Lead scoring
Lead scoring answers “How well does this lead fit our services, and how strong is their current buying interest?” It is a priority indicator, not a probability of purchase, and is separate from opportunity win probability and opportunity risk.
Qualification brings service matching and lead scoring together under Settings → Agents → Qualification. Existing scoring settings, reviewed facts and usage are retained. The former Lead scoring link opens this page. Use All agents to return to the agent cards.
Enable and configure
Section titled “Enable and configure”An Owner or Admin opens Settings → Agents → Qualification → Lead scoring, enables scoring, reviews the rules, and chooses Save settings. Managers can view the settings but cannot change them. Each workspace must be enabled separately. If scoring is unavailable, contact Qonvera support.
| Category | Default criterion | Weight |
|---|---|---|
| Fit | Explicitly confirmed service fit | 15 |
| Fit | Target customer fit | 10 |
| Fit | Within the service area | 5 |
| Buying intent | Demo, consultation, quote, or booking requested | 15 |
| Buying intent | Compatible budget or commercial requirements | 10 |
| Buying intent | Specific purchase or start timeframe | 10 |
| Buying intent | Involvement in the purchase decision | 5 |
| Engagement | Meaningful inbound interaction | 10 |
| Engagement | Answers to qualification questions | 10 |
| Engagement | Customer agreed to a concrete next step | 10 |
Set business-specific criteria beside each rule, such as eligible service locations or what constitutes compatible commercial requirements. Weight 0 disables a rule. Remaining positive weights are proportionally normalized to a total of 100; the breakdown shows each rule’s maximum contribution.
Changing weights recalculates scores using the same facts. Changing a written criterion makes that criterion unknown until reviewed or automatically evaluated again. A rules change while a teammate is editing prevents a stale save; reload and review the current criteria before retrying.
If someone changes the settings while you are editing, saving is blocked. Choose Reload saved settings (discard edits), review the latest values, then make your changes again. A failed save refreshes the saved version without automatically repeating the save. Switching workspaces clears the previous workspace’s editor.
Automatic evaluation
Section titled “Automatic evaluation”In Settings → Agents → Qualification → Lead scoring → Automatic evaluation, enable Evaluate supporting evidence automatically, choose the daily evaluation limit, and Save settings. Both workspace lead scoring and automatic evaluation must be enabled. Existing workspaces remain off until an Owner or Admin opts in.
Qonvera evaluates relevant changes to opportunity details, notes, completed follow-up notes, incoming messages, and intake evidence. Existing records are evaluated gradually, so results may not appear immediately. AI reviews the available evidence, then Qonvera uses your configured weights and engagement aging rules to calculate the score. Repeated messages do not earn repeated points.
A manually created opportunity does not need a chat. For example, record “The customer requested a demonstration next week and confirmed a budget of USD 500” in the opportunity note after your conversation. Automatic evaluation can identify the supported request, timing, and budget facts. A selected service, pipeline stage, opportunity amount, expected close date, or scheduled task alone is not proof of customer agreement.
Evidence includes opportunity notes, completed follow-up comments, directly linked website/intake enquiry text, and inbound text from the opportunity’s linked conversation. If several active opportunities share that conversation, its messages are excluded to avoid mixing their qualification. Conversation and intake scores use their own evidence. Contact/account-wide notes and unrelated chats are not combined. Attachments, voice recordings, and external websites are not analyzed.
Automatic evaluation reviews a limited selection of recent evidence, not the entire customer history. AI findings must be supported by that evidence; ambiguous or missing facts stay unknown. Engagement evidence must come from customer messages or intake, not the date a teammate edited a note. Teammates can still review engagement from an offline conversation manually.
Enabling automatic evaluation shares selected evidence with OpenAI for analysis. Email addresses, phone numbers, website addresses, and recognized credentials are removed, but other personal or confidential information may remain in notes and messages. Follow your organization’s privacy policy before enabling this feature. AI usage charges may apply. The daily evaluation allowance is shared across the workspace and resets at midnight UTC; unchanged evidence can reuse an earlier result.
The card keeps the score, one short status, and review actions visible. Expand Score breakdown and evidence for the evaluation time, incomplete-qualification explanation, missing engagement dates, and partial or limited evaluation details. This section also distinguishes AI evaluated from Reviewed by teammate and shows supporting excerpts. Scoring exclusions and action errors remain visible without expanding it. AI does not send messages, create opportunities or tasks, move stages, or change consent. Human-reviewed Confirmed/Not met facts always take priority. In the qualification editor, Unknown removes that manual override and allows automatic evidence to supply the fact again.
Partial evaluation means some AI findings could not be verified. Verified findings still count; unsupported findings earn no points and remain unknown unless a teammate reviews them. Expand the breakdown, then use Review qualification to fill any gaps.
Delayed or failed evaluations
Section titled “Delayed or failed evaluations”If supporting evidence, scoring settings, or reviewed qualification changes during an evaluation, Qonvera may discard that unfinished result. Your reviewed facts remain in place. Allow the latest changes to settle before using Evaluate again; repeated retries do not bypass the daily allowance.
