The Meta Ads Scorecard
This scorecard tracks the eleven numbers that decide whether Facebook and Instagram spend keeps producing results: creative supply and fatigue, cost trend against your own history, landing page conversion, audience overlap, learning-phase share, and pixel and Conversions API match quality. The media buyer runs it, the creative lead and marketing lead own their rows, and the whole thing takes twenty minutes a week.
Why this scorecard exists
Why Meta needs its own scorecard
A generic scorecard tracks spend, leads, and cost per lead, then calls it a day. On Meta those three numbers are outputs of things the scorecard never sees: how much fresh creative entered the account, how many ad sets are stuck in learning, and whether the pixel is actually matching the people who converted. Cost per lead climbs for three weeks before anyone opens the creative tab. By then the fix is a full rebuild instead of a Tuesday swap.
Meta is a creative-fed system. Google rewards the right keyword; Meta rewards the ad people stopped scrolling for, and that ad has a shelf life. So the scorecard has to measure supply and age, not just results. It also has to measure the plumbing, because Ads Manager can only optimize toward the conversions it can see. A weak match quality score quietly changes who the algorithm goes looking for, and no amount of budget fixes that.
When this scorecard is live, the media buyer knows by Wednesday whether the account needs new creative, a consolidated audience, or a pixel repair, and each of those has a named owner. Arguments about whether Meta is working turn into a review of eleven rows against your own trailing median. Budget decisions get made on trend, not on the last bad day.
The scorecard
11 numbers. Every one has an owner.
| Metric | What it means | How to set the target | Owner | Cadence | Red flag |
|---|---|---|---|---|---|
| Frequency (7-day, top spending ad sets) Rising frequency is the earliest signal that an audience is worn out, and it shows up before cost per result moves. | The average number of times one person in the audience saw your ads over the trailing seven days, pulled for the ad sets carrying most of the spend. | Take the frequency your best performing month ran at and treat it as the ceiling for that audience size. Prospecting and retargeting get separate ceilings, because a retargeting ad set is supposed to run higher. | Media buyer | Weekly | Frequency above the ceiling for two consecutive weeks while cost per result rises triggers a creative refresh or an audience expansion on that ad set. |
| Share of spend on ads older than 30 days It puts a number on creative fatigue at the account level, so the team sees the whole book aging instead of noticing one tired ad at a time. | The percentage of the week's spend that ran through ads first launched more than thirty days ago. | Pull the last six months of ad launch dates and spend, find the share that held while cost per result was stable, and set the target just under it. Adjust the age threshold to your own data if your ads reliably decay faster or slower than thirty days. | Media buyer | Weekly | Share above target for two consecutive weeks moves the next creative batch to the top of the creative lead's queue and adds it to the Friday launch list. |
| New creatives launched this week Fatigue is a supply problem, and this is the supply row; without it the fatigue rows only tell you what already went wrong. | The count of net-new ads (new hook, new asset, or new concept, not a copy tweak) that went live in the account this week. | Count how many creatives per week the account was launching during your most stable stretch of cost per result, and hold that pace as the floor. Raise it when the aged-spend row runs hot. | Creative lead | Weekly | Zero launches in a week, or two consecutive weeks below the floor, puts creative production on the weekly review agenda with a named blocker. |
| Hook rate (3-second plays per impression) It tells you within days whether a new creative earned attention, so the team can kill or scale it long before cost per result has enough data to say anything. | For video ads, the share of impressions that turned into a three-second play, measured per creative in its first seven days. | Rank every video creative from the last ninety days by hook rate and use the median as the bar. New creatives that fall below the median of your own past winners get cut early instead of spending to prove it. | Creative lead | Weekly | A new creative below the account's median hook rate after its first seven days and meaningful impressions gets paused and its concept goes back to the brief. |
| CPM vs trailing 8-week median CPM separates an auction problem from a creative problem; if CPM jumped and click-through held, the fix is not a new ad. | This week's cost per thousand impressions divided by the median CPM of the prior eight weeks, expressed as a percentage change. | Use your own trailing 8-week median as the baseline and set a tolerance band from the normal week-to-week swing in your history. Widen the band around known seasonal auction spikes so the row does not cry wolf every Q4. | Media buyer | Daily | CPM outside the band for three consecutive days starts a check of placements, audience size, and any new campaign competing for the same people. |
| Cost per result vs trailing 8-week median It is the number the business actually cares about, and comparing to your own median instead of a memory of the best month keeps everyone honest. | This week's cost per primary result (lead, purchase, or booked call) compared to your median cost per result for the prior eight weeks. | Set the target at the trailing 8-week median and tighten it ten percent a quarter while creative supply and match quality are healthy. Reset the baseline after any change to the conversion event or the pixel setup. | Media buyer | Daily | Two consecutive weeks above target opens the fatigue, learning-phase, and match quality rows in that order before anyone touches budget. |
| Landing page conversion rate Half of what looks like ad performance is page performance, and this row keeps the media buyer from rewriting ads to fix a slow form. | Results divided by landing page views for Meta traffic, tracked per landing page, using landing page views rather than link clicks so lost loads are excluded. | Take each landing page's trailing 8-week conversion rate from Meta traffic as its baseline. A new page has to beat the baseline of the page it replaces before it takes more than a test share of spend. | Marketing lead | Weekly | A drop of more than the normal weekly swing on a page that took no creative change triggers a page load, form, and tracking check the same day. |
| Audience overlap between active ad sets Overlapping ad sets bid against each other, spread learning thin, and make frequency look fine per ad set while the person on the other end sees the ad five times. | The share of people reachable in one ad set who are also reachable in another active ad set, pulled from the Audience Overlap tool for the top spending pairs. | Map overlap across your top spending ad set pairs at setup and treat that as the starting point. Any pair that overlaps more than your least overlapping pair by a wide margin is a consolidation candidate, and the goal each quarter is fewer ad sets, not more. | Media buyer | Monthly | Any pair whose overlap grows month over month while both stay active gets consolidated or given an exclusion before the next budget change. |
