No pitch until the end. This is the mechanism underneath the money: how the auction settles, what a lead really costs, and why a sales cycle measured in months breaks most of what paid search assumes about a buyer.
Part of the paid advertising system
Nobody at Google sells you a position. Every time someone searches, an auction runs and settles before the page finishes loading. Your ad shows, or it does not, based on what you bid, how well your ad answers the search, and what everyone else did.
Three inputs decide it. Your bid, which is the most you will pay for one click. Your quality score, which is Google's read on how well your ad and your landing page match what was typed. And ad rank, which is roughly those two multiplied together.
The second input is why two schools bidding the same amount do not pay the same price. A school whose ad points at a booking page with a price and a calendar earns a better quality score than a school pointing the same ad at a homepage. A better score buys a higher slot for the same money, or the same slot for less money.
You also do not pay what you bid. You pay just enough to beat the ad below you. Bid $9 against a rival whose ad rank works out to $6, and you pay a little over $6. This is why raising a bid does not always raise your cost per click, and why cutting one can lose you the auction outright.
Here is the whole thing in one search. Someone types discovery flight near me at 8pm. Four schools are eligible. You bid $7 and send the click to a booking page. A rival bids $9 and sends the click to a homepage. Google weighs each bid against its quality read, you come out on top, and you pay about $6.40. The school that bid more sits below you and pays more per click for the privilege.
The median private certificate runs $16,000 to $16,500, and the median student takes about 24 weeks. Those figures come from a 2026 industry survey of 1,180 pilots, 89% of them in the United States.
Most paid search advice is written for things people buy the day they search. Yours is not one of those. Trust builds over the months between the first click and the first payment, and no bid strategy makes that gap shorter.
Aircraft, instructor hours, maintenance downtime, weather and daylight all cap how many students you can fly. A school that cannot absorb more students can be actively harmed by more leads.
That is the reverse of how nearly every other advertiser thinks, and it changes what a correct budget looks like. More is not better once the schedule is full.
Your discovery flight is the paid trial. It is also the only step in the funnel that weather can cancel. A lead who books and never flies looks exactly like a lead who flew, right up until somebody checks.
That single distinction is where most flight school ad reporting quietly goes wrong.
Someone choosing a flight school hands over their money, their calendar and their physical safety at the same time. No other local service asks for all three.
It is the reason pages that answer the fear directly beat pages that list features, and the reason a named instructor on camera outperforms a stock photo of an airplane.
Those four things are why a campaign that would be judged a failure in most industries can be the best money a flight school spends, and why the reverse is also true. To see which one you have, you have to look further down the funnel than the platforms can see.
Five steps run from a click to a certificate. Most schools measure two of them, and the three in the middle are where the money goes.
The chain is: inquiry, discovery flight booked, discovery flight flown, enrolled, certificate finished. Booked and flown are different numbers. In most schools, nobody ties either one back to the ad that produced it.
This matters more than any targeting decision you will make. A campaign can look excellent at the lead stage and be losing you money three steps later. Platform reporting cannot see far enough down the chain to tell you which one you have.
Here is the honest part. The single most important number in this business, how many discovery flights become enrolled students, has never been published by anyone. Every figure in circulation, including the 20% to 40% you will see quoted, is a vendor assertion with no sample, no source and no date. We do not know it either. The difference is that we will measure yours.
So the first job on a new account is rarely a campaign change. It is getting the click identifier onto the form, the form into the CRM, and the flown flight back out of the scheduler. Until that loop closes, every other decision is a guess wearing a dashboard.
The Ultimate Flight School Keywords Guide lists the terms prospective students actually search, sorted by intent, along with the ones that look like students and are not. It is a keyword list, not a sales PDF with a keyword list inside it.
Owners ask how much they should spend, and most agencies answer with a number and no working. Here is the working, so you can run it against your own school.
There is no published cost per lead benchmark for flight training. The closest defensible proxy is Google's Education and Instruction category, drawn from 13,474 US campaigns between April 2025 and March 2026. That category blends tutoring and online courses, so treat it as a ceiling to plan against rather than a number to expect.
Run the arithmetic at both ends. A $4.81 click at a 13.14% conversion rate is about $37 per lead. A $6.23 click at 11.38% is about $55. Push it to an aviation head-term bid of $7.39 at a more cautious 8% and you get about $92.
So a lead costs somewhere between $37 and $92 in media. Google's own guidance is that a bid strategy needs at least 30 conversions in 30 days before anyone can judge it. That puts the floor near $1,100 to $2,800 a month in Google Search media alone. It is why we suggest starting at $3,000 across channels, and why a $500 test proves nothing except that $500 was not enough.
Then keep walking the chain. Cost per lead divided by your show rate gives cost per discovery flight flown. Divide that by your close rate and you have cost per enrollment. Two of those numbers can only come from your own records, which is the whole argument for measuring them.
For a downstream check, two publicly traded career-education companies disclose both advertising spend and new student starts in their annual filings. Across FY2023 to FY2025 that works out to $2,310 to $3,193 of media per new student start, over roughly 50,700 starts. It is a proxy from trade and technical education rather than flight training, and it counts media only, not admissions staff. It is also audited, which is more than any number a competitor will quote you.
Return on ad spend is the number every agency quotes, and the one number nobody in flight training can honestly produce inside a reporting window.
The argument runs in three steps. First, your revenue arrives over 6 to 24 months, across dozens of hourly transactions, rather than in one payment.
