Key takeaways

  • Automate the review work, including opportunity reviews, board reviews, and the quarterly and monthly business reviews. Tran expects this to make the forecast more math than art.
  • Automations only pay off on top of a defined process, agreed rules of engagement, and shared systems. Without those, AI can add noise.
  • A weekly operating rhythm and one forecast language let every manager and rep read the same numbers the same way.
  • Use MEDDIC to test each deal, and coach from shared data about last week's calls.
  • Keep people responsible for the right questions, for reading seller motivation, and for the final edit on anything a buyer will read.

As a chief revenue officer (CRO), you run the week from a forecast call built out of spreadsheets, and your board and quarterly business review (QBR) decks get rebuilt by hand every cycle. On top of that, automations are showing up in your customer relationship management (CRM) system, call recorder, and inbox. Without a defined process underneath, they can add noise. This article covers which automations matter most to a CRO and why they only work when an operating rhythm, agreed rules of engagement, and shared data sit underneath them. It also covers which work should stay with people. It draws on a Revenue Mavericks conversation with Quoc Tran and one with Lesley Young. The guests share their experience and opinions, not measured results, so where this article goes beyond what they said, it is our own inference.

Start with process, rules of engagement, and systems

Quoc Tran, VP of Revenue Operations at Replit, joined Revenue Mavericks and described building a go-to-market motion from scratch.

We didn't have rules of engagement. We didn't have systems. When you're layering on an AI layer without processes, you get a bunch of garbage. You get a bunch of noise.

Quoc Tran, Revenue Mavericks

An automation scores deals against stage definitions that each rep reads differently, or it summarizes a CRM where half the fields are blank. The output looks confident but cannot be trusted, so someone spends the forecast call checking the data instead of discussing the deals.

Tran described the other side on the show, in a quote lightly edited here to remove a repeated word.

Once you have processes in place and you build, you know, that AI layer on top of that, that's just like fuel on the fire.

Quoc Tran, Revenue Mavericks

Start the first automation project by writing down the stages, definitions, field owners, and the conditions that move a deal forward. Once those rules exist, revenue orchestration connects the later automations.

Automate the reviews so the forecast is more math than art

Tran offered a view about where RevOps is heading over roughly the next five years. It is his opinion about the future, not a description of what tools do today.

As you look at the orchestration of AI and the ability to do a forecast that is a lot more math than art. If you look at opportunity management and board reviews and QBRs and MBRs, all of that should be automated.

Quoc Tran, Revenue Mavericks

MBRs are monthly business reviews, and they share a problem with QBRs. Both are often rebuilt by hand from the same pipeline data, and the numbers can drift between versions.

In our view, the review work most likely to suit automation is the repetitive assembly. That includes pulling the changes in pipeline since the last review and flagging deals whose close dates keep slipping. It also includes producing the opportunity review and the QBR from the same source as the forecast. When one dataset feeds all of them, the forecast can become a calculation the CRO can inspect instead of a number to defend. That also helps with building a forecast your CFO will trust, because board reviews go better when the deck and the forecast call no longer disagree.

Asking the right questions at the right time. Sellers will sometimes try to get around the systems to make a buck, so leaders should police them but also understand their motivations. A model can flag a deal that looks off, but a person has to decide why it looks off and what to do about it.

Run a weekly operating rhythm with one forecast language

Lesley Young, who has led sales teams and has recently done a lot of fractional work, described her version of an operating rhythm on Revenue Mavericks as a syllabus.

You get your syllabus, which is sort of like, this is how we operate week to week.

Lesley Young, Revenue Mavericks

In her example, the team reviews slipping deals in week one and locks the forecast in the middle of month two, and each person's job is to build toward the forecast they gave. She presented this as what works for her, not as a cadence every team should copy.

The rhythm needs to be consistent, and it covers the review meetings and the language used in them.

That includes the forecast, that includes the QBRs, that includes how we talk about a forecast. We owe it to them to do that.

Lesley Young, Revenue Mavericks

People deserve clarity about what happens next.

A shared forecast language gives commit, best case, and each stage one meaning, so numbers from different teams add up. Once that exists, the forecasting process can be scheduled and automated, with prompts before each deadline, a locked snapshot at the cutoff, and a comparison against what was submitted.

If those insights are built off of an inconsistent operating model, they're not going to enable you to take advantage of what AI has to offer to go to market, at least that's my very strong opinion.

Lesley Young, Revenue Mavericks

Use MEDDIC to test deals

A common expansion of MEDDIC, from general industry usage and not from the interview, is Metrics, Economic buyer, Decision criteria, Decision process, Identify pain (sometimes Implicate the pain), and Champion. Young was direct about how she thinks it should be used.

MEDDIC is not a forecast tool, it's, in my opinion, it's not a methodology. It's how we test a deal.

Lesley Young, Revenue Mavericks

She learned the distinction from a lost deal. Early in her career, as an account executive, she lost a deal that seemed won. A buying committee had praised her team's demo, and she believed she had a champion. But the champion lacked influence with the executives. She thought it was an incumbent vendor that held the relationship at that level and overrode the committee, though she was not certain. The loss pushed her to start tracking what she knew and did not know in a spreadsheet.

