The CRO is halfway through the forecast slide when the CFO stops him. “Last quarter you said 68 percent confidence on this same number. We closed at 51. What changed in how you’re calling it?” Nobody answers well. There’s a Clari instance running, a Gong seat for every AE, and a 6sense contract that renewed in March. Three of the best products in the category, all live, and the room still can’t agree on what’s real.

We see this a lot. The tools are not the problem. The problem is that “revenue intelligence” got sold as a product when it is actually a practice, and the gap between those two things is where most mid-market and enterprise GTM teams lose a quarter.

The term got stretched until it stopped meaning anything

Ask five vendors what revenue intelligence is and you get five answers that are all sincerely held and all incomplete.

Conversation intelligence is what Gong does. It records calls, transcribes them, and surfaces patterns in what buyers say and what reps say back. Real value, especially for coaching and for catching the moment a competitor’s name enters a deal.

Forecast intelligence is what Clari does. It scores deals, tracks pipeline movement between snapshots, and gives you a defensible call number instead of a rep’s gut. Also real value, particularly if your board is asking harder questions than it used to.

Account intelligence is what 6sense and Demandbase do. It tells you which accounts are in a buying window based on behavior across the web, on your site, and in your own systems.

Three different things. Three different data models. Three different questions being answered. A team that buys one of these and starts calling it revenue intelligence has bought a partial view and given it a total name, which is worse than having nothing, because now there’s a dashboard everyone trusts.

Revenue intelligence, done properly, is the connected view. What buyers said, what accounts are doing, what deals are actually worth, all reconciled against the same account object. That is the Unified Funnel argument in a different costume. It is also considerably harder than a procurement decision.

The constraint is almost never the software

Here is the part vendors will not put on a slide. Every one of these platforms is a reporting layer on top of your data capture. If capture is bad, the output is confidently wrong.

Go look at your SFDC instance right now and count how many open opportunities have a next step field that has not been touched in three weeks. Count how many contacts on a six-figure deal are actually logged as contacts, with roles, versus how many exist only in an AE’s inbox. Count the calls that happened on a mobile phone and never entered Gong at all.

We worked with a company doing about 40 million in ARR that had 78 percent of its closed-won deals showing fewer than four contacts attached. Their real buying groups were nine to twelve people. So every model built on that data, including the attribution model that justified the marketing budget, was reasoning about a third of the buying committee and missing the rest. The forecast tool did exactly what it was designed to do with the inputs it had.

Data capture discipline is unglamorous. Field hygiene, activity logging enforcement, LeanData routing rules that do not silently drop records, and someone in RevOps who owns all of it and has the authority to reject a request that would break it. Nobody wants to fund that project. It has no logo. It is also the entire dependency chain for everything above it.

If you are evaluating a platform right now and you have not audited capture first, you are buying a mirror.

Buyer signals and intent data are not the same thing, and the difference is expensive

This one causes more wasted SDR hours than anything else we see.

Third party intent data tells you that someone at a company with a matching IP range or a matching cookie pool read content about a category. It is aggregated, it is probabilistic, and it is sold to your competitors at the same time. The vendor’s own surge model is doing a lot of interpretive work between the raw signal and the account score that lands in your SDR’s queue.

We would push back hard on treating that as a buying signal. It is a topic signal. Somebody in a 4,000 person company looked at something. Could be your buyer. Could be an intern doing research for a blog post. Could be a competitor.

First party buyer signals are different in kind, not just in quality. Pricing page visits from a known contact at a target account. A second stakeholder from the same account joining a demo. A security questionnaire arriving. Someone from procurement showing up on a thread. A champion changing jobs, which ZoomInfo will flag and which is arguably the strongest single signal in B2B. Those are behaviors by identified people with observable stakes.

The teams getting real lift are weighting first party signals heavily, using third party intent only as a light prioritization tiebreaker within an already defined target account list, and never letting third party alone trigger outbound. We will cover the scoring and weighting scheme separately, because how you weight these signals matters far more than how many sources you feed into them.

The uncomfortable version of this: most orgs are paying six figures for intent data to tell them something their own website analytics and CRM already knew, if anyone had bothered to join the tables.

