Smart Bidding for Seasonal Trades: Storms and Heat Waves

The first 95-degree day of the year, your phone rings off the hook and your Google Ads spend barely moves. Three weeks later the weather breaks, demand dies, and your cost per lead is suddenly double what it was in March. Both problems have the same cause, and the fix is not the one most contractors reach for.

Smart Bidding is an auction-time system. It prices every click against what your account has already done, not against what the radar says is coming. So it under-bids into a surge and over-bids out of one. Google ships two tools that look like the answer — seasonality adjustments and data exclusions — and both are routinely used for the wrong job.

Here is what each one is actually for, and what to do instead when your season lasts four months.

Why does smart bidding lag behind a heat wave or a storm?

Because Smart Bidding prices your bids off conversion data you have already produced, and Google recommends evaluating results over periods containing at least 30 conversions (50 for Target ROAS) — a window measured in weeks, not hours. A storm changes your conversion rate overnight. Your data catches up later.

Google does not publish the exact lookback window its models use, and anyone who tells you the number is guessing. What Google does publish, in its About Smart Bidding documentation, is the list of signals it bids on: device, physical location, location intent, weekday and time of day, the actual search query, and — notably — seasonality only as a Shopping-campaign signal.

That is the whole problem in one line. The seasonality signal Google names is built for a retailer whose Christmas arrives on the same day every year. Your season arrives when a hailstorm hits at 4pm on a Tuesday.

So the lag runs in both directions. Day one of the heat wave, the system is still bidding at the conversion rate of a mild June, and you lose auctions you should have won. Two weeks after the weather breaks, it is bidding at the conversion rate of a panic-buying week, and you overpay for clicks from people who are just browsing.

What is a seasonality adjustment actually built for?

Short spikes. Google's own help page states that seasonality adjustments 'are ideal for short events of 1-7 days' and that 'they may not work as well if you use them for extended periods (more than 14 days at a time).' It also says to use them only if you expect major changes to conversion rates, 'because Smart Bidding already manages seasonal events.'

Mechanically, it is a scheduled conversion-rate adjustment, not a bid multiplier. Per Google's seasonality adjustment docs, if you expect conversion rates to rise 50% during a 3-day sale, you enter a +50% conversion-rate adjustment for those three days. Bids return to normal afterward — Google states no negative adjustment is needed once the event ends.

It is available on Search, Shopping and Display campaigns using Target CPA or Target ROAS, plus Performance Max and App (beta) campaigns on all bid strategies. It is not supported in Travel campaigns, and a single event can hold a maximum of 2,000 named campaigns.

Read that against your actual calendar. A tornado warning weekend is a seasonality adjustment. Roof-replacement season is not. Setting a +40% adjustment for all of July and August is the single most common misuse we see recommended in trade PPC advice, and it runs the tool at four to eight times its documented length.

When should you use a data exclusion instead?

Only when your conversion data is wrong — never when your conversion data is merely bad. Google's data exclusions documentation names three common triggers: tagging issues, website outages, and data import issues. A slow week is not one of them.

This distinction costs contractors real money. A roofer has a dead February, sees a wrecked CPA in the account, and excludes the period so it 'doesn't poison the algorithm.' That is a lie told to the bidding system. February really did convert at that rate, and the model needs to know it.

The reverse mistake is worse: excluding the storm week because the CPA looked weird. You have now deleted your best conversion data from the only model that decides what you pay per click.

If you do have a genuine tracking break, Google's rules are specific. Exclusions apply to clicks, not conversions, so you must cover the clicks that would have produced the missing conversions. Google's own worked example: if uploads broke October 15-18 and your conversion delay is 5 days, exclude October 10-18. Aim to capture at least 90% of the affected clicks.

Verdict: for a seasonal trade, the budget and structure change is the tool that matters. Seasonality adjustments are a scalpel for a 3-day event. Data exclusions are a fire extinguisher for a broken tag. Neither one is a plan for summer.

What do you do for a real multi-month season?

You raise the budget and split the campaign, because your budget — not your bid strategy — is the hard ceiling. Google caps spend at two times your average daily budget on any given day and 30.4 times your average daily budget in any month, per its average daily budget documentation.

