AI call scoring in HubSpot means every sales call is graded automatically against the same scorecard, using the transcript HubSpot already holds. Each call gets a score per skill, the reason for it and the evidence from the call. Coaching stops relying on memory and starts from what was actually said.
Uspeh is a HubSpot Platinum Solutions Partner and HubSpot Software Industry Specialist based in London. We built Call Grading because the sales teams we work with kept telling us the same thing: every call is recorded, but nobody has time to listen.
Here's how it usually goes. A team records its calls, and the transcripts pile up in HubSpot. Then a busy week arrives, and coaching falls back on memory and the odd call someone happened to join. The same mistakes repeat, call after call, and nobody notices until the pipeline does. This guide shows you how to score calls properly, what to score, and how to do it without adding hours to anyone's week.
If you lead marketing, call scores help with a different question: when a deal is lost, was it a bad lead or a bad sales call?
See your own calls scored. We're giving Call Grading free to 20 HubSpot teams in exchange for honest feedback. Book a 20-minute discovery call and mention Call Grading in your message.
AI call scoring is the automatic grading of a sales call against a fixed scorecard, using the call transcript. Instead of a manager listening back and giving a gut score, the AI reads what was said and scores each skill the same way, every time.
It helps to separate it from two things it often gets confused with. Call recording only stores the call, so someone still has to listen. Lead scoring, which HubSpot also does, ranks contacts by how likely they are to buy. That tells you about the buyer, not about how well your rep ran the conversation. Call scoring is about the rep's skill.
Not against a fixed scorecard, based on what HubSpot publishes today. HubSpot records and transcribes calls and helps managers coach, but its public pages don't describe scoring every call against a set of skills with evidence for each score.
HubSpot Conversation Intelligence does a lot well. It records and transcribes calls from supported tools and links them to the right contacts and deals. Managers can find key moments with tracked terms and leave feedback. HubSpot's own comparison of conversation intelligence tools, updated in September 2026, rates its coaching features as strong. HubSpot also offers a Sales Coach Assistant that suggests coaching from call and deal insights. And as we covered in our Fall 2026 Spotlight breakdown, Deal Progression now reads calls to suggest next steps.
What those tools don't publicly describe is one consistent scorecard applied to every call, with a score, a reason and transcript evidence for each skill. HubSpot moves fast, so it's worth checking the latest release notes. But that gap is why we built Call Grading as a layer on top of HubSpot, not a replacement for it. It reads the transcripts HubSpot already holds and adds the scoring.
Score each skill on its own against one fixed scorecard, back every score with evidence from the call, and turn the lowest score into one coaching action. That's the whole method. Here's how to do each part well.
If the scorecard changes every week, the scores stop meaning anything. Choose the skills that matter for your sales motion, then leave them alone for at least a quarter. Consistency is what lets you compare a rep's calls over time, and one rep with another.
"That call was a 6" doesn't help anyone. A rep can open well and ask great questions, then forget to agree a next step. Score each skill out of 10 on its own, so the rep can see exactly where the call slipped.
A score without a reason feels like a judgement, while a score with a line from the call feels like feedback. For every score, point to the moment in the transcript that earned it. This is also how you spot a wrong score, because the evidence won't hold up.
Don't hand a rep eight things to fix. Pick the lowest score and give one action, ideally with a line they can use on their next call. Small, specific changes stick, while long lists get forgotten.
A good B2B scorecard measures how the rep opens the call, uncovers pain, maps the current situation, shows the cost of doing nothing, pitches to what they heard, handles objections, sounds confident and closes on a clear next step. These are the eight skills in our sample report. With Call Grading, the skills and scoring are built around the call structure your own team follows.
| Skill | What it measures | Weak line | Strong line |
|---|---|---|---|
| Opening & frame | Sets the purpose, agenda and time, and gets the buyer to agree | "Hi, just wanted to touch base and see how things are going." | "I'd like to understand how your team handles this today, then decide together if a next step makes sense. Does 30 minutes still work?" |
| Discovery of pain | Uncovers real problems in the buyer's own words | "Would better reporting help you?" | "You said forecasts keep slipping. What happened the last time that went wrong?" |
| Current situation | Maps how things work today: process, tools and people | "So you use HubSpot, right? Great." | "Walk me through what happens after a call ends. Who updates the deal, and when?" |
| Cost of inaction | Helps the buyer see what waiting will cost them | "So, shall we book a demo?" | "If nothing changes by March, what happens to the targets you mentioned?" |
| Pitch relevance | Links the pitch to the pain the buyer described | "Let me show you all our features." | "You mentioned new reps take months to ramp. Here's the one part that helps with that." |
| Objection handling | Explores the concern before answering it | "Actually, we're cheaper than most tools." | "That's fair. When you say it's too expensive, what are you comparing it with?" |
| Confidence & tone | Sounds calm and clear, not rushed or apologetic | "Sorry, I know you're busy, this will only take a second." | "I'll keep this to the point. Here's what I suggest we do next." |
| Closing & next steps | Agrees a specific next step with a date and an owner | "I'll send some info over and you can let me know." | "Shall we book 30 minutes on Thursday with your finance lead to go through the numbers?" |
Call Grading is Uspeh's AI call scoring tool that grades every sales call inside HubSpot against your sales framework, using transcripts you already have, and shows the evidence behind every score.
