Lead qualification

# Writing the lead back to the CRM

## Always search before you create

The most common damage done by automated lead handling is duplicate contacts. Once a customer exists twice, their history splits, the wrong one gets emailed, and merging is manual.

Search in this order, stopping at the first confident match:

1. Email, exact
2. Phone, normalised. Strip spaces, brackets and country code variations before comparing
3. Name plus address or company

Only create when all three miss. When you find a match, update it and add the enquiry to it, rather than creating a second record with newer details.

| System | Search |
|---|---|
| LeadConnector | `leadconnector_search_contacts` |
| HubSpot, Salesforce, Attio | the connector's contact search |

## What to write

**On the contact:**

- Name, email, phone
- Source, specifically. The campaign or the page, not "website"
- Consent state, if you captured it. Enquiring is not consent to marketing

**On the opportunity or deal:**

- What they asked for, in their words. Do not paraphrase the enquiry away
- The score and which criteria produced it
- The band and the routing decision
- Estimated value, if you can reasonably estimate one
- Next step, with an owner and a date

**As a note:**

The full original enquiry text, verbatim. Every summarisation loses the detail that turns out to matter, and the original costs nothing to keep.

## Tags

Useful tags are ones you would filter on later. Everything else is noise that makes the CRM slower to read.

Worth tagging: source, service type, qualification band, and the decline reason where declined.

Not worth tagging: sentiment, urgency as claimed by the lead, anything you would not build a list from.

Agree the tag vocabulary with the customer before creating tags. A CRM with `hot`, `Hot`, `hot-lead` and `HOT` in it is one nobody trusts, and tag lists are rarely cleaned up.

## Pipeline placement

Put the lead in the stage that reflects reality, not the stage that makes the pipeline look healthy.

A cool lead in a nurture stage is honest. The same lead in "Quote Sent" because it will probably get a quote eventually is how a pipeline becomes a work of fiction, and a fictional pipeline leads to hiring decisions.

```
leadconnector_create_opportunity
leadconnector_update_opportunity_status
```

## Declines

Record them, with the reason, in a form you can count.

Declines are the most undervalued data in a CRM. Three months of decline reasons tell you whether your marketing is attracting the wrong work, whether your prices are out of step, or whether you are turning down a service line you should be offering. None of that is visible from wins.

Use a small fixed list of reasons, not free text:

`out of area`, `out of scope`, `too small`, `too large`, `budget mismatch`, `timeframe`, `not licensed`, `no response`, `spam`.

## Do not

- **Do not add to a marketing list automatically.** Enquiring about a job is not consent to a newsletter, and in most jurisdictions treating it as such is a compliance problem, not a style question
- **Do not overwrite existing contact fields with worse data.** A web form with `Dave` should not replace `David Whitfield`. Add, do not clobber
- **Do not create an opportunity for spam or a vendor pitch.** It distorts every conversion number afterwards
- **Do not write a score into a field without also writing what produced it.** A number with no reasoning cannot be reviewed, and an unreviewable model never improves

# Scoring

A simple additive score, out of 100. It is a sorting device, not a truth. Its value is that it is applied the same way every time, which no human does unaided.

## The criteria

### Fit, 30 points

Is this work we want.

| | Points |
|---|---|
| Exactly what we do, in our area, in our size range | 30 |
| We do this, but it is at the edge of our range or area | 20 |
| Adjacent. We could do it, we do not usually | 10 |
| Not our work, or outside our area | 0, decline |

Fit is the criterion worth being strict about. Everything else can be improved by a conversation. Fit cannot.

### Need and timeframe, 25 points

| | Points |
|---|---|
| A specific deadline, or something is currently broken | 25 |
| Planned, with a stated timeframe within three months | 18 |
| Intends to proceed, no date | 10 |
| Researching, comparing, planning for later | 4 |

Weigh demonstrated urgency, not claimed urgency. "ASAP" is worth nothing. "The tenant moves in on the 14th" is worth everything.

### Budget, 25 points

| | Points |
|---|---|
| Budget stated, and it is realistic | 25 |
| Budget stated, and it is low but movable | 15 |
| No budget stated, but the job type is one they will expect to cost this | 12 |
| Expectations clearly unrealistic | 5 |
| Explicitly price shopping across many quotes | 5 |

A lead volunteering a budget is a strong signal even when the number is wrong. It means they have thought about buying.

### Authority, 20 points

| | Points |
|---|---|
| The decision maker, or the owner | 20 |
| Influences the decision, decision maker identified | 13 |
| An intermediary, decision maker unknown | 7 |
| Unclear who decides | 4 |

For consumers this is usually 20 and not worth thinking about. For businesses, property managers and anything with a committee, it is often the criterion that decides whether a deal ever closes.

## Bands

| Score | Band | Response |
|---|---|---|
| 75 to 100 | Hot | Call within the hour |
| 50 to 74 | Warm | Personal reply same day, propose a specific next step |
| 25 to 49 | Cool | Useful reply within 24 hours, add to nurture |
| Under 25 | Low | Reply honestly, decline if out of scope |

## Handling missing information

Most enquiries arrive with a name, an email and a sentence. Do not let that produce a low score by default, because it would rank the least informative enquiries lowest regardless of quality.

**Score what you have, and normalise.** If you can only assess fit and need, score those two out of 55 and scale to 100. Record which criteria were assessed.

**Then ask.** One or two questions, in the first reply, targeted at the criterion that would move the score most. Usually timeframe and budget.

Never invent a value for a criterion you cannot see. A confident score built on three guesses is worse than an honest partial one, because it gets acted on.

## Overrides

The score is overridden, upward, by any of these:

- **An existing customer.** Loyalty outranks the model. Call them
- **A referral from a customer or a partner.** Referred leads convert far above their score, and a bad referral experience costs the referrer's goodwill too
- **A large or strategic account**, even at a poor score
- **A repeat enquirer.** Someone enquiring a second or third time is telling you something the form did not capture

Downward overrides are rarer and should be evidence based, not a feeling:

- A documented history of non payment
- Behaviour towards staff on a previous job
- Work that is unsafe, or that the business is not licensed for

## Adjusting the weights

The weights above are a starting point, not a fact. A business that only ever works locally should drop fit and raise budget. A business with a long sales cycle should raise authority.

Review the model when there is evidence: take three months of leads, compare score against outcome, and see which criterion actually predicted a win. If none of them did, the model is wrong and should be rebuilt around what did.

Record any change to the weights and the date, so a shift in conversion can be attributed to something.