Google Ads Forecasting: The 3 Methods That Actually Work in 2026

Google Ads forecasting is the practice of projecting future campaign performance, including spend, clicks, conversions, CPA, and ROAS, based on historical account data, planned budget and bid changes, and known seasonality. In 2026, three methods cover almost every real-world forecasting job: Google Ads Performance Planner for short-horizon tactical planning, AI-assisted analysis for scenario forecasts in minutes, and spreadsheet modeling for long-horizon custom planning. Most PPC managers use at least two. The one most still rely on, the Keyword Planner forecast feature, is the weakest of the bunch, and we’ll explain why.

TL;DR

• Forecasting Google Ads is fundamentally a math problem. The CPA formula (CPC ÷ conversion rate) and the ROAS formula (conversion value ÷ spend) hold whether you compute them in a spreadsheet, in Performance Planner, or in an AI assistant. The tool you pick affects speed and confidence, not the underlying arithmetic.
• Performance Planner is the most accurate short-horizon method for active campaigns because it uses Google’s auction-level data. The tradeoff: accuracy is strongest at monthly and quarterly horizons and degrades for annual planning, and the tool needs enough campaign history to be reliable.
• AI-assisted forecasting (Genie) produces scenario forecasts with the methodology shown, in minutes. It’s strongest at detecting momentum shifts and inflection points, surfacing graded business signals like ROAS drift or spend volatility, and flagging its own assumptions when the base case looks optimistic.
• Spreadsheet modeling wins for long horizons and custom assumptions, anything stretching into annual planning, anything that needs business-specific revenue logic, anything you need to defend to a CFO.
• 74% of PPC advertisers use Keyword Planner for forecasting, according to a Databox survey. It’s free, it’s familiar, and it’s the wrong tool for forecasting a campaign. It forecasts keywords.

What Google Ads forecasting actually predicts (and what it doesn’t)

PPC managers use the word “forecast” to mean three different things, and conflating them is the most common reason forecasts miss.

Keyword-level forecasting predicts how many impressions and clicks a specific keyword would generate at a given bid. Google Ads Keyword Planner does this. The output works for keyword research and bid estimation. It does not work for campaign planning.

Campaign-level forecasting predicts what an existing campaign will spend, deliver, and convert over a future window, usually 30, 60, or 90 days, assuming current settings hold. Performance Planner does this well. So does a well-built spreadsheet or an AI assistant working off the account’s historical data.

Account-level scenario forecasting predicts what happens to the whole account if you change something: increase budget 50%, pause a campaign, shift from Manual CPC to Target ROAS. The job is harder. Performance Planner handles some of it. AI assistants handle the analytical layer. Real bid-strategy simulation usually still requires running the change.

When someone asks “can you forecast my Google Ads?”, the first question is which of the three they actually mean. Most of the time it’s the second one, campaign-level, but the article that gets shared usually answers the first one. The mismatch is why most published forecasting advice underdelivers.

The three forecasting methods that work in 2026

For campaign-level and scenario forecasting, the work most PPC managers actually need to do, three methods cover the field.

A fourth category, third-party PPC platforms like Optmyzr, exists, but most modern PPC stacks have collapsed those features into either the Google Ads UI itself or into a broader analytics layer. We don’t recommend buying a dedicated forecasting tool in 2026 unless you have an enterprise need (multi-account roll-up, custom attribution models) the three methods above can’t cover.

Method 1: Performance Planner is most accurate at monthly and quarterly horizons

Performance Planner is Google’s built-in forecasting tool inside Google Ads. It uses auction-level data, what Google sees from billions of queries, to project how a campaign will perform under different spend or bid settings. Forecast horizons can stretch monthly, quarterly, or annually, but Google itself recommends not projecting too far ahead because accuracy degrades the further out you go.

It’s the most accurate option for tactical, short-horizon work because it has access to data no third-party tool sees: real-time auction conditions, query-level competition, and Google’s own bidding model behavior. If you’re trying to answer “how much spend can I add to this campaign next month and still hit my target CPA?”, Performance Planner is the right first move.

The blind spots are real, and they’re the reason it can’t be the only method you use:

Accuracy degrades for longer horizons. Monthly and quarterly forecasts are based on stable accounts. Annual forecasts are possible in the UI, but Google’s own guidance is “don’t project too far ahead.” For annual planning that has to hold up in a meeting, build a spreadsheet model alongside.