Do the math before you touch a bid. A campaign set to $100/day cannot spend more than about $3,040 in a month no matter how many roofs the hail took off. If July demand is triple June, no seasonality adjustment on earth gets you a third roof — the money is not in the account.

Structure is the other half. Emergency and storm-driven services should live in their own campaign with their own budget, separate from your steady-state maintenance and replacement campaigns. Then a surge gets funded without inflating spend on the terms that did not spike, and you can cut it in a day when the weather turns. Our bid strategy automation guide covers which strategy belongs in each of those campaigns.

• Raise the average daily budget on the seasonal campaign first — the 2x daily / 30.4x monthly ceiling is calculated from it.
• Keep storm and emergency terms in a separate campaign so the surge is funded and killable on its own.
• If you use Target CPA, raise the target during the season rather than leaving a January target on a July auction.
• Do not swap bid strategies mid-surge; you are throwing away the learning right when you need it.
• Log the start and end date of the season in writing so next year's prep is not a memory exercise.

How do you prepare a bid strategy before storm season, not during?

You get 30 conversions of clean history into the campaign before the season starts, because that is the volume Google names for evaluating Smart Bidding performance (50 for Target ROAS). A campaign you switch on the day the hail falls has no data, and it will spend your surge budget learning what your normal is.

The prep work is boring and it is the whole game:

• Fix conversion tracking first: count calls over 60 seconds and booked jobs, not every form ping and every 8-second wrong number.
• Know your conversion delay — the days between click and booked job. You need it for exclusions and for reading the surge.
• Build the storm campaign now, paused, with ad copy, negatives and a budget already set. Enabling a built campaign takes two minutes.
• Set a budget headroom number: what you are willing to spend per day at peak, written down before your phone starts ringing.
• Decide the kill trigger in advance: what CPA or what booked-job rate ends the surge budget.

If your account cannot answer 'what does a booked job cost us' in one number, no bid strategy will save the season. That is a measurement problem, and it is the first thing we fix in a paid ads engagement.

Should you change budgets or bids when demand spikes?

Budget first, target second, and never both on the same day. Budget sets how much volume you can buy; the target sets how selective the system is. A campaign that is limited by budget will ignore a generous target, because it ran out of money before it ran out of auctions.

So the order is: raise the daily budget, watch two or three days, then — if the campaign is spending its full budget and still leaving qualified auctions on the table — raise the Target CPA to what a storm lead is actually worth to you. A booked roof replacement can carry a far higher cost per lead than a gutter clean, and your target should say so.

Move targets in steps of 10-20%, not 100%. A large target jump behaves like a new strategy and forces the system to re-learn at the worst possible moment. If you are not sure what your peak budget should be, our guide on how much Google Ads budget a service business needs has the math.

One naming note so you are not confused in the interface: Google announced that from June 2026, 'Maximize conversions with a Target CPA' is simply labeled 'Target CPA,' and 'Maximize conversion value with a Target ROAS' is labeled 'Target ROAS.' The bidding behavior did not change.

What happens to your CPA for the two weeks after the surge?

It goes up, and you should expect it to. The system just spent three weeks learning a conversion rate produced by people whose basement was flooding. Those bids do not fit a normal Tuesday, and the account will keep paying storm prices for a while after the storm.

So walk the budget back down the way you walked it up: in steps, over about the same number of days, not to zero overnight. Judge the result on a window that contains at least 30 conversions rather than on yesterday, or you will be reacting to noise and re-training the model on your own panic.

Resist two temptations here. Do not exclude the surge weeks — they are real data. And do not pause the seasonal campaign entirely in the off-season if you plan to run it again; a campaign with no recent conversions comes back cold. Drop it to a floor budget on your highest-intent keywords instead. The HVAC paid ads playbook works through what that floor looks like for a trade with two peaks a year.

If your ads spend is riding a season you cannot control and nobody has shown you what a booked job costs, that is a fixable problem and a 90-day one. We run paid ads month to month with no lock-in, and you keep the account. Start with our paid ads service, or get my free audit and we will tell you what your seasonality is actually costing you.