Key takeaways
- A funnel multiplies, so one stage 25% under benchmark removes a quarter of the customers while every other number on the sheet still looks healthy.
- A high CPM is a narrow audience, not a broken account. Forecasting on the published average shows four times the volume and is wrong in the direction that sells.
- The Meta CTR benchmark most B2B articles quote is measured on a different channel. On Meta the 2026 median is 2.19%, not 0.7%.
- The cheapest number on the whole sheet is the minutes between a form submission and a phone ringing. Five minutes closes at 32%, twenty four hours closes at 12%.
- Month one is usually zero. The benchmark describes month two, sometimes month three, and the weeks in between should climb rather than sit flat.
A Meta ads funnel benchmark is the conversion rate a single stage should produce in an average month, from CPM at the top to close rate at the bottom. It describes a month once the account is working, not the first month, which is usually zero.
You have seen the sheet. Five funnel shapes, three rows each, a minimum in amber and a benchmark in green, and a number at the end that says how many customers a month of spend should produce. It sits at the bottom of my services page and people play with the sliders for a while before booking a call. Then on the call they ask the only question that matters: where do these numbers come from, and what happens if mine are lower. This article answers both, stage by stage, and tells you what to do about each one.
Here is the sheet itself, so you do not have to keep a second tab open. It is the same component that runs on the services page, so it is never out of date with it.
A funnel multiplies, which is why the leak is never where you are looking
Nobody reads a funnel as multiplication, and that is the whole problem. Every stage takes a percentage of the stage above it, so the final number is the product of eight or nine decimals. Move one of them and the bottom moves much further than your intuition says it should.
$5,000 a month, SDR to closer
Here is the arithmetic on a real template. The SDR to closer funnel at $5,000 a month produces 66,667 impressions, 1,467 clicks, 1,173 visits, 235 leads, 127 people actually reached on the phone, 70 demos booked, 42 attended, and 13 customers. Drop the connect rate from 60% to 45%, which is a fifteen point move on a single stage nobody photographs for a dashboard, and the bottom goes to 9 customers. That is a quarter of the revenue, gone, and the ad account still looks completely normal. CPM is fine. CTR is fine. Cost per lead is fine. The reports everyone stares at are all green.
So when your numbers come in under the sheet, do not start at the top. Divide every stage by its benchmark, sort the list, and go to the worst ratio first. It is almost always one stage carrying most of the damage, and it is almost never the one being discussed in the meeting.
CPM: mine is triple the published figure and that is the safe direction
The published Meta CPM for B2B and SaaS in 2026 runs somewhere between $16 and $22. The sheet assumes $54 to $99 depending on the funnel. That gap is the biggest single deviation on the page and it is deliberate.
$5,000 buys 0 impressions
$5,000 buys 0 impressions
Narrow B2B audiences, static images, and a frequency cap. The blended average is set by broad ecommerce, which is not the account being forecast.
Three things make an impression expensive: a small audience, a static image instead of video, and a frequency cap that stops you showing the same person the same ad nine times. Every account on this sheet has all three. The blended benchmark is dominated by broad consumer ecommerce, which is a genuinely different auction, and quoting it at a B2B founder is how forecasts end up four times too optimistic.
The direction matters more than the number. A high CPM shrinks every count below it, so if I am wrong about CPM, the forecast is too pessimistic. That is the only kind of wrong worth being in public. If you want the opposite view, run the numbers at $19 and watch the customer count quadruple. That version would be much easier to sell and I would not be able to defend it on a call.
When your CPM is genuinely too high: the fix is almost never the creative. It is audience size, placement mix, and frequency. A CPM that climbs week over week on a fixed audience is frequency, and the answer is more creative variants, not a bigger bid.
CTR: most B2B benchmark posts are measuring a different channel
This is the number I was wrong about, so it is worth being precise. If you search for B2B CTR benchmarks you will find 0.6% to 0.8% quoted everywhere. That figure is real, and it is measured on search and LinkedIn. On Meta the 2026 median across industries is 2.19%, rising to 2.59% on lead generation objectives. Technology sits lower at about 1.04%.
