When leads are plentiful and sales are few, don't start by changing the ad. Put last month's numbers into a five-stage table, then split it by lead source and by salesperson. In an hour or two, that shows whether your lead conversion problem sits with the ad, the page or the follow-up.
A lead is someone who left their details or messaged you and hasn't bought yet.
Step 1: Fill in the stage table from last month
Sit with whoever handles leads and write the real count at each stage. If a number doesn't exist, that's your first finding: you can't judge the campaign until it's recorded. These figures are hypothetical.
| Stage | Hypothetical count | Share of previous stage |
|---|---|---|
| Leads received | 200 | |
| Actually contacted | 150 | 75% |
| Serious leads | 60 | 40% |
| Reached a quote | 25 | 42% |
| Paid | 8 | 32% |
Define "serious" before you count: has the budget, needs the service within an agreed period, and can make the decision. Left to each salesperson's judgment, the numbers can't be compared.
No rate is right for every business, so skip the hunt for a general benchmark. Compare each stage with your own earlier months, and start with the one that fell most or where the drop costs the most money.
Step 2: Split the numbers to separate the ad from sales
This is the test that matters most here, because it identifies who owns the problem and ends the debate between marketing and sales. Repeat the table for each split:
| Split | What you see | What it means |
|---|---|---|
| By source | One source has a far lower serious rate than the others | The fault is in that source's targeting or message |
| By source | Every source drops by about the same rate | The fault is probably after the lead arrives: reply, offer or price |
| By salesperson | One closes clearly more on similar leads | The fault is in the selling method, and that person's method is what to spread |
| By reply time | Leads answered quickly convert more | The fault is in reply speed and how leads get assigned |
Make sure each row holds enough leads. Don't judge a source from ten of them. Combine two or three months if numbers are low.
Step 3: Read the biggest drop and match it to its causes
- Received to contacted: wrong numbers, accidental form fills, or leads landing in an inbox one person opens once a day. Send yourself a test lead and time the reply.
- Contacted to serious: broad targeting, or a campaign optimized for the cheapest lead, so the platform finds people who readily leave details. Check the ad's promise too: if the page doesn't deliver it, people leave disappointed.
- Serious to quote: the budget question comes late, or nobody has time to prepare quotes. Add one budget question to the first call.
- Quote to paid: an unclear offer, a price the value doesn't support, or no follow-up after sending the quote.
What about reply speed?
Slow replies are a common cause and easy to test. A study published in Harvard Business Review in 2011 concluded that most companies don't respond to online leads nearly fast enough, per the study record. Work out your own figure: the average time from a lead arriving to first contact, split between working hours and after hours. Agree on a time you can keep, send each new lead straight to a named person with an alert, then watch whether the contact rate moves.
Step 4: Review 30 leads and 10 calls yourself
Numbers show where the fault is, and people tell you why. Take 30 random leads from last month and note each one's source, what they asked for, whether they could buy at your price, and the last thing that happened. Then listen to ten calls, or read ten WhatsApp chats, and record:
- Who talks more, the salesperson or the customer?
- Did the salesperson know which ad the lead came from?
- What next step did they agree on?
- Which objection came up most, and how was it answered?
Pass each lead to sales with its context: source, page visited, service asked about, and the answer to one screening question. That changes the first sentence of the call.
Step 5: Send sales outcomes back to the ad platform
If the platform sees the lead but never the sale, it keeps looking for more people who leave details, buyers or not. Google offers enhanced conversions for leads, which returns hashed data about who became a customer, per the Google Ads Help Center. Meta lets you send events that happen off the website through its Conversions API, per its developer documentation. All of this needs sound measurement first, covered in how to measure conversions after the privacy changes.
A decision table: what do you fix first?
| Diagnosis | Fix first | Don't start with |
|---|---|---|
| Leads nobody called | Lead assignment and reply alerts | Changing the ad |
| One source brings weak leads | That source's targeting and the form's questions | Sales training |
| Every source is weak on seriousness | The message, the offer and price clarity on the page | Raising the budget |
| Many quotes, no payment | Quote wording and a follow-up plan after sending | Cutting the price right away |
| A wide gap between salespeople | Documenting the best method and training the others | Moving budget |
Common questions
Does a longer form bring better leads? It may lower the count and raise seriousness, and the result varies by business. Test one screening question and compare serious leads before and after, ignoring the total.
Should I pause the campaign while diagnosing? Usually there's no need, unless leads have nobody to answer them. In that case pause or reduce spend until replies are possible, because you're paying for leads nobody will call.
To place this diagnosis in the wider plan, see digital marketing for companies, and use customer acquisition cost to turn any improvement into money.
Next step
Book two hours this week with whoever owns sales and whoever owns ads, fill in the stage table for last month, and split it by source. If you'd like a team to tie campaigns to sales results and give you this table monthly, take a look at PlanKit's digital marketing and campaigns service.