# How to forecast ad results

> Forecast ad results by turning budget into impressions, clicks and orders with CPM, CTR and conversion rate. Six steps, a rupee example and a low-base-high range.
- **Author**: Marxx
- **Published**: 2026-11-18
- **Category**: Reporting
- **URL**: https://marxx.ai/library-guides/how-to-forecast-ad-results

---

Forecast ad results by turning your budget into impressions with CPM, impressions into clicks with CTR, and clicks into orders with conversion rate. Use your own last 30 days of numbers, and build a low, base and high case instead of one guess.

## The chain

| Step | Formula |
|---|---|
| Impressions | Budget / CPM x 1,000 |
| Clicks | Impressions x CTR |
| Orders | Clicks x conversion rate |
| Revenue | Orders x average order value |
| ROAS | Revenue / budget |

CPM is the cost of 1,000 impressions. CTR is the share of impressions that become clicks. Conversion rate is the share of clicks that become orders.

## Step 1: pull your own numbers

Open the last 30 days in Ads Manager. Note CPM, CTR, conversion rate and average order value for the campaign you plan to run. If the account is new, run a small test first and forecast from that. Industry averages tell you little about your product.

## Step 2: set the budget and the dates

Pick a budget and a window, such as INR 1 lakh over 30 days.

## Step 3: run the chain

Work down the table. Round at the end, not at each step.

## Step 4: build three cases

Make a low case where every input is worse and a high case where every input is better. Keep the swings realistic. Your own last four weeks show how much each number moves.

## Step 5: compare with break-even

Your forecast ROAS must beat your [break-even ROAS](/glossary/break-even-roas), which is 1 / your profit margin before ad costs. If even the high case misses it, fix the offer or the margin before you spend.

## Step 6: re-forecast every week

Replace guesses with real numbers each week. Late orders arrive after the click, so allow for that when you read the first days.

## A worked example

Say you sell a INR 1,400 product with a 40% margin. Your break-even ROAS is 1 / 0.40 = 2.5. You plan INR 1,00,000 of spend.

| | Low | Base | High |
|---|---|---|---|
| CPM | INR 220 | INR 180 | INR 150 |
| CTR | 0.9% | 1.2% | 1.5% |
| Conversion rate | 1.5% | 2.0% | 2.5% |
| Impressions | 4,54,545 | 5,55,556 | 6,66,667 |
| Clicks | 4,091 | 6,667 | 10,000 |
| Orders | 61 | 133 | 250 |
| Revenue | INR 85,909 | INR 1,86,667 | INR 3,50,000 |
| ROAS | 0.86 | 1.87 | 3.50 |
| Cost per order | INR 1,630 | INR 750 | INR 400 |

The numbers are made up. The base case returns INR 1.87 for each INR 1 spent, which is below 2.5, so you would lose money. Only the high case clears break-even. That tells you to improve the page or the offer before you scale. Check your own figures in the [ROAS calculator](/roas-calculator).

## Forecasting Google Search

For Search, Keyword Planner gives estimates. Google says it draws on historical search data to estimate what a set of keywords might get for your spend, and that forecasts reflect your bid, budget, seasonality and ad quality (Google Ads Help, checked 2026-10-20).

## Mistakes to avoid

- Using one number instead of a range
- Forecasting from a different product's results
- Treating the forecast as a promise to your boss or client
- Assuming the cost per result stays flat as you raise spend
- Ignoring [conversion lag](/glossary/conversion-lag)

## FAQ

### How accurate can a forecast be?

It is only as close as your inputs. A range from your last four weeks is far better than one guess.

### What if I have no past data?

Run a small test budget first, sized with [how to build a creative testing budget](/guides/how-to-build-a-creative-testing-budget). Then forecast from the real numbers, using the [ad budget template](/templates/ad-budget-template).

### Which tool should I use?

A spreadsheet is enough. Put the chain in one row per case and change one input at a time.

---
- [More Reporting articles](https://marxx.ai/library-guides/category/reporting)
- [All articles](https://marxx.ai/library-guides)