That is a threefold difference caused entirely by quoting the wrong channel, and it will make a Meta account look broken when it is performing normally. Every CTR on the sheet sits between 1.8% and 3.6%, which is inside the Meta band, and the demo-led funnel is deliberately set below the median because high intent B2B offers do not get scrolled into the way a consumer offer does.
When your CTR is under 1%: on Meta that is a creative problem, not a targeting problem, and it is the fastest thing on this list to fix. It is also the reason I run an engine that makes a hundred ads from one brief rather than four.
Page loading: the visitors you paid for and never met
Every sheet has a stage called page loading at 80%, with a 70% floor, and it is the one people skip past. It is the share of clicks that survive the trip to your page. The other 20% tapped the ad, waited, and left before anything rendered.
You paid for all of them. On a $5,000 month at a 2.2% CTR that is roughly 290 people a month who cost you money and saw nothing. It is the only number on the sheet that is fixed with engineering rather than marketing, and it is usually fixed in a day: compress the hero image, drop the third party scripts nobody audits, and stop redirecting the ad click through a tracking hop.
Anything under 70% is not a benchmark problem, it is a broken page, and no amount of creative testing above it will help.
The middle stage is what actually defines your funnel
Everything so far is the same for every business. The middle is where the five shapes on the sheet separate, and picking the wrong one is the most common mistake people make with the tool.
- Self-serve SaaS converts visits to signups at 15%. That sounds high against the 2 to 5% figure quoted for SaaS websites, because that figure is blended across organic traffic. Opt-in trials from paid traffic convert at about 17.4%, and paid traffic is what this sheet is about.
- Demo-led sales takes 20% of visits to a lead and 29% of those to a booked demo. Two separate stages, because asking for a call and turning up for one are different decisions.
- SDR to closer takes 20% of visits to a contact form. This is my number, not a published one, and it is the thinnest figure on the sheet.
- Digital course takes 31% to an opt-in and 20% of those to an initiated checkout.
- Webinar or launch takes 35% to a registration and 42% of those into the live room.
Choose by shape, not by industry. The question is how many steps sit between an impression and money, and whether a human is on any of them. A dental practice and a B2B services firm run the same shape.
Connect rate: the cheapest number on the entire sheet
If your funnel has a phone in it, this is where the money is, and it is the section I would read twice.
Called within 5 minutes
0
close
Called after 24 hours
0
close
A study of 939 companies found leads contacted within five minutes close at 32%, against 12% for leads contacted after twenty four hours. Same leads, same offer, same sales team. The only variable is how long the form sat in an inbox. Nothing else on this sheet moves that far for free.
The sheet assumes 90% of leads get called at all and 60% of those get reached. Cold outbound connects at 5 to 8%, but this is not cold outbound: these people filled in a form and asked to be called. 60% over a same day, multiple attempt cadence is optimistic and reachable. 70% is not, unless they recognise the number on the screen.
The fix is operational, not strategic. A notification that fires on submission, a rule that the first attempt happens inside five minutes during business hours, and three more attempts across forty eight hours. That is it. I have watched this single change do more for an account than a month of creative work, which is a slightly humiliating thing for someone who sells creative to admit.
Close rate: why 12% and 30% are both correct
Here is where the obvious answer is wrong. People look at the sheet, see demo-led sales closing at 12% and SDR to closer at 30%, and assume one of them is a typo. Neither is. They are measuring different pipelines.
Published qualified opportunity win rates land at 15 to 30%, with about 29% for qualified opportunities only. Demo-led at 12% sits below that band on purpose, because a self-serve inbound demo request has not been screened by anybody. Half the people on those calls were never going to buy and the closer finds out in minute four.
SDR to closer sits at 30% because an SDR already threw the bad ones away. That is the entire point of the role, and it is worth saying clearly because most people hire an SDR to book more meetings. The client this template is modelled on hired one to book fewer. The closer’s calendar got emptier and the close rate roughly doubled.
