaign, frequent use served as a rough proxy for utility. More importantly, that person could explain why the software mattered. At renewal, the champion translated interface activity into budget language.

Agents loosen that relationship. A product can record thousands of actions because one agent is traversing APIs, generating assets, and updating systems. The work may be excellent. The reduction in manual effort may be substantial. Yet the volume of machine activity does not tell a budget owner what changed for the business.

An agent can invoke an API all quarter. It will not show up at the budget meeting.

That is the mismatch I described as Hollow Usage. Automation can increase delivered value while weakening the old evidence used to sell that value. The dashboard gets busier while the human advocate spends less time inside the product.

Human champions will not disappear from every workflow. High-stakes work remains hybrid. People set intent, make tradeoffs, and accept or reject what the system produces. Their role may simply shift from daily operator to objective owner and output judge. The software can become more useful even as its interface becomes less visible.

At renewal, a budget owner is allocating scarce capital among plausible claims. If a product can show only sessions, calls, and tasks completed, the buyer must infer the connection to business value. That inference becomes less generous when orchestration layers can route work among several models and tools.

My read is that pricing power will begin to migrate from activity a vendor can count toward outputs a customer can recognize, circulate, and defend. This is a market hypothesis, not a conclusion Google’s feature can prove. But it follows the money.

Put the renewal case inside the work

Consider an agentic campaign platform. Its account page might report assets generated, agent runs completed, and hours estimated to have been saved. Those numbers are useful for capacity planning and product operations. They do not, by themselves, show what the marketing organization chose to put into the market.

A renewal-grade campaign package would look different. It would contain the agreed objective, approved audience and spend logic, final editable creative, significant decisions made by the agent and human reviewers, launch status, and the downstream performance record. It would be the core object used to launch the campaign, brief partners, review results, and prepare the next cycle.

Portable, in this context, means the package can move across the customer’s organization without losing its reason for existence. Marketing can send it to finance. An agency can share it with a client. The next campaign team can reuse it. The commercial context remains attached.

Defensible does not mean pretending the agent caused every sale. Marketing rarely offers such clean causality. A stronger artifact preserves the objective, decisions, accepted state, and observed consequence without turning correlation into theater. It gives the budget owner something the organization can honestly stand behind.

I previously argued in Define ‘Good’ Before You Ship the Agent that teams should establish the renewal-grade artifact before deployment. The next step is commercial: treat that artifact as a possible pricing surface. It can carry the vendor’s claim on budget into the room where renewal is decided.

Google’s label puts origin context beside the creative. An agentic SaaS product can go further by keeping business context beside completed work. Workday’s Agent Passport offers a related signal from enterprise software: agents are being tied to bounded identities and operations. Once an action has a defined product boundary, its result can have one too.

A meter is not pricing power

Agentic software gives vendors a legitimate reason to meter activity. Model calls, external tools, memory, and other services consume resources. Credits, tasks, and agent runs can help protect gross margins and make variable costs legible.

A meter still describes what the supplier spent or executed. It does not automatically establish what the customer is willing to pay. A token is a cost unit. A completed run is an activity unit. If the output remains generic, pricing conversations can drift toward cost comparisons. That is where commoditization begins.

A durable business artifact opens a middle ground between seat pricing and pure outcome pricing. Seat models become less intuitive when machines perform more of the work. Pure outcome contracts can collapse into arguments about attribution. An accepted campaign package is more concrete.

Some vendors may eventually charge for accepted work products or completed business states. More will probably retain familiar contracts while using durable outputs to support premium tiers, expansion, and renewal. Either way, the value unit moves closer to work the customer chose to keep.

Portability creates a useful tension. If the customer can export the campaign package, the vendor gives up some technical lock-in. Yet an artifact that circulates through planning, finance, agencies, and subsequent campaigns makes the vendor’s contribution more visible. Forced retention and pricing power are different assets. One obstructs exit. The other sustains willingness to pay.

As AI expands the supply of adequate creative at lower marginal cost, generic generation becomes easier to compare. Pricing power may accrue to the product that consistently turns generation into campaigns the customer approves, launches, learns from, and reuses.

The valuable position is not necessarily the tool logging the most calls. It is the product responsible for the business object the budget owner recognizes.

What remains after the interface

The thing I would watch is whether SaaS companies add another outcome tab to their dashboards. That would tell us very little. Watch the product object itself. Does the campaign package retain its objective and decision context? Can the customer accept it, circulate it, and reuse it? Does it connect the agent’s work to a business consequence without overstating causality?

Then watch the contracts. A move from seats or raw consumption toward accepted work units would be a strong signal that the commercial model is changing.

Not every category will follow this path. Some work resists neat completion states, and people will remain central to judgment and acceptance. The thesis does not require every SaaS company to adopt outcome pricing. It requires recognizing that the old usage proxy loses force as the agent becomes the operator.

That should make AI-native software more ambitious, not less. Agents can execute more of the workflow while the product leaves behind a stronger object for human judgment. Faster operations and clearer commercial memory can coexist.

Google’s label is modest. An ad now carries a route to its making. An agentic product may need its output to carry a route to its value.

At renewal, the decisive question will not be whether the agent was busy. When the interface, the session, and the usage chart disappear, what remains valuable enough to renew?

Sources and further reading