Key takeaways
- Close every Q3 test with a written scale, kill, iterate, or escalate verdict before Q4 budget is set.
- A clean loss on AI Max for brand search and a clean win on form field reduction are not equal: one frees budget, one changes your CPA target.
- A 40 percent CPA improvement on a partner page means nothing without disbursal data. Escalate, do not scale.
- Too many open tests with no end dates is the root cause of ambiguity carrying over quarter to quarter. Cap concurrent tests at three per campaign type.
- Q4 targets should reflect Q3 learnings. A confirmed conversion rate improvement that does not lower your CPA target is a sandbagged target.
A Q4 plan is only as good as the Q3 verdicts underneath it. If you carry ambiguous tests into the new quarter without a clear disposition, you carry the same confusion, the same budget drag, and the same arguments about what to prioritize.
This template gives you a repeatable structure: review every open test from Q3, assign a scale, kill, or iterate verdict, and use those verdicts to set Q4 allocations. The examples below are real enough to be instructive. Adapt the structure to your own account.
Why verdicts before budgets
Most quarterly planning sessions start with budget. How much do we have, how do we split it, what do we want to try. That order is backwards.
Budget decisions are downstream of learning decisions. If a test ran in Q3 and you do not formally close it, it sits in the account consuming spend while the team debates it informally for another quarter. Multiply that by five or six open tests and you have a meaningful share of your budget allocated to ambiguity.
The right order: close Q3 tests with a written verdict, then allocate Q4 budget against things you have evidence for.
The verdict framework
Every test from Q3 gets one of three dispositions before Q4 planning begins.
Scale. The result was clean, the sample was sufficient, and the direction is clear. Move budget here in Q4.
Kill. The result was clean in the other direction, or the test ran long enough that no signal emerged. Remove it from the account and stop the debate.
Iterate. The result was directional but not conclusive, or a confounding variable means you cannot trust the read. Design a cleaner version for Q4 with a specific hypothesis and a defined end date.
A fourth category worth adding: Escalate. The test result is meaningful but the decision requires data or a commitment you do not currently have. Name the blocker and name who owns it.
Working through Q3 examples
The examples below come from a composite account. The numbers are illustrative. The structure is the template.
Test 1: AI Max on brand search
What ran. AI Max was applied to brand search campaigns in Q3 to test whether automated query expansion could find incremental volume without cannibalizing direct brand traffic.
Result. Clean loss. Brand CPA rose, impression share on exact brand terms dropped, and the incremental volume that did come in converted at a significantly lower rate than the control. No ambiguity in the read.
Verdict: Kill. Remove AI Max from brand search in Q4. Document the finding so the question does not get relitigated when a new team member joins or a vendor rep suggests it again. The specific failure mode (query expansion pulling in non-brand intent) is worth keeping in the account notes.
Q4 implication. Budget that was allocated to AI Max brand expansion goes back to exact match brand. No net new spend required.
Test 2: Form field reduction, 14 fields to 8 fields
What ran. The lead gen form on the primary landing page was shortened from 14 fields to 8 fields. The hypothesis was that friction was suppressing conversion rate more than the additional qualification data was worth.
Result. Clean win. Conversion rate increased 31 percent. Lead quality, measured by downstream sales qualification rate over a six-week window, held within two percentage points of the control. Statistically significant at the sample sizes involved.
Verdict: Scale. Roll the 8-field form to all active lead gen landing pages in Q4. If there are campaign-specific forms still running 14 fields, those are the first priority.
Q4 implication. This is a CPA improvement that compounds across the whole account, not just the test cell. Model the projected CPA reduction across Q4 volume and use it as a baseline when setting targets with leadership. A 31 percent conversion rate lift on the primary form changes what a profitable CPC looks like.
Test 3: Partner page CPA down 40 percent, no disbursal data
What ran. Traffic was sent to a co-branded partner page as an alternative to the owned landing page. CPA on the partner page came in 40 percent lower than the owned page.
Result. Directionally strong, but there is a critical blocker. The partner has not shared disbursal or fulfillment data, so you cannot confirm whether the leads that converted on the partner page actually completed the downstream process. A 40 percent CPA improvement means nothing if fulfillment rate is materially lower.
Verdict: Escalate. This does not get a budget allocation in Q4 until the disbursal data is in hand. Assign one owner to get that data from the partner before the Q4 budget lock date. If the data confirms that fulfillment holds, this becomes the highest-priority scale decision of the quarter. If it does not, the 40 percent CPA improvement is an illusion.
Q4 implication. Hold the partner page budget flat or at minimal maintenance spend until the escalation resolves. Do not scale on an unverified signal.
Test 4: Tier 2 channel mix, display at 1.8 percent conversion versus search at 9 percent
What ran. Display was added to the channel mix as a lower-CPM awareness and retargeting layer. Conversion rate on display came in at 1.8 percent versus 9 percent on search.
