“My cost per lead has doubled” — diagnose before you cut

If your cost per lead has doubled, platform price inflation explains almost none of it. Meta’s own quarterly accounts report average price per ad up 12% year-on-year in Q2 2026 — not 100%. A doubling is nearly always one of four causes: auction pressure, creative fatigue, tracking loss, or audience exhaustion. Each has a different confirming check, and a different fix.

• Check the number is real first. CPL is spend divided by leads. At 10 leads a week, losing 3 of them raises CPL 43% with no change in cost.
• Subtract the platform. Published ad-price inflation is a low-double-digit number (Meta: +12% YoY, Q2 2026). What is left is yours.
• Four causes, four checks. Auction pressure shows in Search lost IS (rank); creative fatigue in click-through rate; tracking loss as a CRM-versus-platform gap; audience exhaustion when you exclude the last 180 days of leads and CPL falls back.
• Do not cut budget on day one. Cut when cost per sale exceeds gross profit per sale, not when CPL rises. At $40 CPL and an 8% lead-to-appointment rate an appointment costs $500; at $80 CPL and 20%, $400.

First: is the doubling real, or is it a small-numbers problem?

Cost per lead is an arithmetic ratio, not a measurement: CPL = spend ÷ leads. The denominator is a count of events, and counts wobble. Hold spend at $500 a week: ten leads gives a $50 CPL, five leads next week gives $100. It doubled and nothing happened. A weekly lead count swings by roughly its own square root — ±3 at 10 a week, ±10 at 100 a week, a 10% swing rather than a 30% one.

The decision rule: compare two consecutive 28-day windows, never two weeks. Under about 30 leads a month, one bad window is noise and you need two consecutive windows above baseline before it is a trend worth paying to fix. Above 200 leads a month, a 28-day doubling is real; act today.

The four causes of a rising cost per lead, and the check that confirms each

Most teams guess, change three things at once, and destroy the evidence. Each cause below has one confirming check that rules the others out. Run them in this order — tracking loss first, because it is the only one where CPL never actually moved.

The four checks are mutually exclusive by design: each produces a result the other three cannot. Google defines Search lost IS (rank) as “the percentage of time that your ads weren’t shown on the Search Network due to poor Ad Rank in the auction”, and Search lost IS (budget) as the same figure “due to insufficient budget”. Opposite problems, opposite fixes.

“Ads have just got more expensive” — how much of that is true?

Partly true, and worth quantifying. In its Q2 2026 results Meta reported revenue of $60.80 billion, ad impressions up 14% year-over-year and average price per ad up 12% year-over-year — Meta’s own price for its own inventory, filed under securities law, not a market-wide CPM index.

Call it the 12% test: subtract published platform inflation from your increase, and what remains is yours to fix. Twelve points inside a hundred-point problem is about an eighth of it. The honest caveat: 12% is a global blended average, so your category, city and season can all run hotter, and a single competitor entering your auction is invisible in it. That is what Auction insights measures, and why it is the second check rather than the first.

What to do in the next 24 hours

All five are free, take about an hour, and involve nobody but you.

1. Rebuild the number from source records. Spend from the billing tab, leads from the CRM, two consecutive 28-day windows. Not a reporting dashboard — that is the thing under suspicion.
2. Split the CPL into its parts. Pull CPM, CTR and landing-page conversion rate for both windows and compare the three multipliers. Worked example: CPM $18.00 → $21.00 (×1.17), CTR 1.4% → 0.9% (×1.56 on cost), landing-page conversion 12.0% → 11.5% (×1.04). Multiply: 1.17 × 1.56 × 1.04 = 1.89, and CPL goes $10.71 → $20.29 — the half-point CTR drop is more of the rise than everything else combined. Substitute your own numbers; the largest multiplier is your cause.
3. Check the trivial explanations first. A paused ad set, a declined card that restarted learning, a form field added last month, an expired offer, a pixel or conversions-API event that stopped firing.
4. Check the calendar. End of financial year, school holidays, a public holiday week, an election, a competitor’s launch sale. Compare with the same period last year, not just with last month.
5. Change nothing else today. Not bids, not budget, not targeting. Simultaneous changes destroy your ability to attribute the recovery, and you will be back here in six weeks.

What to do in the next 7 days

Days 1–2: run the four checks in order. Day 3: make one change and only one — a new creative concept rather than five variations of the old one, or a bid-strategy change, or an audience exclusion. Days 4–7: hold budget flat through a full learning window, read on a 7-day rolling basis, decide on day 14.

