Lead Qualification: How to Decide Which Leads Get Your Time

Lead qualification means checking each lead against criteria you wrote down in advance, then deciding whether it gets a sales conversation now, a follow-up later or a polite no. Nearly every framework tests the same four things in a different order: a problem you can solve, a person who can decide, money that can be spent and a reason to act soon. Which acronym you pick matters less than recording the answers in the same fields every time, so that "qualified" means the same thing to everyone on the team.

Customermates publishes this guide and is the CRM in the setup example further down. It has no lead scoring, and the sections below say where that matters.

Lead, MQL, SQL, opportunity: the order of stages

A lead is anyone who has shown interest: they sent an inquiry, were referred to you or replied to an outreach message. It counts as qualified once someone has checked it against your criteria and the answer was yes. Companies with a marketing team split that check in two:

HubSpot's default lifecycle stages run in the same order, from subscriber and lead through MQL, SQL and opportunity to customer and evangelist, according to its lifecycle stage documentation. In Salesforce, converting a lead uses the lead record to create a business account, a contact and an opportunity, as its Trailhead module on leads describes. The lead always comes first. The opportunity is what it turns into once it qualifies.

A team of three without a marketing department can drop the MQL stage. Every inquiry goes to the person who sells, and the only line that matters is the one between lead and qualified.

Two kinds of criteria: fit and readiness

Fit asks whether you want this client. It comes from your ideal customer profile: industry, company size, location, type of project and whether you have done similar work. Most of it can be checked before the first call, from the company website or a LinkedIn profile.

Readiness asks whether they can and will buy soon. That is need, authority, budget and timing, and you only learn it in conversation.

Keep the two apart. A lead can be ready and still a bad fit: a company that wants the project next month, at half your day rate, in a field you don't know. Write down two or three fit criteria that end a conversation early. A branding agency might settle on B2B companies with 20 to 500 employees, someone in-house who owns marketing, and project budgets from €15,000.

Lead qualification frameworks compared

The frameworks differ mainly in where the conversation starts and in how many people they assume take part in the purchase.

HubSpot's sales qualification guide credits MEDDIC to Jack Napoli at PTC and GPCTBA/C&I to HubSpot itself, and names the RAIN Group as an advocate of FAINT. SPICED comes from Winning by Design.

How to choose:

• BANT or ANUM if you sell a defined package at a known price and one person signs. ANUM if your leads come from outbound.
• CHAMP if you sell services and the first call is about the client's problem anyway. For consultants and agencies it matches how a good first call already runs, and it's where I would start.
• MEDDIC or SPICED if three or more people decide, procurement is involved and deals take months.
• One at a time. Map one framework's letters to fields, use it for a quarter, and change it if the same fields keep coming back empty.

A warning about BANT: asking for the budget in the first call works for a product with a price list. For a service it often doesn't, because the client has no budget until they know the scope. Ask for a range, or for what the problem costs them today.

Questions that produce usable answers

Write the answer down in the client's words. "Board approved 30k for Q1" in the deal notes is worth more than a ticked Budget box, because the next person can judge it for themselves.

When to say no, and when to park a lead

Disqualify when one of these is true:

• The lead fails your fit criteria.
• There is no real need, only curiosity or research for someone else.
• Nobody you can reach is able to decide, and nobody will introduce you.
• They want something you don't offer, and stretching for it would hurt the work.

Park the lead instead when the need is real but the timing isn't: the project starts in six months, this quarter's budget went elsewhere, or your contact can introduce you to the decider later. A parked lead needs a reason from a fixed list and a date to look again.

Count those reasons every month. If most refusals say "budget below range", the problem sits in where your leads come from or how your prices are presented. Better sales calls won't fix it.

Handing leads from marketing to sales

When marketing passes leads on and sales turns down a large share as unqualified, the usual cause is that nobody wrote down what qualified means. A written handoff takes an afternoon:

1. Define the MQL in one sentence

   For example: "A contact at a company in our target profile who asked for prices or a call." Add what doesn't count: newsletter sign-ups, job applicants, students, suppliers.

2. Define what sales must confirm

   List the SQL criteria and the minimum answer for each. Both teams work with the same fields.

3. Agree on a response time

   Harvard Business Review published The Short Life of Online Sales Leads in March 2011. Its authors, James Oldroyd, Kristina McElheran and David Elkington, concluded that most companies were not responding nearly fast enough. Pick a target you can keep, such as the same working day, and measure it.

4. Make rejection a field

   Sales may send a lead back only with a reason from a fixed list: wrong fit, no need, unreachable, duplicate, too early. A reason in a field can be counted. A complaint in a meeting can't.

5. Recycle instead of deleting

   Leads marked "too early" go back to marketing with a date to look again.