As soon as a call transcript arrives in HubSpot, Call Grading scores it automatically, so nobody has to choose which calls to listen to. Each grade is then ready for a manager to review, edit or approve before it's used for coaching.
Call Grading reads transcripts that are already in HubSpot, from HubSpot calling or from a note-taker such as Fireflies that saves transcripts to HubSpot. There's no new login, no extra bot joining your calls and nothing new for reps to learn. HubSpot's own transcription needs Sales Hub or Service Hub Professional or Enterprise. On other plans, a note-taker that saves transcripts to HubSpot does the same job.
Each graded call gets:
Sample report from a client delivery call. Download the full sample report (PDF)
A manager does. Every grade is ready for manager review, and managers can review, edit or approve each AI grade before it's used for coaching. The skills and scoring are built around the call structure your team is meant to follow, so they can match MEDDIC, Sandler or your own sales method.
It's better you hear this now than after a discovery call. Call Grading doesn't give live help during a call, because it works afterwards, from the transcript. It doesn't offer a deal-risk or forecasting view. If live coaching or forecasting is your main need, one of the tools in the next section may suit you better.
Be one of 20 teams to try Call Grading free. We're opening Call Grading to 20 HubSpot teams at no cost. In return, we ask for honest feedback on what works and what doesn't.
Book your discovery call and mention Call Grading in your message.
Avoma, Jiminny and Gong all score calls, and they're mature products. The main difference is where the scores live. They run as separate platforms that connect to HubSpot, while Call Grading works inside HubSpot on transcripts you already have.
| Tool | Rubric scoring | Where results live | Published price | What it needs |
|---|---|---|---|---|
| HubSpot Conversation Intelligence | Coaching with tracked terms and feedback; scoring every call against a fixed scorecard isn't described in public docs | Inside HubSpot | Included in Sales Hub Professional, from $90 per seat a month (annual) plus a $1,500 onboarding fee | Sales Hub or Service Hub Professional or Enterprise |
| Avoma | Yes, AI call scoring against your criteria | In Avoma, with notes saved to HubSpot | From $19 per seat a month, plus $29 per seat a month for the Conversation Intelligence add-on (annual) | An Avoma plan plus the add-on |
| Jiminny | Yes, automated call scoring | In Jiminny, with a native HubSpot integration | Quote-based; recording seats reported from $83 a month | A Jiminny licence |
| Gong | Yes, scorecards | In Gong, with a HubSpot integration | On request | A Gong licence plus a platform fee |
| Call Grading | Yes, every call against your sales framework, with evidence and manager review | Inside HubSpot | Not published yet | Call transcripts in HubSpot |
Prices are in US dollars as published, and they change often, so check with each vendor before you buy.
If you want one platform for recording, live help and forecasting, a full platform like Gong or Avoma makes sense. If your calls already land in HubSpot and you simply want them scored there, a layer is the simpler choice.
HubSpot records, transcribes and supports coaching, and it does those things well. What it doesn't publicly describe is scoring every call against your sales framework with evidence. Call Grading adds that layer, and you keep using HubSpot for everything else.
No scoring is perfect, whether a human or an AI does it. That's why every Call Grading score shows its reason and the evidence from the transcript. You can check any score in seconds, and a manager can review, edit or approve each grade before it's used for coaching. A gut score from a busy manager rarely comes with that much proof.
Neither do your reps. Call Grading has no new login and no bot on calls, and the results sit in HubSpot, where your team already works.
They will if scores feel like surveillance. They won't if every score comes with evidence and one useful action. As we wrote in Human-Centric RevOps, it's worth asking your team where they feel judged or exposed by the data you capture, then designing around the answer. Frame grades as coaching, share the example lines, and point out strengths as often as gaps.
Your recorded calls already show what your reps do well and where they slip. The problem is that nobody has time to listen to them. Scoring every call against the same scorecard, with evidence, turns those recordings into coaching your reps can use on their very next call.
Claim one of 20 free places. Get Call Grading free in exchange for honest feedback. It starts with a 20-minute discovery call. Book yours now and mention Call Grading in your message.