So a low close rate is often a lead quality problem wearing a sales costume. Before you retrain anybody, check what share of booked calls were qualified against a written definition. If there is no written definition, that is the finding.
CAC, LTV and the only ratio that decides anything
The four tiles under the sheet are customers, cost to acquire each one, revenue, and LTV to CAC. The last one is the only one that tells you whether to keep spending.
CAC is the month’s budget divided by the customers it produced. LTV depends on whether the sale repeats. For a one-off, the tool values a customer at exactly one purchase, which is the harshest possible reading and deliberately so. For a subscription it is your price multiplied by how long a customer stays, and that comes from the churn slider: months of life equals 100 divided by monthly churn percent.
The default is 10% churn, which means ten months. Published median monthly B2B SaaS churn is about 3.5%, and even SMB self-serve runs 3 to 7%. So the default assumes a customer worth roughly a third of what the median business actually gets. If your real churn is 4%, move the slider and watch the ratio nearly triple.
Three is the number people quote as healthy. Under 1 means the ads cost more than the customer is worth and you should stop. Between 1 and 3 usually means the price is wrong rather than the ads, which is a conversation about the offer and not about Meta. Above 5 with a sane payback period generally means spend more.
What the first eight weeks actually look like
None of the above happens in week one, and this is the part I insist on saying before anyone signs anything.
Month one is usually zero. The budget is buying information: which angle a stranger responds to, which audience the delivery system should be hunting in, which landing page survives contact. That is not a waste, it is the only way to get the data, but it does mean the first invoice buys learning rather than customers.
The line gets crossed in month two. In a December 2025 study of 14 client accounts, all of them starting from zero without a validated offer, 12 hit benchmark numbers before the end of the second month. The other two landed in month three. What you should see in between is the weekly numbers climbing, every week, toward the line. Flat weeks are the warning sign, not low weeks.
You can see what that looked like on real accounts: HyperTask took a $280 cost per lead down to $10 over thirty two weeks, Thryve Clinic went from 175 leads and zero sales to a 7.96% page-to-lead rate in four weeks, and Fuse Health filled a CEO’s calendar six weeks out on $100 a day. The full set is here, including the ones that took longer than anybody wanted.
Frequently asked questions
Why is your CPM three times the published benchmark?
Because the published average is set by broad ecommerce, and the accounts on this sheet are not that. Narrow B2B audiences, static image ads and a frequency cap all push the price of an impression up. Assuming the expensive impression also shrinks every number below it, which is the honest direction for a forecast to be wrong in.
If month one is usually zero, what am I paying for?
Information. Month one buys the data that tells the delivery system which people to find and tells us which angle a stranger responds to. The budget is not wasted, it is spent on learning, and the sheet only describes what happens after that learning is done. Anyone promising sales in week one is selling you the wrong thing.
My numbers are under every line on the sheet. Is the account broken?
Probably not, and one stage is usually responsible for most of it. A funnel multiplies, so a single stage at 45% instead of 60% removes a quarter of the customers at the bottom while every other number still looks fine. Find the worst ratio against its benchmark first, and fix that one before touching anything else.
Are these your results or are they industry averages?
Both, and the sheet says which. Most rates are published 2026 figures held still. CPM, the contact form rate and the connect rate are mine, from running the accounts, and they override the published numbers where the two disagree. The blend is stated on the page rather than hidden, because a prospect finding it themselves is worse.
What if my business does not match any of the five funnels?
Pick the one with the same shape rather than the same industry. What matters is the number of steps between an impression and money: no call, one call, two calls, a live room, or a checkout. A dental practice and a B2B agency both run the SDR to closer shape. The industry changes the rates, not the structure.
Run your own numbers first. The sheet is free, nothing is fetched, and it takes about a minute. If the ratio at the end is under three, bring it to a Revenue Diagnostic and we will find which stage is doing the damage. More on how I work is on the consultancy page, and the rest of the writing is in the blog.
Open the benchmark sheet