6. Review once a month

   Count rejections by reason and by source. If one source produces most of the "wrong fit" rejections, change the source rather than the rules.

Any CRM in which both teams see the same fields and the same reason list can carry this agreement. The written definitions do most of the work.

Lead scoring, AI and who makes the decision

Lead scoring is a separate step from qualification. It adds points for attributes such as job title and for actions such as a visit to your pricing page, then passes a lead to sales above a threshold. That needs website and email tracking, plus enough leads to tune the threshold. HubSpot's lead scoring tool, for example, offers fit, engagement and combined scores and needs Marketing Hub or Sales Hub at the Professional or Enterprise tier. Scores built with AI need Marketing Hub Enterprise and a sample of at least 50 contacts, 25 converted and 25 not, according to HubSpot's documentation.

With a few dozen inquiries a month, I would skip points entirely. You can read every inquiry yourself, and a threshold tuned on thirty leads tells you very little.

Where AI helps a small team is the reading. A language model such as ChatGPT or Claude can map an inquiry or an email thread to your criteria, point to the sentences it relied on and apply the same rules to every lead. It does badly with thin input, such as a name and "tell me more", and it cannot weigh a referral from a client you trust. The bigger risk is guessing: unless told otherwise, a model fills gaps with plausible answers. This prompt works in any AI client:

You qualify inbound inquiries for [company], which sells [service] to [ideal client].
Read the conversation below. For each criterion, pick one option and add the sentence that supports it.
- Fit: Yes / No / Unknown
- Need: Confirmed / Vague / None / Unknown
- Decision maker: Talking to the decider / Can reach the decider / No access / Unknown
- Budget: In our range / Below our range / Unknown
- Timing: This quarter / Later / Unknown
Answer Unknown whenever the text does not say it. Do not infer budget from company size.
Finish with the two questions that would settle the most Unknowns.

Let the model fill the fields and let a person decide. Never let it reject a lead on its own.

Setting up lead qualification in Customermates

Customermates has no separate lead record. A new inquiry is a contact, linked to an organization, plus a deal that carries the qualification fields. Contacts come with a Sales Pipeline field (New, Contact, Qualified, In Progress, Won, Lost) for a quick status before a deal exists. If your team says "opportunities", rename Deals in Configure: open the list and change Name and Plural name under General; records and links stay as they are. The lead management page shows the same model from the contact side.

Create the fields in Configure for Deals with Add › Field:

Add a Source field (Single choice: Referral, Website, LinkedIn, Event, Outbound) on contacts, so the monthly review can count by source. The percentages are stage probabilities, set as Probability % on each option of the Stage field in Configure for Deals. They drive the weighted pipeline value, and Parked at 0% keeps parked deals out of that total.

Then save three views on the Deals list. Saved views are personal, so each person saves their own:

• To qualify: Stage is Prospecting or Qualification, sorted by created date, oldest first.
• Missing answers: the same filter as a board grouped by Decision maker, so the Unknown column shows where calls are needed.
• Parked: Stage is Parked, sorted by Look again on. Customermates has no reminders, so check this view each week.

On Pro and higher, the unified inbox puts email, LinkedIn, WhatsApp, Instagram and Telegram conversations next to the contact. Pro includes 1 connected account per user and Business 3. Starter has no inbox.

In Customermates Cloud, Mate can fill the fields for you, using hosted AI credits. Open a deal and ask:

On Starter, which has no inbox, Mate can only work from what is already on the deal and its contacts, such as notes you pasted in.

To do that for every new deal, create a routine with the trigger When something changes and the event Record created for Deals:

A routine runs as its owner, with that owner's access and channels, and Mate sends messages when an instruction tells it to, so the last sentence matters. Starter includes 1 routine per user and Pro 5; each routine may only run a limited number of times per hour, so when many deals are created within the same hour, some of them get no run. The routines documentation lists the current limits. With your own AI client instead, such as Claude or ChatGPT, the same fields are reachable through the MCP server, billed by your AI provider. A self-hosted installation keeps fields, stages, views and MCP, but has no inbox and no Mate.

What it costs, billed monthly:

Each hosted plan includes monthly hosted AI credits per active user, see pricing. The trial runs 7 days.

Measuring whether qualification works

In Customermates, a dashboard chart can count deals grouped by Stage or by Disqualification reason and sum their values. Rates and response times aren't calculated in the app: export the deals as an XLSX workbook and work them out in Excel.

FAQ

Vendor and framework details checked on 8 October 2026 against HubSpot's sales qualification guide, lifecycle stages and scoring tool documentation, Salesforce's Trailhead module on leads, Winning by Design's SPICED framework, Sierra Interactive's LPMAMA guide and Harvard Business Review's article on online sales leads.