Cost per lead is the wrong number to manage

Ask an advertiser how their campaigns are doing and you will usually get a cost per lead. It is the number on the front of the dashboard, it moves quickly, and it feels like a measure of efficiency. It is also the number you can improve fastest while making less money, which makes it a poor thing to steer by.

Why the number is so easy to game

Cost per lead falls whenever leads get cheaper to acquire. That happens for good reasons, like a better landing page or sharper targeting. It also happens for bad ones.

Broaden your match types and you will get more clicks from people with vaguer intent. Some of them fill in the form. Cost per lead drops. Add a low-friction offer, a guide or a checklist or a discount code, and your form fill volume jumps. Cost per lead drops again. Count newsletter signups as conversions and it drops further.

None of those changes sold anything. Two of them actively made things worse, because now the sales team is spending its week on people who were never buying. But the report looks better than it did last month, and if cost per lead is what gets reviewed, the account will keep being pushed in that direction.

What to manage instead

Cost per sale, or cost per booked job, or cost per closed deal. Whatever the last step in your process is called, that is the number that connects marketing spend to revenue.

The objection is always the same: we cannot measure that. Sometimes that is true, usually it is a matter of nobody having set it up. Most businesses can get closer than they think.

If the sale happens fast, as it does in ecommerce and in urgent home services, you already have the data. The purchase or the booked job is a trackable event and it can be sent back to the ad platform with a value on it.

If the sale happens in a CRM weeks later, the connection is a data problem rather than an impossible one. Stamp lead source and click identifier on the record when the lead arrives, then export closed deals back to the ad platform as offline conversions. Google Ads has supported this for years and hardly anyone running lead generation uses it.

If most of your leads arrive by phone, and in home services and healthcare most of them do, you need call tracking with dynamic number insertion before any of this works. Without it, a large share of your outcomes are invisible to everything doing the optimising.

An example of how far apart the two numbers get

Take two campaigns in the same account, both spending five thousand dollars a month.

Campaign A produces 100 leads. Cost per lead is fifty dollars. It looks like the strong performer, and in most reports it is the one held up as the benchmark.

Campaign B produces 40 leads. Cost per lead is a hundred and twenty five dollars. It looks like the one to cut.

Then you look at what closed. Campaign A converted at four percent, which is four sales, at a cost per sale of one thousand two hundred and fifty dollars. Campaign B converted at twenty percent, which is eight sales, at a cost per sale of six hundred and twenty five dollars.

Campaign B is twice as efficient at the only thing that matters, and every cost-per-lead report you produced said to cut it. This is not a hypothetical shape. It is what usually happens when a campaign targeting broad research terms sits next to one targeting specific buying terms, and it is why blended lead metrics quietly misallocate budget for years.

Not all conversions deserve the same weight

If you are going to keep multiple conversion actions, and most businesses should, give them values that reflect what they are worth to you.

A booked consultation is not the same as a guide download. If both are counted as one conversion, and the platform is using automated bidding, it will chase whichever is cheapest to produce. That will always be the download. You have effectively instructed Google to buy you the least valuable outcome on your site, and it will do it very efficiently.

You do not need perfect numbers here. Rough relative values, informed by your actual close rates, work far better than treating everything as equal. If a demo request closes at twenty percent and a download closes at one percent, a twenty-to-one value ratio is closer to the truth than one-to-one, and the bidding will act on it.

Where cost per lead is still useful

It has a place. It is a fast diagnostic, it tells you quickly whether something broke this week, and it can be compared across campaigns that produce the same kind of lead.

The rule worth holding is that cost per lead is a monitoring metric and cost per sale is a decision metric. Watch the first one weekly. Make budget decisions on the second one.

Where to start

If none of this is currently measured, the order that works is:

1. Get call tracking in place if calls matter to you. Everything downstream depends on knowing what happened.
2. Write down what counts as a lead, and get sales to agree to the definition. This is a business conversation, not a technical one, and skipping it is why so many tracking projects produce numbers nobody trusts.
3. Stamp lead source into the CRM on every enquiry.
4. Export closed deals back into the ad platforms as offline conversions.

Step four is where the compounding starts, because at that point the automated bidding is learning from your revenue rather than from your form fills. It takes a few months of data before it changes behaviour, which is a good reason to start it before you need it.