Result. The raw conversion rate comparison is not the right read here. Display and search serve different roles in the funnel and you should not expect them to convert at the same rate. The real question is whether display is generating assisted conversions or pipeline that search alone would not have produced, and whether the CPM efficiency makes it worth running even at lower conversion rates.
Verdict: Iterate. The test as designed does not answer the right question. In Q4, run display with proper view-through attribution and a holdout group so you can measure the assisted conversion contribution. Set a specific efficiency threshold before the test begins: if display-assisted pipeline does not reach a defined CPA within eight weeks, kill it.
Q4 implication. Keep display spend contained, at roughly the same level as Q3, until the cleaner test produces a verdict. Do not expand it based on the current read.
The meta-lesson: too many open tests
Across Q3, the account had too many tests running simultaneously with no defined end dates and no pre-committed verdict criteria. That is how you end up carrying ambiguity into Q4.
For Q4, set a maximum number of concurrent tests per campaign type. Three is a reasonable ceiling for most accounts. Define the verdict criteria before each test launches, not after the results come in. Assign an owner and an end date to every open test at the start of the quarter.
The Q3 to Q4 handoff template
Use this structure for every test in your account review.
Test name. What ran and when. Primary metric and result. Sample size and statistical confidence. Confounding variables or data gaps. Verdict: Scale, Kill, Iterate, or Escalate. Q4 budget implication: specific dollar or percentage shift. Owner and deadline (for Iterate and Escalate verdicts).
Run through every open test before the Q4 budget conversation begins. If a test does not have enough data to receive a verdict, that is itself a finding: the test was under-resourced or ran too short, and the Q4 version needs a larger budget or longer runway to be worth running.
Setting Q4 targets after verdicts are in
Once you have closed Q3 tests, your Q4 targets should reflect what you learned.
If the form field test produced a 31 percent conversion rate improvement and you are scaling it across the account, your Q4 CPA target should be lower than Q3. If you are not adjusting the target to reflect the improvement, you are either sandbagging or you do not believe the test result.
If you killed AI Max on brand search and are returning that budget to exact match, model the expected CPA on the recaptured volume before committing to a target.
Targets set before verdicts are arbitrary. Targets set after verdicts are defensible.
Where Almond fits
The hardest part of this process is not the template. It is that the Q3 context required to fill it in is scattered across Slack threads, meeting notes, ad platform dashboards, and the memory of whoever ran the test.
Almond's Business Context Engine captures tests, results and decisions as they happen across ad accounts, Slack, meetings and docs, and keeps that record current and versioned. When Q4 planning arrives, the verdicts are already documented. The form field result from Q3 does not live in one person's head. The AI Max failure is not relitigated because the account lead who remembers it is on leave.
Almond 2.0 early access is open at almond.ai. Every member gets a detailed 1:1 Meta and Google ad account audit with the CEO, recorded.
Frequently asked questions
- When should I start Q4 planning for paid media?
- Most performance marketers should start Q4 planning in mid to late September, after Q3 data has at least six weeks of post-click attribution. The priority is closing Q3 test verdicts before budget conversations begin, not the other way around. If your Q4 budget lock date is October 1, your Q3 review needs to be done by the third week of September.
- How do I decide whether to scale, kill, or iterate a test?
- Scale when the result is statistically clean, the sample is sufficient, and the direction is unambiguous. Kill when the result is a clean loss or no signal emerged after adequate runtime. Iterate when the result is directional but a confounding variable or data gap prevents a confident read. Add an Escalate category when the decision requires external data or a commitment you do not yet have.
- What is a quarterly business review for a performance marketing team?
- A quarterly business review for performance marketing is a structured session where the team closes every open test with a written verdict, documents what was learned, adjusts targets to reflect those learnings, and allocates next-quarter budget based on evidence rather than assumption. It should happen before budget is set, not after.
- How many tests should a paid media account run at once?
- Three concurrent tests per campaign type is a reasonable ceiling for most accounts. Running more than that dilutes budget across each test cell, extends the time needed to reach statistical significance, and makes it harder to isolate variables. The bigger problem is tests without defined end dates or pre-committed verdict criteria, which is what causes ambiguity to carry over quarter to quarter.
- How do I set Q4 CPA targets after a Q3 test improves conversion rate?
- If a Q3 test produced a measurable conversion rate improvement and you are scaling it, your Q4 CPA target should be lower than Q3 by roughly the same proportion. A 31 percent conversion rate lift on the primary form changes what a profitable CPC looks like across the whole account. Targets that do not reflect confirmed improvements are either sandbagged or signal that you do not trust the test result.
Almond 2.0 early access
Get a 1:1 Meta and Google audit with the CEO.
Every waitlist member gets a 1:1 audit of their ad accounts. Work email only.