That spreadsheet is the manual version of an automation worth building. An automated deal test checks each open opportunity against the criteria, using what was said on calls and in email, and flags what is missing. Checks include whether the economic buyer is identified and whether the champion has access to that person. Those checks go to the rep and the manager as questions to resolve, and they do not change the forecast number on their own. Conversation intelligence can help with the test, since much of the evidence sits in calls and email, not only in CRM fields.

Coach from shared data

Young described a coaching habit she calls leading with data.

How do you take the emotion out, but provide people with feedback that they can activate? What is the data we're using?

Lesley Young, Revenue Mavericks

She gave an example of a rep who believed she was a poor salesperson. Young showed her a tracked history of what she had forecast at the start of each quarter against what she delivered, and the rep saw that she consistently landed close to her number. At that time the tracking lived in a spreadsheet.

Managers cannot sit in on every call, and a Monday recap reflects what the rep remembers. Shared data changes the conversation, because the manager and the rep look at the same call evidence and the same forecast history. Automation can help prepare that view and surface moments worth discussing, so managers can coach reps without sitting in on every call. At Terret, our AI Agents coach reps, and the manager still decides what to raise.

The conversation itself stays with the manager. In Young's telling, the manager's conversation carried the message. Our own inference is that a data summary alone would not have given the rep the same confidence.

Use AI for research and prep, not final emails

Tran gave an example of where the line sits. A field engineer at Replit came up and shared an example of an email he had received.

You can tell that it was AI slop. How do we use AI tools to research?

Quoc Tran, Revenue Mavericks

He forwarded an AI-generated email to his marketing and sales team.

Research and prep are where AI helps without much risk. It can summarize an account, pull the key points about a contact, and assemble the background for a call or a review. The seller then decides what matters and writes the message.

You can't replace the human touch of, you know, reviewing it.

Quoc Tran, Revenue Mavericks

For a CRO, the rule is to automate what comes before the message and keep a person responsible for the message a buyer receives.

How we approach answer-to-action

The automations above share a data problem. Revenue data lives in fragments across the CRM, email, calls, and the data warehouse, and each source needs a different language to access. No single AI sees the whole picture, so its answers stay incomplete. A data lake leaves insight disconnected from execution. We built Terret Nexus as an answer-to-action engine to solve that problem. You ask the question and get a complete answer. We then turn that answer into action when it is needed.

Our Revenue Graph reasons across structured and unstructured data from every system. It captures the full context of the selling environment: who is buying, what business problem they are trying to solve, and how the deal is actually moving, not only the records stored in each system. Our AI Architects analyze it to deliver the answers and to design the GTM system, including sales processes, playbooks, and competitive strategies. For AI agent orchestration, our AI Agents execute what the Architects design. They deploy workflows, coach reps, score deals, and generate forecasts. Your sales team decides what to do with what the Architects and Agents produce.

Frequently asked questions about CRO automation

Which automation should a CRO start with?

Start with review preparation. Opportunity reviews, the QBR, and the weekly forecast roll-up often pull from the same pipeline data, and rebuilding them by hand is one place where numbers can drift between versions. Before you automate them, write down the stage definitions, the forecast categories, and who owns each field, because the automation will read from those.

Why do automations in the CRM and inbox add noise?

Often because the process underneath them is not defined. On Revenue Mavericks, Quoc Tran described layering AI on top of missing processes as a bunch of garbage and a bunch of noise. If reps read stage definitions differently or fields are blank, an automation will report on that mess with confidence.

What should be agreed before we automate anything?

Agree on the stages and what must be true to move a deal forward, the meaning of each forecast category, the owner of each CRM field, and the weekly rhythm for reviews and forecast lock. These are the rules of engagement. Without them, two managers can submit the same word and mean different things.

Will automation make our forecast more accurate?

It can help, mostly by removing manual rebuilds and keeping every review on one dataset. Neither guest offered measured evidence on accuracy, and accuracy still depends on shared definitions and on honest inputs from the field. Teams that want to go further can work on how to improve forecast accuracy over time.

Should we use MEDDIC to produce the forecast number?

Lesley Young said on Revenue Mavericks that, in her opinion, MEDDIC is not a forecast tool and is how a team tests a deal. Following that view, an automated check can flag missing items such as an unidentified economic buyer and send them to the rep and manager as questions. Teams that want a model-based view can add deal scoring and keep it separate from the deal test.

What work should stay with people?

People should keep the judgment work, including asking the right questions, reading why a seller is working around a system, coaching the person, and editing anything a buyer will read. Automation can help prepare the data, but a person decides what to do with it. This is close to what RevOps owns in most organizations.

Is it safe to let AI write outbound emails?

It is safer to let AI handle research and prep and keep a person on the message. Tran described an unpersonalized email as AI slop. A person still has to review what a buyer will read. A sound rule is to automate account summaries and call background, and have the seller write or edit what the buyer receives.

Plan the key automations for your revenue team

Run the review work from one dataset, with a rhythm and definitions the whole team shares. The 2-day POC below is a concrete first step.

If you want to talk through which of these automations fit your current process, and what would need to be in place first, we can talk it through. Request a demo and ask us about the 2-day POC.