When Sales Premium actually pays for itself

What it does

Sales Premium is Sales Enterprise with the AI capacity meters removed and relationship intelligence extended to Exchange data. Enterprise already includes the four headline digital selling capabilities — sales accelerator, conversation intelligence, predictive scoring and relationship intelligence — but three of them are capped or degraded. Premium is a capacity purchase dressed as a feature tier.

Key facts

Sales Enterprise includes sales accelerator at 1,500 sequence-connected records per month. Premium has no stated limit.
Sales Enterprise includes predictive lead and opportunity scoring at 1,500 records per month. Premium has no stated limit.
Conversation intelligence carries no capacity limit on either Enterprise or Premium.
Relationship intelligence is the one genuine feature difference: Enterprise gives basic relationship insights and basic who-knows-whom drawn from Dynamics 365 data only. Premium gives the advanced versions, drawn from Dynamics 365 and Exchange.
Setup for all of it sits in Sales Hub under App Settings > Overview > Sales Settings, with a quick setup and an advanced settings page per capability.
Accelerator usage is measured as records connected to sequences in the current calendar month. When you go over, a warning symbol appears and the excess over 1,500 shows in red on the usage graph.
The usage panel is powered by the Sales insights sequence daily usage flow in Power Automate and refreshes every 24 hours. Without that flow running, the usage figures are wrong.
Premium forecasting is GA, not preview. It adds a prediction column and prediction factors to forecasts built on the Opportunity rollup entity.
Premium forecasting needs more than 10 closed opportunities carrying Actual Value, Actual Close Date, Estimated Value and Estimated Close Date, plus open opportunities with estimates, and every underlying opportunity must be owned by a system user. Predictions appear roughly two hours after activation.
Premium forecasting is not available in Government Community Cloud, France or India.

When to use / skip

Do the arithmetic before you do anything else. Count how many leads and opportunities the team will connect to sequences in a month, and how many records need a predictive score. If both numbers are under 1,500 with headroom, Enterprise gives you the same capability and Premium buys you nothing the sellers will feel. That is the common case for a team of twenty or fewer running a considered-sale motion with long cycles.

Premium starts earning its keep in high-velocity inside sales — SDR pods working large inbound lists, where 1,500 sequence-connected records a month is a fortnight's work. There the cap is not a nuisance, it is a hard stop on the operating model, and the upgrade is obvious.

The other honest reason to buy Premium is advanced relationship intelligence, because the Exchange-sourced who-knows-whom is a materially different product from the Dynamics-only version. If the client's actual pain is "we don't know who in our firm already knows this buyer" — professional services, complex accounts, big partner ecosystems — that alone can justify the tier, independent of volume.

Skip Premium for forecasting alone unless the pipeline history is genuinely there. On a fresh implementation you will not have ten properly closed opportunities with clean actuals for months, and predictive forecasting will sit there producing nothing. Buying Premium in month one for a feature you cannot switch on until month nine is a bad trade, and clients remember it.

My default advice: land on Enterprise, turn the capped features on properly, watch the usage panel for a quarter, and upgrade the specific users who are hitting the ceiling. Premium is a per-user licence — you do not have to buy it for the whole sales organisation because two SDRs are busy.

Configuration decisions

Which users, if any, go to Premium, and whether that split follows a role (SDR versus field seller) that will stay stable as the team grows.
Whether the Sales insights sequence daily usage flow is confirmed running before you make any capacity argument from the usage panel.
Whether relationship intelligence draws on Exchange, which is a data-access conversation with the client's security team, not a toggle you flip quietly.
Whether predictive scoring is scoped to all leads and opportunities or to a filtered subset, since the filter is what keeps Enterprise under 1,500 records a month.
When premium forecasting gets activated relative to go-live, given the closed-opportunity threshold and the requirement that opportunity owners are system users.

Gotchas

The 1,500 cap counts records connected to sequences, not sequence steps or emails sent. A bulk connect of a purchased list burns the month's allowance in one action.
The usage panel is 24 hours stale and depends on a Power Automate flow that people disable during environment clean-ups. You can be over the cap for a week before the graph admits it.
Exceeding the cap surfaces as a warning and a red bar, not a clean block, so nobody notices until a licence true-up conversation.
Premium forecasting silently produces nothing when opportunities are owned by teams or by disabled users rather than system users. There is no error — the prediction column is just empty.
Basic relationship intelligence being on by default on Enterprise means a Premium demo can look identical to what the client already has. Demo the advanced who-knows-whom explicitly or the upgrade case falls flat.

Consultant notes

Never quote Premium off a feature grid. Quote it off two numbers the client can verify: sequence-connected records per month and scored records per month. It turns a licence argument into an evidence argument.
Demo the Exchange-based who-knows-whom against real people in the client's own tenant during a trial. It is the single most convincing Premium moment and the hardest to fake with sample data.
Push back when someone wants Premium at kickoff "so we have everything". Sell the trial, instrument it, and revisit at the end of the first quarter.
Before go-live, check the daily usage flow is on and that opportunity ownership is clean — those two things decide whether your capacity data and your forecasting predictions are worth anything.
Set the expectation early that predictive forecasting needs pipeline history. Say the number out loud in the kickoff so it does not land as an excuse in month three.

Worth another look if Microsoft changes the Enterprise capacity allowances, or when the client's inside-sales headcount jumps.