Allow time for an evaluation to finish after changing a record. If it fails, your reviewed facts remain unchanged. An Owner, Admin, or Manager can choose Evaluate again, allowing five minutes between retries. Work deferred by the daily limit resumes after the allowance resets. If a record stays queued or evaluation remains unavailable, share its reference and the visible status with Qonvera support.
Where to review a score
Section titled “Where to review a score”Find Lead score in:
- Opportunities → open an opportunity → Overview.
- Inbox → open a conversation → Overview in the right-hand panel.
- Intake → open an enquiry, including enquiries awaiting review.
Scores belong to the opportunity, conversation, or intake enquiry itself. They do not belong to contacts or accounts and are not displayed on those detail pages. A conversation or enquiry does not need a linked contact to be scored; scoring never creates a contact.
Owners, Admins, and Managers choose Review qualification to open a popup. Its list scrolls independently, with Save and Cancel kept visible. For each enabled rule choose:
- Unknown: not enough evidence; this does not mean poor fit.
- Confirmed: evidence supports the criterion.
- Not met: reviewed evidence does not support the criterion.
The Evidence / source reference and When observed fields are optional. You can save Confirmed or Not met without filling either field. Add supporting detail when useful; if you supply a date, use when the customer signal actually happened, not today’s review date for old evidence. Supplied dates must be valid and not in the future.
Leaving a date empty does not invent an observation time. Confirmed fit and buying-intent facts still earn their configured points; confirmed engagement facts need an observation date to earn points because their value depends on age. The card identifies undated engagement. This does not weaken the evidence requirements for AI-derived facts.
Choose Save qualification to save and close the popup; Cancel or Escape closes without saving. Clicking the background does not dismiss your draft. While saving, closing and duplicate submission are blocked. A failed save keeps the popup and your edits open. Qonvera still records who saved each revision and when. Agents can view scores and evidence but cannot edit qualification.
With automatic evaluation disabled, scores use teammate-reviewed facts only. With it enabled, evidence-backed AI findings supplement—not overwrite—those facts. Website visits, message counts, saved AI service suggestions, and creating an opportunity or follow-up do not by themselves qualify a lead.
Separate records, separate qualification
Section titled “Separate records, separate qualification”Each opportunity has independent facts, even when several opportunities share the same contact or account. Conversations and intake enquiries also have their own qualification. Review service fit against the service relevant to that specific record; there is no contact-wide or service-scope selector.
Creating an opportunity from a conversation or intake enquiry does not copy or combine its score. Automatic evaluation independently reviews available, directly linked evidence for the opportunity. Linking or changing its contact/account does not import another record’s reviewed facts.
Prioritization status → Opted out or Disqualified suppresses the numeric score for that record only. These are scoring exclusions, not a replacement for channel consent or sending restrictions. An eligible score never authorizes outreach, overrides an opt-out, or sends anything.
Score labels and aging
Section titled “Score labels and aging”| Score | Label |
|---|---|
| 0–29 | Early qualification |
| 30–59 | Developing interest |
| 60–79 | Qualified interest |
| 80–100 | High priority |
The three engagement rules earn full points for the first 7 elapsed days, half points from day 7 until day 30, and no points from day 30 onward. Owners and Admins can adjust these boundaries. Fit and confirmed buying-intent facts remain until corrected or replaced by a fresh evaluation. Scores reflect aging when you open them and refresh approximately once a minute while displayed.
Each criterion contributes once, not once per message or form. A website submission and its email notification must not be recorded as multiple scoring events. Scores do not rise when Qonvera sends messages, when AI replies, or when an agent creates a task. Contact merges do not combine scores.
Open Score breakdown and evidence for the individual contributions, unknown facts, aged engagement, source notes, observation dates, rules version, and recent qualification revisions. Qualification history records who made each change and is retained for audit. Deleting a conversation or archiving an opportunity is not blocked by scoring; its score becomes inaccessible through the app while evidence is retained.
Test the configuration
Section titled “Test the configuration”With the default rules and recent evidence, confirm service fit, target fit, a demo request, and all three engagement criteria. The score should be 70 — Qualified interest. Leave budget, timeframe, decision involvement, and location unknown; qualification should still show as incomplete. An engagement observation older than 30 days contributes zero. Change this record’s prioritization status to Opted out: it should show no numeric score, while another opportunity for the same contact remains unchanged.
If saving reports a conflict, cancel the editor and reload before reviewing again. Failed saves retain your draft. If Lead scoring unavailable persists, contact Qonvera support. Scoring does not change sales details, tasks, amounts, or send messages.
To test automatic evaluation, enable it in a test workspace and create an opportunity with a note describing real customer qualification facts. Wait for the card to update, then inspect the AI-labeled evidence. Leave unsupported criteria unknown. Add or correct a fact manually and confirm a later automatic evaluation does not replace it. Without supporting notes or enquiry text, Qonvera cannot infer qualification facts safely.