Second, the attribution windows close long before the money finishes arriving. Google's click-through conversion window maxes out at 90 days. Meta's standard Conversions API rejects any event older than 7 days, and errors the whole request when you send one.
Third, and this is the part that costs schools money: a 30-day return figure on a healthy flight school account reads as failure. A student who clicks in March, flies in April, enrolls in May and pays across the next six months adds almost nothing to any return the platforms can compute in month one.
If an agency quotes you a return on ad spend multiple in month two, ask which conversion window produced it. There is not one that could.
Cost per thousand impressions, link click-through rate, cost per link click, cost per lead. These say whether the machine is running.
They are not performance. Treating them as performance is how good accounts get taken apart in well-meaning Monday meetings.
Cost per qualified lead, cost per discovery flight flown, show rate, speed to lead, and students per aircraft.
Show rate settles most arguments about lead quality, and it can only come from your records. It is the number we ask for first and the one schools most often cannot produce yet.
Cost per enrollment, fully loaded acquisition cost with agency fees included, and lifetime value against acquisition cost by path.
Reported by click cohort rather than by calendar month. A calendar month mixes students who clicked at wildly different times, which is exactly the error that makes a good quarter look flat.
This is a limit built into your lead volume, not a sales position. Knowing it will save you from a lot of expensive advice.
At 2 to 8 leads a week, detecting a 20% improvement in conversion rate takes between 1.8 and 8 years at standard confidence and power. That is arithmetic, not pessimism. The sample does not arrive fast enough to see a change that small.
Checking a test daily makes it worse. In a simulation where every apparent winner is noise by design, daily peeking over 90 days pushes the false positive rate from 5% to somewhere between 27% and 37%. Look at random data often enough and you will always find a winner.
Meta has the same problem from the other side. Its learning phase wants about 50 optimization events per ad set per week. A school producing 8 leads a week delivers 16% of that with its entire lead flow aimed at a single ad set.
So the working rule is this. Do not test anything you do not believe could move the number by 30% to 50%. Below that, your account cannot tell a real gain from noise, and neither can anyone selling you one.
Capture the Google and Meta click identifiers on every form, store them in your CRM, and send the outcome back when a student enrolls. This is the only way an enrollment ever reaches the campaign that caused it.
The windows are hard limits. Google accepts an offline conversion against a click identifier for 90 days, and enhanced conversions for 63. Past that the click is gone and the enrollment cannot be attributed by any native mechanism.
Flight training inquiries skew heavily toward calls. An untracked phone number is the most common reason a paid search account looks like it is failing when it is working fine.
One mechanic worth knowing: a Google forwarding number keeps connecting for 60 days after a caller's first call, and the clock resets each time they call again.
Asking how did you hear about us is worth doing, but not for judging paid social. One study found 20 people naming Facebook against 195 tracked Facebook conversions, an undercount of roughly ten to one.
That figure is from business software and it is old, so treat the direction as real and the size as unknown. Use the question to find word of mouth and referrals. Do not use it to evaluate Meta.
PPC buys a position in the auction at the top of the results page, and the visibility stops the day you stop paying. SEO earns a position in the organic results below it, takes months to build, and keeps working after the invoice stops. For a flight school the practical split is that paid search covers you now, while SEO and your Google Business Profile lower what you have to pay later. Schools running both usually see cost per lead fall over time, because the organic listing and the ad reinforce each other on the same search.
Plan on $37 to $92 in media per lead, based on the Education and Instruction proxy benchmark and aviation head-term bids. Google needs at least 30 conversions in 30 days before a bid strategy can be evaluated, which puts the practical floor around $1,100 to $2,800 a month in Google Search alone. We suggest starting at $3,000 a month across channels. A $500 monthly test does not succeed or fail, it simply cannot produce enough data to tell you anything.
Leads arrive within days of launch. A readable account takes about a quarter, because at 2 to 8 leads a week you cannot reach statistical significance monthly. Payback depends on your numbers rather than ours. A school that converts discovery flights well can cover a month of media with a single enrollment, and a school with a weak show rate can spend for a quarter and see very little. That is why we instrument the show rate from the first week.
Yes, and some owners do it well. The work that decides the outcome is unglamorous: a negative keyword list, targeting set by drive time rather than a map radius, a landing page with a price and a scheduler, and the offline conversion loop that sends enrollments back to Google. If you have a few hours a week and the patience to leave a campaign alone long enough to learn something, running it yourself is a real option. Most owners hire it out because the account punishes inattention, not because the buttons are hard.
Nobody has published a credible answer, and any agency quoting you a discovery flight conversion rate is guessing. The figures in circulation carry no sample, no source and no date. What we can tell you is how to work out your own: count the discovery flights flown in a period, count how many of those students enrolled, and divide. Once you have that ratio you can price a lead properly, and every budget decision after it becomes arithmetic instead of opinion.
Pick by the symptom you actually have, not by the channel that sounds most appealing.
That is a search problem. Google Ads for flight schools covers the keyword landscape, the intent clusters worth paying for, and the negative list that keeps you out of the wrong auctions.
That is a reach problem. Facebook and Instagram ads run about a third of Google's cost per lead, with a lead quality tradeoff and a compliance exposure most agencies will not raise with you.
That is a trust problem. YouTube ads reach someone before they know they are a buyer, and they are the only place you can answer the safety question at the length it deserves.
This page is the mechanism. If you want the layer above it, which campaign goals to pick and how to split a budget between them, the owner's guide to PPC advertising covers that ground.
If you would rather have someone walk your account and tell you which of the three problems above you actually have, that is what the call is for.