Zoom now sells a forecast. So do the CRM, the engagement platform and the conversation-intelligence tool. Each is built from the signal its vendor happens to own, and each goes deaf in a different part of the deal.
On 15 September Zoom announced what it calls an AI-powered revenue operating system. Zoom Revenue Accelerator gains two capabilities: Engage, which builds multichannel sales sequences across email and phone and prioritises follow-up, and Forecast, which turns deal-level signals into a live forecast that can be drilled from company commitment down to single opportunities, configured by RevOps without engineering support and written back into the CRM. Common Room, the buyer-intelligence company whose acquisition Zoom announced in July, sits alongside as a source of product-usage, digital and community signals. A new Elite tier bundles the lot.
It was not the only forecast announced that week. Pipedrive launched meeting intelligence the following day, and Salesforce spent Dreamforce describing agents that update opportunities. The pattern matters more than any single product: the forecast has become a feature that every layer of the sales stack now ships, and each version is built from whatever data its vendor already collects.
Every forecast is built from somebody's exhaust
A CRM-based forecast is built from stage, amount, close date and the rep's commit call. An engagement platform's forecast leans on activity: emails opened, sequences completed, replies. A meeting platform's forecast is built from what was said on calls: who attended, what topics came up, how the conversation went and what happened next.
Conversation data is genuinely richer than a stage field. It is also unevenly distributed across a deal, and the unevenness is not random. The early and middle stages of a B2B deal are full of calls: discovery, demos, technical sessions, the champion walking the seller through the org chart. The late stages, where deals are won or lost, move off the recording. Procurement works over email and supplier portals. Legal exchanges redlines. Security reviews run through questionnaires. The economic buyer decides in an internal meeting the seller is not invited to.
The call recorder hears the champion's confidence. It does not hear the CFO who has not been met.
Four places a conversation-led forecast goes quiet
Talk volume. Deals with more meetings generate more signal and tend to score as healthier. Meeting volume reflects seller effort and champion enthusiasm at least as much as buyer commitment. A rep who books a lot of calls can look like a rep with a lot of momentum.
Late-stage silence. The stage with the highest consequence produces the fewest recorded conversations. How a model treats that silence is the most important design decision in it, and it is rarely documented for customers. Silence read as risk will flag healthy deals in procurement. Silence ignored will carry dying deals on stale optimism.
Motions without calls. Renewals handled by customer success, expansion driven by product usage, partner-sourced deals and low-touch inbound often have little or no recorded conversation with a seller. Common Room's product and community signals are presumably meant to cover some of that ground, which is a reasonable answer and also a second model with its own gaps.
One side of the table. Recorded calls capture what the buyer chose to say to a vendor. Sentiment analysis on those calls measures how well the conversation went, which is not the same as how the buying group feels.
Figure 1
Forecast built from | Sees well | Goes quiet | Typical failure |
|---|---|---|---|
CRM stage and rep commit | Deal structure, amount, rep judgement | Anything the rep did not enter | Sandbagging and happy ears |
Engagement activity | Response and cadence | Offline and multi-threaded buying | Activity mistaken for progress |
Recorded conversations | Discovery depth, stakeholders met, objections raised | Procurement, legal, internal buyer decisions | Talkative deals over-scored |
Buyer-side and usage signals | Expansion, churn risk, account interest | Deal-specific commitment and timing | Interest mistaken for intent to buy now |
No single source covers the whole deal. The categories are generic, not assessments of any named product; vendors increasingly blend several sources, which changes where the gaps sit rather than removing them.
Disagreement becomes the real design problem
Once two or three systems each produce a forecast, they will disagree, and the disagreement is more useful than any of the numbers. A deal the rep has committed and the conversation model scores as at risk is exactly the deal a pipeline review should spend its time on.
That only works if someone has decided in advance what happens next. Which system's number goes to the board? Which one the manager inspects? What does the rep have to do when the model disagrees with them? Most sales organisations have not written that rule, and a vendor forecast that writes back to the CRM will write one by default. If the model's view overwrites the forecast category field, the CRM stops recording what the rep believes and starts echoing what the model inferred, and the comparison that made the second forecast valuable disappears.
Questions to settle before switching a vendor forecast on
Measure coverage first.What share of last year's closed-won deals had a recorded call in their final 30 days? That number is the ceiling on what a conversation-led forecast can see late in the cycle.
List the motions with no calls.Renewals, expansion, partner and inbound. Decide which forecast owns each.
Ask how silence is treated.Get the answer in writing. It will tell you more about the model than any accuracy claim.
Protect the rep's field.Keep the rep commit and the model score in separate fields. Write-back should add, not overwrite.
Backtest before trusting.Run the model against the last four quarters and look at which deals it got wrong, not its aggregate error.
Zoom's announcement cites a customer reporting 15% more closed deals and faster deal velocity using Revenue Accelerator. Those are claims about the selling tools, not about the new Forecast capability. No vendor in this category has yet published forecast accuracy against a holdout period in a form buyers can check. Until one does, the honest way to judge any of these forecasts is the old one: look at how it was wrong last quarter, and whether anyone would have acted differently if they had believed it.
Sources and notes. Zoom announcement of 15 September 2026 describing Zoom Revenue Accelerator with Engage and Forecast capabilities, an Elite tier combining Revenue Accelerator Premium with both, and Common Room by Zoom, which remains available as a standalone product; Zoom announced its agreement to acquire Common Room on 2 July 2026. The customer result cited (Barracuda Networks: 15% more deals closed, 5% faster deal velocity) is Zoom's own claim about Revenue Accelerator. Pipedrive's Nova meeting intelligence launched 16 September 2026. Product details are taken from Zoom's press release and trade coverage and have not been independently tested; availability and pricing should be confirmed with the vendor. SalesHubMedia has no commercial relationship with any vendor named. Journalism, not procurement advice. Corrections welcome.