| Learning-phase share of spend Spend in learning is spend the algorithm is still guessing with, and a high share usually means too many ad sets, too many edits, or budgets too small to exit. | The percentage of the week's spend that ran in ad sets marked Learning or Learning Limited. | Look at the weeks where cost per result was steady and note the learning share they carried; that is the ceiling. Structural changes push the share up for a week by design, so the row is judged on the following week. | Media buyer | Weekly | Learning share above the ceiling for two consecutive weeks freezes non-essential edits and starts an ad set consolidation review. |
| Event match quality (pixel and Conversions API) The algorithm can only find more of the people it can match, so a falling score changes who sees your ads before it changes any number in Ads Manager. | Meta's match quality score for your primary conversion event, plus the share of that event arriving through the Conversions API and being deduplicated against the pixel. | Record the score at setup and treat any drop as a defect, not a trend. The setup goal is to send every customer identifier you already collect (email, phone, name, location) through CAPI, and the ongoing target is no decline from your own baseline. | Marketing ops lead | Weekly | A score drop, a missing CAPI event, or a deduplication warning in Events Manager triggers a tracking fix ticket before the next creative launch. |
| Ads Manager results vs CRM-verified results Ads Manager counts what the pixel saw, not what the sales team received, and the gap between the two is where budget quietly leaks. | The number of results Ads Manager reports for the week divided into the number of leads or sales your CRM can tie to Meta over the same window. | Compute the ratio for the last eight weeks and treat the median as normal. The goal is a stable ratio you can plan around, and any move outside the historical swing is a tracking or lead quality question before it is a media question. | Marketing lead | Weekly | A ratio outside its historical range for two consecutive weeks starts a joint review of the conversion event definition, form spam, and CRM source tagging. |
Targets are set from your own history, never from someone else's benchmarks. The builder does the arithmetic once you plug in your numbers.
The weekly review
Twenty minutes, same order, every week.
// 01
Cost per result vs your own median
Open with the one number the business cares about, shown against the trailing 8-week median and not against last week. The media buyer states whether it is inside the band, and if it is not, says which of the next three rows explains it.
// 02
Creative supply and fatigue
Frequency, aged-spend share, new launches, and hook rate get read together. The creative lead reports what shipped, what is in production, and what the hook rate said about last week's launches, and the media buyer names which ads are getting paused.
// 03
The plumbing
Match quality, learning-phase share, audience overlap, landing page conversion, and the CRM ratio. The marketing ops lead and marketing lead each own a row here, and this is where the review catches the problems that look like media problems but are not.
// 04
Decisions and owners
Every red-flag row leaves the meeting with an action, an owner, and a date, written in the scorecard notes. Budget only moves if the fatigue and plumbing rows are clean, and next week's review opens by checking whether last week's actions closed.
Where scorecards die
Mistakes we see constantly.
Judging a new creative on cost per result in week one
A new ad has too few results in seven days to say anything about cost per result, so teams either kill winners early or let losers spend. Hook rate and click-through against your own medians are the week-one call; cost per result is the week-three call.
Letting a winner run until it dies
The best ad in the account keeps its budget because it still works, and nobody notices frequency doubling until cost per result follows. The aged-spend row exists so the replacement is in production while the winner is still winning.
Slicing audiences into overlapping ad sets
Five ad sets that reach the same people look like testing but act like one ad set bidding against itself with a fifth of the learning data. Fewer, larger ad sets with real exclusions almost always exit learning faster and read cleaner on the scorecard.
Treating Ads Manager as the source of truth
Ads Manager reports what the pixel matched, and the pixel does not know about spam forms, duplicates, or the phone call that came from the ad. The CRM ratio row is the only thing that keeps the scorecard tied to revenue.
Chasing someone else's benchmark
A CPM or cost per lead number from a blog post has nothing to do with your offer, your audience, or your landing page. Every target on this scorecard comes from your own trailing history, and that is the only comparison the team is allowed to make.
Other scorecards
Same discipline, different numbers.
By platform
By business type
Want it set up live in your dashboards?
We install scorecards on real data, wire the owners and alerts, and coach the weekly review. Scorecards and accountability is the service.
Questions we get
Straight answers.
Which conversion event should the scorecard be built on?
The one closest to revenue that still fires often enough to optimize on, which for most lead-driven businesses is a form submit or booked call and for ecommerce is a purchase. Pick one, name it on the scorecard, and reset every baseline if you ever change it.
How do we set targets if the account is new and has no history?
Run the first eight weeks with the rows in place but no targets, then set every target from the median of that window. A new account will spend more time in learning and see more swing, so the bands start wide and tighten each quarter.
Does the scorecard change if we use Advantage+ campaigns?
The rows stay the same, but frequency and audience overlap are read at the campaign level since Meta controls the targeting. Creative supply matters more, not less, because the system has fewer levers left and the ads carry the whole job.
Why is match quality on a marketing scorecard instead of a developer checklist?
Because it drives who the algorithm looks for, which makes it a marketing outcome with a technical cause. Giving it a row and a weekly owner is the only way we have found to keep it from silently decaying after the site gets updated.
Who runs the review if there is no in-house media buyer?
Whoever is closest to the account runs it, whether that is the owner, a marketing lead, or an outside partner, and the owner column still names one person per row. If we manage the media, we run the review and the client owns the accounts, the data, and the scorecard.
A scorecard is only as good as the system feeding it.
Start with a diagnostic call. We will tell you which of these numbers you can trust today and what it takes to trust the rest.