Forecasting with Google Sheets & Vertex AI

The PPC landscape is constantly evolving, and the ability to predict campaign performance in the future will be invaluable.

This article will discuss some of my favourite PPC forecasting methods using Google Sheets, and Vertex AI. These techniques help to paint a more accurate picture of the future for clients and provide them with useful insights.

The methods described here are a good way to get a sense of what the potential future of PPC campaigns could be.

Google Sheets FORECAST Function: Basics

Google Sheets has a reliable and easy-to-use forecasting feature using the formula:


=FORECAST(z, known_y values, known_x values)


Where:

• The data point that you are trying to predict the corresponding y value for is.
• known_y is a range of dependent data (usually your previous results or outcomes).
• known_x is a range of independent datapoints (usually the variable that you believe might influence your outcome).

If you have only two dimensions, this function is an excellent tool.

It uses linear regression. This is good for a quick peek at forecasting, but not for anything too complex that takes into account external circumstances or data sources.

Imagine you want to make a budget forecast for the future based on historical data.

In this example we want to forecast the future sales between September and December based on current sales data.

You can quickly see that this method has its drawbacks if you visualize the forecasts.

The blue line shows the actual sales figures up to August and the red line the forecasted sales figures.

The prediction is nothing more than a trendline. It might be useful to have a quick glance at something, but it is nothing in comparison to the blue line. This is how actual business data will appear.

Google Sheets FORECAST Function – Supercharged

There are many ways to fix linear regression using advanced methods.

You can also add trend data and other market predictions to the forecast formula instead of using just the linear =FORECAST().


=FORECAST()*3rdParty_Trend_Data


You can get trend data from Google Trends or other public sources such as Google Keyword Planner, Dataset Search, Google Industry Reports (PwC, EY and McKinsey). Export it to a CSV file or another format that you are familiar with.

You can clean up the datasets so that they match your original spreadsheet’s structure. For example, you could use data on a daily, weekly, or monthly basis.

Then, you can add the FORECAST feature to your computer’s forecasting software for a more accurate prediction than a simple line that goes up or down.

In this example we have added trend data that shows an increasing trend towards the Q4 of the current year. These numbers differ from predicted sales that do not include trend data.

Visualizing these new data allows us to see more details and better insights than a flat line.

It is a good rule of thumb to back up your forecasts as much as possible with data and provide data for more specific time frames, such as day-today or week-toweek.

Vertex AI – Advanced Forecasting

Vertex AI from Google could change the way you forecast your PPC campaigns if they involve large datasets and multiple variables.

Vertex AI, unlike simpler tools, allows for more complicated models that take into account a variety of factors, including seasonality, different advertising platforms, and even global market trends.

You’ll need to first upload your PPC historical data to Google Cloud Storage.

You can then access Vertex AI AutoML Tables and build a machine-learning model that is tailored to your dataset.

You can use the built-in metrics after training to evaluate how well it performs. This will ensure that your forecasting needs are met. Once you are satisfied with the model, it is simple to deploy.

You can now use this model for predicting future outcomes such as clicks or impressions based on different levels of advertising spend, placements or any other variables you consider important.

What’s the best part? This doesn’t require you to be an expert in machine learning. You can get more accurate PPC forecasts with a little setup and fine tuning.

Vertex AI’s capabilities are endless. Let’s start by looking at a simple framework.

After you have created your Vertex AI project and set up your Google Cloud account, you can start creating data sets.

Data sets are basically the data points that you will use to forecast.

The data set includes a budget dimension, as well as some revenue and time dimensions. The data sets may contain different data depending on the goal.

Choose Regression, Forecasting and Tabular for your data types.

Forecasting is more focused on predicting the future of a series. While regression can be used to analyze relationships, it also works with different data types.

Both tools are vital in data science, and they are used to make different decisions and analyses. In most cases, you will be able to forecast the outcome.

It’s now time to train a brand new model. AutoML is a great option for beginners. You must then set the forecast period, target and data granularity.

Set the budget and duration of training once you have completed this step. You’ll be notified when the model is finished learning.

This is the last step. The option will only be available after the training.

You must submit the data on which you will base your forecast. Use more recent data.

The model will forecast future target values based on the training set of data and your forecast set.

The job may take some time depending on the volume of data. You shouldn’t have to wait longer than 5-10 mins for PPC.

Vertex AI will provide you with an output file containing new columns of predicted values. You can then use this information to make further decisions.

Vertex AI may seem like a lot of software for certain forecasting tasks. However, you can train the model with years of historical data and inventory insights.

Vertex AI allows you to build a machine-learning forecasting model that is tailored to your company, and which is far more powerful than any static formula.

PPC Forecasting to improve campaign performance

The choice of these tools ultimately depends on the goals you have and the complexity level of your PPC campaign.

Google Sheets is a simple and easy way to get started with PPC forecasting. It may have some limitations, but it is a good starting point for most advertisers.

Vertex AI, on the other hand takes your PPC Forecasting to the next step with its ability of handling large datasets and complicated models. You can now take into account seasonality, global trends and other variables to make accurate business predictions.

Forecasting PPC advertising no longer requires guesswork.

Now you can make informed decisions about your PPC campaigns based on data.

Dig deep: A guide for effective PPC projections

The post Forecasting PPC with Google Sheets & Vertex AI first appeared on Search Engine land.