It requires campaign history. New campaigns, recently restructured campaigns, or campaigns with sparse conversion data get weak forecasts. Google needs roughly 30 days of conversion data to produce a reliable projection.

It can’t model cross-campaign effects. Performance Planner forecasts one campaign at a time. If you’re shifting budget between campaigns, you’re stitching forecasts together manually.

It doesn’t surface business signals. Performance Planner will tell you the projected spend. It won’t tell you that your ROAS has collapsed from 1.2x to 0.5x in the last month, or that your conversion tracking might be underreporting. Pattern questions like those need a different tool.

It’s a black box on methodology. You see the projection. You don’t see the math. The opacity is fine for tactical use, but harder to defend in a board deck.

For campaigns you actively manage at monthly and quarterly horizons, Performance Planner is the strongest single tool. Use it. But don’t end there.

Method 2: AI-assisted forecasting with Genie produces scenarios in minutes

For most PPC managers, the bulk of forecasting time goes to pre-forecast work: pulling 90 days of clean data, segmenting it by campaign, calculating week-over-week trends, spotting where the trend broke, building scenarios at different spend levels. The work eats afternoons.

Genie, Databox’s AI analyst, collapses the prep into a conversation. You ask for a forecast. It pulls the data, detects regime changes, runs the scenarios, flags business signals, and shows its work.

Here’s what a real Genie forecast looks like. We asked it: “Forecast my Google Ads performance for the next 30 days based on the last 90 days of data. Include scenarios for spend, clicks, conversions, conversion value, and ROAS. Flag any regime changes or business signals worth knowing about. Show me low, base, and high scenarios.”

Baseline: what the last 90 days tell us

🚨 Regime changes and business signals

🟢 Signal 1: Strong efficiency inflection, late-April breakout. The week of Apr 27 was the clearest inflection point in the entire 90-day window: $19,525 conversion value on $3,988 spend = 4.9x ROAS. The pattern wasn’t a spending spike. It was a conversion efficiency jump. Campaigns either found better-performing audiences, got quality score improvements, or hit a demand surge.

🟡 Signal 2: Spend volatility is extremely high. Daily spend ranges from $40.80 to $2,159 (σ = $505, nearly equal to the daily mean). The pattern doesn’t look like a stable managed budget. It behaves like campaigns with variable bidding, paused/unpaused campaigns, or multi-budget management. The volatility inflates scenario uncertainty.

🟡 Signal 3: Weekend drop pattern. Consistent across the 90 days: Fridays and Saturdays show significantly compressed spend and conversions (e.g., May 12: $71, May 19: $67, Apr 22: $40). Likely audience behavior or bid scheduling. The next 30 days include ~9 weekend pairs, so expect 8–10 natural low-activity days in the forecast period.

🟢 Signal 4: ROAS recovery trend is real. After a mid-period dip (weeks of Mar 9 and Apr 13–20 saw ROAS fall to ~1.6–2.1x), the last two weeks have pushed ROAS back into 5.2x–5.4x territory, a genuine efficiency recovery.

🔴 Signal 5: One anomalous spike (Mar 6). A single day produced $11,731 in conversion value on $943 spend (ROAS: 12.4x). The number is a statistical outlier, almost certainly a one-time high-value transaction, tracking event misfire, or attribution quirk. Don’t extrapolate it into the forecast.

30-day forecast scenarios (2026-05-27 to 2026-06-25)

Method: the base scenario uses the last-30-day run rate adjusted for the confirmed MoM momentum trend. Low applies a -20% haircut (competition pressure, budget pullback, quality score decay). High applies a +20% lift (momentum continues, budget expansion, seasonal demand).

All three scenarios yield essentially the same ROAS (~6.1x) because spend and revenue scale together with the same MoM growth rate applied proportionally. The variance across scenarios reflects volume risk; efficiency risk holds constant across all three. If spend grows faster than conversion value (rising CPCs), ROAS will compress toward the 3.8–4.0x recent realized rate. The 6.1x figure is optimistic. It assumes the late-April efficiency breakout sustains. A more conservative ROAS assumption of 3.8–4.5x (the last 30-day actual) is a safer operational planning figure.

