The 5 Data Druid Reports We Check Every Monday
· Tutorials/Guides

Over the past 2 weeks, one account in our portfolio showed an average frequency of 1.57 per ad set. The people inside that audience were seeing its ads 6.39 times. Nothing in the standard Meta Ads Manager columns flagged the gap, because its 81 ad sets each reported a healthy number while chasing the same broad audience.
That gap is why our Monday review runs through Data Druid, our account intelligence platform, before anyone touches a budget. Ads Manager reports how each ad set performed last week. The 5 reports below show whether an account still finds new people, whether creative produces winners, and where budget sits in the wrong place. We run them in a fixed order because each one sets up the question the next one answers.
Report 1: Trends shows you where to look
The Trends report compares each core metric's 7-day moving average against its 30-day average and logs every crossover. That gives you a one-screen read on whether an account is improving or slipping. Of all 5 reports, this one is the fastest to read and the most likely to give you false comfort if you stop there.
On Account A this week, 7-day CPM sat at $19.99 against a 30-day average of $21.38. Over the same window, 7-day ROAS reached 1.59 against 1.40. Read alone, that looks like an account in recovery.
The same report logged early crossovers on October 5, when the 3-day ROAS average dropped below the 7-day while 3-day CPA climbed above it. That pattern is usually the first sign a recovery is losing pace. We use Trends as triage, pointing us to the accounts that deserve the deeper reports without explaining why any metric moved.
Report 2: The Reach report splits CPMr into its 2 causes
The Reach report puts deduplicated weekly reach next to impressions, frequency and spend. Those inputs give you CPMr, or cost per 1,000 people reached, a metric popularized by Phil Kiel, our Director of Paid Social. CPMr equals CPM multiplied by frequency, which makes it the true cost of reaching 1,000 unique people (full calculation here).
When CPMr climbs week over week, your budget is cycling through people it has already reached. That shows up here weeks before CTR softens and well before ROAS moves.
Account A is the reason we never stop at Trends. Between the week of August 24 and the week of September 14, its CPMr went from $65.51 to $117.96, an 80% increase. Spend rose only 6% over those 3 weeks while weekly reach fell 41%, from 5.83M people to 3.42M.
By the week of September 21, CPMr had eased to $100.30. That figure still sat 53% above late August, even though Trends now shows CPM below its 30-day average.
The headline number only becomes useful once you split it. Frequency rising while CPM holds flat points to audience saturation, which calls for exclusions and consolidation. CPM rising while frequency holds flat points to auction pressure, a different problem with a different fix. On Account A, CPM and frequency climbed together over those 3 weeks, up 34% and 35% respectively, so the account needed work on both sides.
Report 3: Audience Overlap finds the frequency each ad set hides
When the Reach report points to frequency, Audience Overlap shows where it comes from. The report groups every active ad set by the targeting fields that decide who can be reached. It then pulls Meta's deduplicated reach for each group, which gives you the frequency a real person experiences.
Account B, the account from the opening, had 81 ad sets chasing a single broad audience definition over 14 days. Together they spent $262,678. Summed ad set by ad set, they reached 5.94M people. The deduplicated figure was 1.45M, an overlap factor of 4.08. Each ad set reported an average frequency of 1.57, while the true frequency per person was 6.39.
No single ad set in that group would have tripped an alert, which is exactly why this report earns its Monday slot. The usual fix is consolidation into fewer ad sets carrying more creative each. Those 81 ad sets held a median of 3 live ads apiece, so splitting one audience 81 ways added frequency without adding reach.
Report 4: Creative Yield, judged at the same age for every cohort
Creative Yield groups launches by month. For each cohort, it shows how many creatives went out, how many got enough spend for a fair test, how many hit the CPA bar and how many went on to scale. Read the hit rate at day 14. Lifetime hit rate flatters older cohorts, which have had months longer to bring CPA down.
Account B runs against a $75 CPA target. Here is how its last 3 cohorts performed.
Cohort | Launches | Day-14 hit rate | Day-14 CPA | Scaled (over $1k spend) | Scale rate |
July | 328 | 30.3% | $71.14 | 42 | 12.8% |
August | 359 | 26.4% | $81.81 | 40 | 11.1% |
September | 432 | 30.9% | $76.10 | 74 | 17.1% |
September launched 20% more creatives than August while lifting its day-14 hit rate back to 30.9%, with CPA close to target. The number we watch most is the scaled count, which rose from 40 creatives in August to 74 in September.
Scale rate is scaled creatives divided by total launches, a separate measure from hit rate. All 3 cohorts land inside the 10% to 20% range we treat as a healthy scale rate. Volume only counts as progress when yield holds up alongside it. If launches rise while the scale count stays flat, the problem sits in the brief, where more budget will not fix it.
Report 5: Graduation Candidates lists winners still stuck in testing
Graduation Candidates compares your testing containers against your scaling containers. It lists every creative that beat your thresholds in testing but is not yet running where the budget lives. Matching on both creative ID and post ID keeps a copy that already graduated with its engagement intact from being proposed twice.
On Account B, we ran a 30-day window with a $75 CPA ceiling and a $500 minimum spend. The report returned 30 creatives that had beaten target and were still running only in testing ad sets. The top 3 had spent $70,610 between them, proving they work in containers that were never built to scale them.
A fast-track view also flags early winners converting at target below the spend floor. Those carry a low-confidence label, which makes the view a quick scan for day-one breakouts.
This is where the Monday review loops back to Report 3. Graduate winners into an existing scaling campaign instead of a fresh ad set, or you end up feeding the same overlap problem you just found.
Run in order, the 5 reports move Monday from triage to diagnosis to action. Trends picks the accounts that need attention. From there, the Reach report and Audience Overlap explain whether budget is saturating an audience or fighting the auction. Creative Yield and Graduation Candidates close the loop by showing whether the fix already sits inside the account. If CPMr has climbed 3 weeks running on any account you manage, start this week's review there.
Data note:All data anonymized. Figures come from Data Druid for 2 accounts in the Taikun portfolio. Account A Trends cover the 60 days to October 6, 2026. Its Reach report uses weekly 7-day windows from the week of July 27 to the week of September 21, 2026. Account B overlap covers September 23 to October 7, 2026. Creative yield uses monthly launch cohorts from July to September 2026 against the account's $75 CPA target. Graduation candidates cover the 30 days to October 7, 2026. CPMr is calculated as spend divided by reach, multiplied by 1,000, which equals CPM multiplied by frequency.