Buying groups break the lead-based funnel, and revenue intelligence makes that visible

Your funnel probably still counts leads. MQL to SQL to opportunity, with a conversion rate at each stage and a demand gen lead who reports on volume.

Meanwhile your actual enterprise deals involve eight to fourteen people, three of whom never fill out a form, one of whom is the economic buyer and only appears in week nine, and at least one who is quietly against you and will not say so on a recorded call.

When you turn on real account level reporting, the lead-based model stops holding together. A single MQL from a 12,000 person enterprise gets the same weight in the funnel as an MQL from a 40 person startup. Six people from one target account fill out six forms and generate six leads, which routing splits across three SDRs, who send three separate sequences to the same buying committee in the same week. Everyone in that committee now knows your process is not coordinated. That is not a data problem showing up in a dashboard. It is a customer experience problem that the dashboard finally lets you see.

The fix is structural. Score and work the account, not the person. Define the buying group per segment before you build routing rules, not after. Modeling committee roles in SFDC without turning the object model into a science project is a longer subject we will take on its own, but A Practical Account Planning Template for Sales Teams is built around exactly this shape.

Anything that treats a lead as the unit of revenue will eventually mislead you about where revenue comes from.

AI made the analysis cheap and the bad inputs louder

Every vendor in this space shipped an AI layer in the last two cycles. Some of it is genuinely useful. Automatic call summarization saves an AE real time. Deal risk scoring against historical patterns catches things a manager scanning a pipeline report would miss. Clay and Apollo have made research and enrichment workflows that took an SDR an hour now take four minutes.

But the analysis was never the bottleneck. Most GTM teams did not fail because nobody could compute a correlation. They failed because the underlying record of what happened with the customer was incomplete, and running a model over an incomplete record produces a fluent, specific, well formatted answer that is wrong in ways that are harder to spot than a blank field.

There’s a version of AI in GTM that is worth the spend, and it starts with using it to fix capture rather than to summarize what capture missed. Auto-logging activity. Reconciling contact records against actual meeting attendees. Flagging opportunities where the logged buying group does not match the people who have appeared on calls. AI GTM: What Changes When AI Runs Inside Your Go to Market Motion goes further into this, and the short version is that the boring applications are outperforming the impressive ones by a wide margin.

What changes in the QBR when this is working

The tell is not the tooling. It is what the meeting sounds like.

In a QBR without revenue intelligence, marketing presents MQL volume and cost per lead, sales presents pipeline and closed-won, and the two sets of numbers touch nowhere. Somebody says “marketing sourced” and somebody else visibly disagrees but does not want to spend the political capital. The CMO leaves with no defensible claim on revenue, which is why the marketing budget is the first thing a PE investor cuts in a downturn.

In a QBR where this works, the conversation changes shape. Marketing reports on accounts engaged in the target list, buying group coverage within those accounts, and pipeline contribution in dollars against a coverage ratio the CRO agrees with. When somebody asks why Q3 coverage is at 2.8x instead of 3.5x, there is an account level answer, not a lead volume answer. Forecast confidence intervals come with the reasons attached, because the deal scoring is visible and arguable rather than a black box number.

The board deck gets shorter. It stops being a wall of activity metrics and starts being three or four numbers with a mechanism behind each one. That is the real deliverable. Not a dashboard. A shared model of the business that survives a CFO’s questions.

We have watched teams get to that state with a fairly modest stack and teams fail to get there with 400,000 dollars of annual platform spend. The variable was never the license count.

Where this actually starts

If you are considering a revenue intelligence investment this year, we would spend the first six weeks on three things instead of a vendor evaluation. Audit what your systems currently capture versus what actually happens in a deal. Define the buying group for each of your core segments and make that definition the unit of your funnel. Decide, explicitly and in writing, which signals you will act on and which you will ignore.

Then buy the software. It will work better and you will need less of it.

We are curious where this breaks for you specifically. The pattern we see most often is that the capture audit surfaces something nobody expected, usually in routing or in how activity gets logged on multi-threaded deals, and the whole roadmap reorders around it. If you have run this and hit something different, we would like to hear it.