What to watch for

• ROAS hold: if ROAS drops below 3.0x week-over-week, the efficiency regime may be reverting. Set an alert.
• Conversion rate: you’re converting at a higher rate than spend growth, so protect it by not over-scaling budget too fast. Aggressive scaling can dilute quality audiences.
• Weekend/weekday split: don’t let weekend anomalies (single-day zeros or spikes) throw off weekly pacing reviews.
• Mar 6–style spikes: if another high-ROAS day appears, verify conversion tracking before declaring a new performance regime.

Key takeaway: Your Google Ads account is in a strong upward momentum phase. The last 30 days significantly outperformed the prior 30 across every metric: spend up moderately (+13%) while conversions (+56%) and revenue (+78%) grew much faster, compressing your effective cost-per-conversion. The base forecast of $98k in conversion value at $16k spend (6.1x ROAS) is achievable if the current efficiency regime holds. Plan operationally around the last-30-day actual ROAS of 3.87x as your floor. The biggest risk is overspending into lower-quality audiences if budgets scale faster than audience quality can support.

Four things in that output are worth pointing out, because they’re what a manual forecast usually misses:

Inflection point detection. Genie identified the late-April efficiency breakout as the most important signal in the 90-day window: conversion value jumped while spend held flat, a sign of efficiency improvement rather than volume growth. A spreadsheet that averages the whole period would smooth the signal away entirely.

Five graded business signals, with severity coding. Each signal carries a severity (🟢🟡🔴) and an interpretation. The Mar 6 outlier flagged as 🔴 is the kind of catch that prevents a forecast from absorbing a one-day anomaly into the baseline. The weekend pattern flagged as 🟡 prevents the operator from over-reacting to a single low-activity day.

The ROAS-same-across-scenarios observation. Genie explicitly flagged that all three scenarios produce the same 6.1x ROAS because spend and revenue scale together. The scenario variance reflects volume risk; efficiency risk is held constant. A naive forecasting tool produces three different ROAS numbers and never explains why they came out identical.

Self-skeptical honesty. Genie called its own base case optimistic and recommended using 3.87x (the last-30-day actual) as the operational floor rather than the 6.1x scenario figure. The willingness to flag its own assumptions is what makes the output defensible in a meeting.

Genie shows you the method it used to produce the forecast. The transparency makes it citable in a board meeting and separates it from “AI forecasting” products that ask you to trust the output without showing the math.

The right fit for Genie is 30–90 day campaign and account-level forecasting, scenario planning, and the exploratory analysis that precedes any formal forecast. For tactical, in-platform bid optimization on a single campaign with 30+ days of data, Performance Planner is still more precise. The two methods are peers. Use both when the question calls for both.

Method 3: Spreadsheet modeling handles long horizons and custom assumptions

For annual planning, or any forecast that needs to incorporate business-specific revenue logic, build a spreadsheet model.

The structure is straightforward. Pull 12–24 months of weekly or monthly campaign data. Apply a baseline method: naive averaging works for stable accounts, linear regression for trending accounts, ARIMA or exponential smoothing for accounts with clear seasonality. Layer in business assumptions: planned budget changes by month, expected conversion rate improvements from landing page work, expected CPC inflation, sales-cycle lag from ad click to closed revenue.

The method is one you build once and re-use. It’s slow the first time, then fast. It also holds up to CFO scrutiny because every assumption is on the page and every formula is auditable.

The right fit is annual planning, multi-quarter scenario modeling, and any forecast where the answer needs to defend itself in a meeting where someone will ask “where did this number come from?”

The wrong fit is anything you need in the next 30 minutes.

The forecasting math every PPC manager should know

You can’t audit a forecast you don’t understand. These are the formulas that sit underneath every Google Ads forecasting tool:

Cost per click (CPC) = Total spend ÷ total clicks

Click-through rate (CTR) = Clicks ÷ impressions

Conversion rate (CVR) = Conversions ÷ clicks

Cost per acquisition (CPA) = Spend ÷ conversions, or equivalently, CPC ÷ CVR

Return on ad spend (ROAS) = Conversion value ÷ spend

Forecasted clicks = Budget ÷ projected CPC

Forecasted conversions = Forecasted clicks × projected CVR

Forecasted revenue = Forecasted conversions × average order value (or average deal value, for B2B)

Forecasted ROAS = Forecasted revenue ÷ budget

The forecast is the chain. If you know your CPC trend, your CVR, your average order value, and your budget, you can derive every other metric. Where forecasts go wrong is usually in the input estimates: using a CPC pulled from a peak period, assuming a CVR that hasn’t been hit since Q1, or using a 6-month-old AOV in a market where prices have changed.

