Every morning the Ads Optimisation agent reads 7 and 28 days of paid performance per ad and proposes pauses, scales, budget shifts and creative refreshes against each plan's objective and cap. Each one is a gate with the numbers and the rule behind it. Nothing changes until you decide.
Today’s proposals · 4
0 of 6 raised today · the agent is working now
Ads Optimisation Agent · Tuner proposes · Scale up
Mook — first bite (Spark Ad): THB 5,953.10 → 7,143.72 a day
livecostly to reverse
Mook · first bite (Spark Ad) buys the cheapest view on the TikTok account: THB 0.15 per view over 7d and THB 0.15 over 28d, against an account median of THB 0.49 (all simulated). That is 0.31× the median, well inside the 0.6× scale trigger. It is out of learning (29 days, 2,102,824 impressions, simulated). Raise the daily budget from THB 5,953.10 to THB 7,143.72 · a 20% rise, exactly the per-ad daily change limit; no plan cap applies to this ad. No compliance scan is on file for this post; a human published it. Costly to reverse: spend already delivered is not refundable.
Rule
cost per result 0.31× the account median after learning (scale below 0.6×); raised 20%
THB 0.49 · median across the account's ads with results
Learning
learned · 29 day(s), 2,102,824 impressions · guardrails: 3 days or 1,000 impressions
Compliance
no scan on file for this post; a human published this ad · compliance_scans, latest verdict
Expected effect
+20% on views · Assumes cost per result holds at the higher budget; auction prices usually rise with spend, so the true lift is likely lower. Not observed yet; the rail is simulated.
Asked
07/09/2026 23:04
Ads Optimisation Agent · Tuner proposes · Scale up
Mook — first bite (Spark Ad): THB 5,953.10 → 7,143.72 a day
livecostly to reverse
Mook · first bite is the cheapest view on this TikTok account: THB 0.15 (simulated) cost per view over 7d and 28d, versus an account median of THB 0.49 (simulated) · a 0.32× ratio, well under the 0.6× scale threshold. It's past learning (29 days, 2,100,548 impressions, simulated). Raising daily budget from THB 5,953.10 to THB 7,143.72 (a 20% rise, at the per-ad daily change limit) buys more of a result that's already cheap. Note: no compliance scan is on file for this post · a human published it, so this is flagged per guardrails rather than blocking the scale (compliance verdict is "none"/not a fail or review).
Rule
cost per result 0.32× the account median after learning (scale below 0.6×); raised 20%
THB 0.49 (simulated) · median across the account's ads with results
Learning
learned · 29 day(s), 2,100,548 impressions (simulated) · guardrails: 3 days or 1,000 impressions
Compliance
no scan on file for this post; a human published this ad · compliance_scans, latest verdict
Expected effect
+20% on views · Assumes cost per result holds at the higher budget; auction prices usually rise with spend, so the true lift is likely lower. Not observed yet; the rail is simulated.
Asked
06/09/2026 23:01
Ads Optimisation Agent · Tuner proposes · Scale up
Mook — first bite (Spark Ad): THB 5,953.10 → 7,143.72 a day
livecostly to reverse
Mook · first bite (Spark Ad) is the cheapest view on this TikTok account: THB 0.15 cost per view (simulated) against an account median of THB 0.49 (simulated), a ratio of 0.30× · well under the 0.6× scale threshold. It's past learning (29 days, 2,105,977 impressions, simulated). Raising daily budget 20%, from THB 5,953.10 to THB 7,143.72 (simulated), puts more money behind the result that's already working. No compliance scan is on file for this post; a human published it, and there's no fail/review verdict blocking the raise.
Rule
cost per result 0.30× the account median after learning (scale below 0.6×); raised 20%
THB 0.49 (simulated) · median across the account's ads with results
Learning
learned · 29 day(s), 2,105,977 impressions (simulated) · guardrails: 3 days or 1,000 impressions
Compliance
no scan on file for this post; a human published this ad · compliance_scans, latest verdict
Expected effect
+20% on views · Assumes cost per result holds at the higher budget; auction prices usually rise with spend, so the true lift is likely lower. Not observed yet; the rail is simulated.
Asked
05/09/2026 23:09
Ads Optimisation Agent · Tuner proposes · Scale up
Mook — first bite (Spark Ad): THB 5,953.10 → 7,143.72 a day
livecostly to reverse
Mook · first bite (TikTok Spark Ad, kfc-zinger-ugc) is the cheapest view on the TikTok account: THB 0.15 cost per view both 7d and 28d (simulated) against an account median of THB 0.50 (simulated, 7d) · a 0.30x ratio, well under the 0.6x scale threshold. It's past learning (28 days live, 2,031,170 impressions to date, threshold is 3 days or 1,000 impressions). Raising the daily budget from THB 5,953.10 to THB 7,143.72 is a 20% increase · the maximum allowed per-ad daily change · and puts more budget behind the result that's already working. Compliance verdict is "none" (no fail/review on file), but note: no compliance scan is on file for this specific post; it was published directly by a human, so this isn't a blocked case but also isn't a confirmed clean scan.
Rule
cost per result 0.30× the account median after learning (scale below 0.6×); raised 20%
THB 0.50 (simulated) · median across the account's ads with results
Learning
learned · 28 day(s), 2,031,170 impressions · guardrails: 3 days or 1,000 impressions
Compliance
no scan on file for this post; a human published this ad · compliance_scans, latest verdict
Expected effect
+20% on views · Assumes cost per result holds at the higher budget; auction prices usually rise with spend, so the true lift is likely lower. Not observed yet; the rail is simulated.
Asked
05/09/2026 09:16
Per-ad performance
simulated7 and 28 days from the stored daily ad metrics · cost per result against the account median · learning per guardrails (3 days or 1,000 impressions)
Ad
Objective
7d spend
7d results
7d cost / result
28d cost / result
vs median
Learning
Compliance
Daily budget
Mook — first bite (Spark Ad)
tiktok · live
video views
THB 42,823
279,182
0.15
0.15
0.31× · under the scale line
learned · 29 days
no scan on file
THB 5,953.10
JJay — no bun (Spark Ad)
tiktok · live
video views
THB 22,697
26,954
0.84
0.83
1.69×
learned · 29 days
no scan on file
THB 3,244.50
Brand cut 16:9 — Reels
instagram · paused
video views
THB 28,273
101,162
0.28
0.29
1.00×
learned · 17 days
no scan on file
THB 3,838.48
simulated Attributed orders are counted where the brand holds commerce rows. The median is per connected account across ads with results in the window.
Applied log
0 changes this week · 0 denied · 0 reverted
No change has been applied, denied or reverted yet.
Weekly summary
The agent writes one at the end of each run. None saved for this brand yet.