The festive-quarter ad account: six adjustments before Diwali

Adjustment 1. Budget re-baselining

The instinct is to raise the daily cap and move on. That is not budget re-baselining; that is permission-granting. The re-baseline is a full recalculation of what the account is trying to do with money in the festive window.

Start with actual conversion volume from the last two festive quarters (2024 and 2025 if available), corrected for account changes since. Model the target CPA or ROAS you are willing to accept during peak, remembering that both drift during festive: CPA rises with auction competition, ROAS falls as new-customer share increases. Set a total budget for the window, not just a daily cap. Reserve the capital in the account. Then, and only then, raise the daily caps.

Accounts that skip re-baselining and just raise the cap either underspend (bidding cannot find the volume) or overspend (bidding wins auctions the account does not want).

Adjustment 2. Bidding recalibration

Most bidding strategies were tuned for a non-festive baseline. In festive weeks, CPMs and CPCs typically rise between 30 and 80 percent depending on category. A target CPA left at the pre-festive number will win fewer auctions each day, and the account will look like it is under-pacing when it is actually being priced out.

Two workable moves. First, raise the target CPA (or lower the target ROAS) proportionally to expected auction inflation, category by category. Second, on the campaigns where volume matters more than efficiency for the peak, switch to Maximise Conversions or Maximise Conversion Value with a bid cap, so the account chases the volume that is available while the cap keeps the worst auctions out.

Either move, made deliberately. Not both on the same campaign, and not none.

Adjustment 3. Audience refresh

Remarketing lists built in Q1 and Q2 are stale for festive intent. A visitor from July who did not convert is a different signal now than they were then. Refresh the audiences the account is actually spending against.

Rebuild the 30-day cart abandoner list, the 60-day high-intent-search converter list, the newsletter-subscriber-not-yet-purchased list. Add any signal specific to festive: users who viewed the "gift" category, users who added multi-item baskets, users who searched a wedding-adjacent keyword. Exclude segments that historically do not convert in festive: newsletter-only signups, support-ticket contacts, employees on the office IP range.

Cold reach for the festive window is a separate exercise; do not fund it from remarketing budget without segregating the reporting.

Adjustment 4. Creative rotation

Non-festive creative underperforms in festive auctions for a specific reason: it looks generic against the seasonally-tuned inventory the account is bidding alongside. The bar is not "make it look festive." The bar is "make it look like this brand chose to show up for the season."

Prepare three to five festive variants for every major asset group. Vary the message across at least two intents (gifting and self-purchase). Test culturally specific visuals that stop short of cliche. Rotate the primary variants on Dhanteras morning; have a second-wave set ready for the days between Diwali and the following weekend, because intent shifts and the creative that landed on the 6th will fatigue by the 10th.

Every festive creative variant carries a matched landing page. Skipping the landing-page match is where most festive campaigns lose their gains.

Adjustment 5. Tracking check

Festive traffic exposes every latent tracking bug the account has been quietly living with. Conversion tags that fire correctly at 1,000 sessions a day misfire at 10,000. Tag Manager environments diverge between staging and production because the last deploy was in July. Offline conversion imports that run weekly cannot inform bidding during a window where a week is the whole peak.

Before the window opens, verify: the conversion tag fires exactly once per counted outcome under peak load; the Tag Manager production container matches the reviewed configuration; offline imports run daily (or hourly if the CRM allows) for the peak weeks; cross-device conversion linkage is intact; enhanced conversions are configured correctly if the account relies on them.

Fix now. The account cannot be re-instrumented during peak; it can only be over-adjusted based on wrong numbers.

Adjustment 6. Dayparting

Festive purchase behaviour clusters differently. Late-evening spikes (post family time), mid-morning secondary peaks (research from the office), the pre-Diwali weekend surge, and the day-after Diwali drop-off are all category-specific. What is universal is that the flat 24-hour schedule the account has been running for the rest of the year does not reflect festive intent.

Pull the last two festive quarters' hourly conversion data. Weight upward the hours that historically convert; do not weight downward the hours that do not, because absence of past conversions in a bad hour may reflect low past spend rather than low intent. Set explicit schedules for Dhanteras, Diwali, and the two days after, because those days behave unlike any other days of the year.

Dayparting comes last because the other five adjustments change the data dayparting is calibrated against.

Why the order matters

Budget re-baselining first, because bidding recalibration without a budget frame either overshoots or undershoots. Bidding second, because audience refresh without recalibrated bids optimises against the wrong signal. Audience third, because creative rotation without a refreshed audience shows the right ad to the wrong person. Creative fourth, because tracking must be verified against real festive-shape creative flow to be trustworthy. Tracking fifth, because dayparting off broken tracking is worse than no dayparting. Dayparting last, because the data all five other adjustments produce is what the schedule is set against.

Reversing any two adjacent items in this list makes the sequence less reliable. Reversing three or more produces the reactive festive account most operators end up with by the second week of November.