The math is straightforward. Producing honest input estimates is the hard work.

A 5-step process to forecast a Google Ads campaign from scratch

The process works for a new campaign, a campaign you’ve just inherited, or any campaign where you need to defend the forecast to someone else.

Step 1: Pull the historical baseline. Last 90 days minimum, 12 months if available. Segment by week. Pull spend, impressions, clicks, conversions, and conversion value at a minimum. If you have Databox or a similar tool connected to Google Ads, pull it there. If not, export from Google Ads UI to a sheet.

Step 2: Check for regime changes. Look at week-over-week trends in CPC, CVR, and ROAS. If any metric has changed by more than 25% in the last 30 days versus the prior 60, you have a regime change, and the baseline can’t be a simple average. Forecast off the current regime instead. Most manual forecasts skip the step.

Step 3: Pick your baseline method. For stable accounts, the last 4-week average works. For trending accounts, use the trend line from the last 12 weeks. For seasonal accounts (e-commerce, B2B with calendar-driven buying), use the same period from the prior year as the baseline, adjusted by current trend. If you’re using Genie, it picks the method based on the data and tells you what it picked.

Step 4: Build three scenarios. Pessimistic (assume volume continues to compress), base case (current trends hold), optimistic (assume planned improvements land). Use ±1 to ±1.5 standard deviations from the base as the range. A forecast with three numbers per metric is more useful than a forecast with one, and far more honest.

Step 5: Layer in known changes. Budget changes are linear within a band: a 20% budget increase usually delivers close to 20% more clicks at the same CPC, until you hit saturation. Bid strategy changes are non-linear and harder to predict. Conversion tracking changes invalidate prior baselines entirely. Document every assumption inline so the forecast is auditable later.

If you’re using Performance Planner, steps 1–3 are automated, and step 5 is partial. If you’re using Genie, all five steps happen inside the conversation. If you’re spreadsheeting, you build all five from scratch: slower the first time, faster after.

How to pick the right method for your situation

Three filters cut the choice in most cases:

Monthly or quarterly horizon, single active campaign, plenty of conversion history → Performance Planner. It’s the most precise tool for this exact job because it has Google’s auction data underneath it.

Cross-campaign forecast, regime change suspected, what-if scenarios needed, time pressure → Genie. Conversational forecasting handles the question well. You can iterate in real time, asking what if I move $5k from Campaign A to Campaign B.

Annual planning or longer, board-level defensibility, custom revenue logic → spreadsheet. Slow to build, fast to re-use, and every assumption is on the page.

A real workflow usually uses two or three. Run Performance Planner for monthly and quarterly tactical views. Ask Genie to surface what’s actually happening across campaigns and run the what-ifs. Move the cleanest numbers into the annual model. The forecast that holds up rarely comes from a single tool.

Three forecasting mistakes that wreck accuracy

Mistake 1: Using a blended baseline when a regime change has happened. If CPC has jumped 40% in the last month, the last-90-day average CPC is a lie. Your forecast will systematically understate spend and overstate volume. Always check for regime changes before forecasting.

Mistake 2: Treating point predictions as truth. Single-number forecasts feel confident and underdeliver. Scenario forecasts (low, base, high) set expectations honestly and let you stress-test budget asks against the pessimistic case. The Databox survey found that 44% monitor Google Ads weekly and 37% monitor daily. The cadence makes sense only if you have a forecast range to compare actuals against; without one, you’re watching numbers move without context.

Mistake 3: Forgetting that the most common forecasting tool isn’t a forecasting tool. Google Ads Keyword Planner, used by 74% of advertisers in that same Databox survey, forecasts keyword performance only. It tells you how many clicks a keyword might generate at a given bid. It doesn’t tell you what your campaign will spend or convert. Many published forecasting workflows lean heavily on Keyword Planner and end up answering the wrong question.

Forecasting Google Ads in 2026 comes down to three things: matching the method to the question, being honest about what you don’t know, and showing the math.

Genie handles the prep work in minutes and produces scenario forecasts you can defend, with the methodology shown, the regime changes flagged, and the business signals surfaced. Try it on your Google Ads account: connect your data, ask for a forecast, and see what your numbers actually say.