Then do the part almost nobody does: pull cost per booked appointment and cost per sale for the same two windows. Run paid acquisition against a sales team and those two decide everything; our cost per lead benchmarks by industry give you a comparison point. Doing this yourself costs three to four hours of one analyst’s time, a CRM export and a spreadsheet — no tooling spend. The hard version is proving the spend caused the leads at all, which needs a geo or holdout incrementality test and costs you deliberately suppressed volume for two to four weeks.

Cutting budget is the move that most often makes it worse

Two mechanisms. Conversion-optimised campaigns need a minimum weekly conversion volume to stay out of the learning phase, so halving spend can push an ad set under that floor and raise CPL further. And if you scaled spend deliberately before the rise, part of the increase is the ordinary efficiency decay that comes with buying a colder audience: arithmetic, not a fault.

The one threshold that justifies cutting today: if cost per sale now exceeds gross profit per sale, cut immediately, because every additional lead loses money. If cost per sale is still below it and only CPL has moved, cutting is a decision to buy less of something profitable. One other reason is legitimate: if you cannot ring, qualify and service the leads you already have, buying more is the most expensive way to annoy people.

A doubled cost per lead matters far less if the downstream numbers move

Take a business at $40 per lead, booking 8% of leads into a sales appointment and closing 25% of those. Before: $40 ÷ 0.08 = $500 per booked appointment; $500 ÷ 0.25 = $2,000 per sale. After: CPL doubles to $80, lead-to-appointment goes 8% → 20%, close rate unchanged; $80 ÷ 0.20 = $400 per appointment and $400 ÷ 0.25 = $1,600 per sale. Cost per lead doubled and cost per sale fell 20%. The currency does not matter; the ratios do. It is also why a business with a strong lead-to-appointment rate can outbid a competitor for the same lead and still earn more from it.

The levers that move that middle number are contact rate, speed to lead and set rate: how many leads you actually reach and how many of those book a call. In our own client work we typically see speed to lead alone worth around 3x, doubling contact rate about 2x, and doubling set rate about 2x. The honest wrinkle: those do not multiply. 3x × 2x × 2x is 12x, and 12x is not what happens — they overlap, because fixing speed to lead is part of how contact rate improves and contact rate is part of how set rate improves. Our methodology page states a 7x average sales lift, median closer to 4x, measured as trailing three-month closed-deal revenue at month six against the three months before launch; an average is not a forecast. If your database is large, reworking dormant records is usually cheaper per conversation than replacing a lead that just got dearer.

When a rising cost per lead is not a marketing problem

Three of these are common and none is solved by an agency. If your ad account was disabled or restricted on policy grounds, the fix is the platform’s appeals process, not a creative brief — CPL looks infinite because delivery stopped. If your margin moved, and CPL only looks worse because a price or supplier increase changed what you can afford per sale, that is a conversation with your accountant about unit economics before it is one about ads. If the rise followed a consent or list-sourcing change — a purchased list, a new SMS or calling programme, an opt-in you cannot evidence — check the rules with the regulator: the ACMA in Australia, the FCC and FTC in the United States. General information, not legal or financial advice.

Frequently asked questions

Why is my cost per lead increasing when nothing changed in my ad account?

Because two of the four causes happen outside your account. Competitors entering your auction change your price without touching your settings, which the Google Ads auction insights report shows via overlap rate and outranking share. The other invisible cause is measurement: Google reports both observed and modelled conversions, and modelled conversions “use data that doesn’t identify individual users to estimate conversions that Google is unable to observe directly”. Count leads in your CRM before you believe a dashboard.

How much should cost per lead go up each year?

There is no single answer, but there is a published floor. Meta’s Q2 2026 results report average price per ad up 12% year-over-year across its whole network. Treat low-double-digit annual inflation as normal, and anything materially above it as specific to your account, your category or your season rather than to the market.

Should I pause my ads if my cost per lead has doubled?

Only if cost per sale now exceeds gross profit per sale, or you cannot service the leads you already receive. Pausing a conversion-optimised campaign can push it back into a learning phase and raise CPL further. Hold budget flat for 14 days while you run the four checks, changing one variable at a time.

How do I tell creative fatigue from audience exhaustion?

Run the tests in sequence. Duplicate the ad set with one genuinely new creative concept at identical budget for seven days: if CPL recovers, it was creative fatigue. If not, exclude everyone who entered your CRM in the last 180 days and run seven more days: if CPL recovers then, the audience pool was exhausted. The two tests never both pass.

Is a rising cost per lead always bad?

No. CPL is an input metric. If lead-to-appointment rate rises faster than CPL does, cost per booked appointment and cost per sale both fall while CPL climbs — $40 at 8% is $500 per appointment, $80 at 20% is $400. Businesses that convert well can afford to outbid competitors for the same lead.