How to forecast Google Ads results, by hand or in a spreadsheet

Two different questions hide behind the word forecast. The first is where the account is heading if nothing changes. The second is what would happen if you spent more or less.

Google's own tools answer the second. Keyword Planner forecasts, Performance Planner and the budget simulators model the auction at other budgets or bids. They are useful before a decision, and they say nothing about where this month will land at today's settings.

This guide is about the first question. It uses only your account's own daily history, so you can build it in a spreadsheet and check every number. It is the forecast a client means when they ask whether the month will hit its goal.

Export one row per day: date, cost, clicks, conversions and conversion value. Take at least eight weeks, and 13 months if you want to use last year's season. Campaigns segmented by day, or a saved report in the report editor, both work.

Keep the campaign column if budgets differ a lot between campaigns. Then note the days that break the pattern beside the figures: a budget change, a campaign paused or launched, a sale, a tracking fault. You will need them later, and they are hard to find again once the export is closed.

The run rate is the forecast most people do in their head.

It assumes every day of the month is worth the same. Few accounts work that way. Take Ferncliff Plumbing, a made-up business that spends about €500 on a weekday and €320 on a Saturday or Sunday.

October 2026 starts on a Thursday and has 22 weekdays and 9 weekend days. At those rates the month will end at €13,880. On the morning of the 5th, Ferncliff has four complete days: Thursday, Friday, Saturday and Sunday, worth €1,640.

The run rate says €1,640 ÷ 4 × 31 = €12,710. That is 8% short, because half of the days read so far were weekend days, against 29% of the month. In a month that opens on four weekdays, it overshoots instead.

The error shrinks as the month fills, since the days already spent are real. A run rate read on the 25th is usually fine. Read on the 5th, it can send a manager to raise a budget that was on track.

Keep what is already spent, and forecast only the days left, each from the average of its own weekday.

Read each weekday's average over the last four weeks if the account moves quickly, or eight if it is steady. Use the median rather than the mean when a public holiday sits in the window, so one odd Monday does not set every Monday ahead.

On the 5th, Ferncliff has 20 weekdays and 7 weekend days left. The forecast is €1,640 + 20 × €500 + 7 × €320 = €13,880, which is where the month will actually end if nothing changes.

Weekday averages assume the next weeks look like the last ones. For a month that differs every year, such as December for a gift shop or January for a gym, they will be wrong in a known direction.

Use it only when last year's account was comparable: the same campaigns, similar budgets, tracking that counted the same things. A year in which budgets doubled in March teaches you about the budget, not the season. Watch moving dates too. Easter switches between March and April, and Black Friday falls in late November.

Conversions are harder than spend for two reasons. They are lumpy, and they arrive late. Google Ads credits a conversion to the day of the click, so recent days keep gaining conversions for days or weeks.

The steadier way is to forecast clicks the same way as spend, then multiply them by a conversion rate. Read that rate over the last eight weeks, leaving out the newest days whose conversions are still arriving.

Then deal with the days already in the month. Segment conversions by days to conversion to see your own lag. Say 70% of a day's conversions are in after one day and 90% after seven. A day that shows 14 conversions the next morning will likely end near 20, because 14 ÷ 0.7 = 20.

Back to Ferncliff, on the morning of the 14th. Thirteen days show 61 leads, and the lag curve says about 7 more are still owed to the last week. The clicks forecast for the 18 days left is 1,900, and the settled conversion rate is 5%. That gives 61 + 7 + 95 = 163 leads for October, about 5 a day on both sides of the 14th.

Below about 30 conversions a month, a conversion forecast is mostly noise. Forecast spend and clicks with care, and give conversions a wide range rather than a precise number.

A trend can promise spend the budgets will never allow. Google Ads charges a campaign at most 30.4 times its average daily budget in a calendar month, and at most twice that budget on one day. After a budget change, the month's limit becomes what was already spent plus the new budget for each day left. The budget pacing guide covers these limits in detail.

• Limited by budget. If a campaign loses a large share of searches to budget, it will spend its room. Forecast its spend from the budget, not the trend.
• A budget changed last week. The history was spent at the old budget. Use the new one for the days left.
• Paused campaigns. Take their share out of the trend. A paused campaign spends nothing ahead.
• Campaign total budgets. Since January 2026, Search, Shopping and Performance Max campaigns can also run on one total for set dates, from 3 to 90 days. Forecast those against what is left of the total.

Some days should not feed the forecast at all, or only from a given date.

• What counts as a conversion changed. A new primary action, a longer call threshold, a switch from every to one per click. Restart the conversion series on that day. Spend and clicks keep their history.
• A long pause. After four weeks or more without spend, start the history again from the restart.
• A one-off sale or an outage. Leave those days out of the averages, or forecast them apart if they will happen again.

Google's seasonality adjustments and data exclusions tell Smart Bidding about short events. They change the bidding, not your forecast. Your spreadsheet still needs its own notes.

A single number invites the wrong reading. €13,880 sounds exact, and nobody knows whether €13,200 would be a surprise. A range says it.

Build it from your own past. On the 10th of each of the last few months, forecast the month end with the days you had, then compare with how the month really ended. If your forecasts usually landed within about 7% of the truth, show this month as €13,880, give or take 7%: roughly €12,900 to €14,850.

Expect wider ranges early in the month, further ahead, and for conversions more than for spend. On a small account, a few late sales can move the whole month, so keep the range generous.

The same past months tell you which method to trust. Forecast each of them with the run rate, the weekday method and last year's season, and see which one came closest to the real month ends, month after month.

Keep a plain 14-day average as your yardstick. If a cleverer method does not beat it over several months, use the plain one: it is easier to explain to a client. When no single method wins every month, taking the middle of their forecasts is a safe choice, because it ignores whichever one goes wild.

A forecast says where the account goes if nothing changes. Set it against the goal or the budget. If the target sits inside the range, the month is on course and a change is a bet. If it sits outside, the gap is real and worth a decision.

Re-forecast every day if you like, but act on the trend over several days. A month end that moves €200 overnight on a €14,000 month is noise. To turn a forecast into a goal pace, the goal tracking guide takes it from there.

Optizads builds this kind of forecast for you every night, from each account's own daily history, with the safeguards of this guide built in: late conversions, budgets, paused campaigns and long pauses. You see the month end and its likely range on the Overview and on Goals each morning.

How the Optizads forecast works explains it in plain words.