Opensource Model: https://github.com/LocalSEOGuide/lsg-forecasting
Forecasting isn’t always straightforward but it’s essential for stakeholders & SEOs. Stakeholders require forecasts from SEOs to predict revenue and figure out what they are going to get for their investment so they can determine if the juice is worth the squeeze.
Furthermore, many SEOs find forecasting to be troublesome due to the unpredictability associated with some SEO initiatives. This, in turn, can lead some to justify budget spending to stakeholders by using forecasting methods that don’t have a lot of sound science behind them.
So the question is, how are stakeholders supposed to know if they are being sold a bill of goods or a big opportunity? And, how can forecasting be made better? More accurate? And predictable at different levels of investment?
About this SEO Forecasting Model
This problem of how to improve forecasting is what this tool is all about. Its purpose is to provide a good foundation for connecting historical data and performance to different levels of SEO investment.
This will allow you to have a better understanding of your SEO ROI at various levels as well as help you prioritize SEO projects and get good forecasting predictions instead of questionable fortune-telling.
Our Goal: As we develop and refine this tool through industry collaboration we will be able to make better predictions and help forge an industry standard for SEO forecasts. In the spirit of collaboration, we have open-sourced our model so you can have access. This will allow you to do some DIY forecasting and benefit from the progress we have made thus far.
What you’ll need to use the forecasting tool:
- Someone with Python experience
- Cleaned up historical data
- Statistical Modeling Libraries
Resources for build:
- ARIMA Model – Complete Guide to Time Series Forecasting
- How to Create an ARIMA Model
- Introduction to ARIMA
- Local SEO Guide! 🤙
- Our Forecasting model
The Future of the SEO Forecasting Modeling
This forecasting model is an ongoing project that will be continuously updated as we collaborate with others and refine it with new features and data points.
Where we want to take this:
Additional Model Optimizations on the Horizon
Currently, we have four buckets for how to move forward with the model and iterate for a more optimal version.
Becoming more accurate
- Use historical data and see how the automating test fits
- Modified calculations based on projections/insights
- Break clients down to cohorts
Incorporating other data points
- Incorporate Google trends
- RICE Model addition: the RICE model help prioritize SEO projects by impact and cost
Having cleaner data
- Having historical data // **need to have
- Warehouse and classify your GSC data
- Improved front end and user experience
Want to know more about SEO forecasting?
If you missed out on our forecasting presentation at Local U you can catch the recap here. Andrew Shotland, CEO of Local SEO Guide, walks through the current state of forecasting in the industry, common types of forecasting, and how we attempted to improve on some of these forecasting methods.
Additionally, Local SEO Guide is holding a live panel of SEOs to discuss forecasting methods and optimizations which will help inform new refinements. It will also help stakeholders better understand what goes into a good methodology vs. a bad one so you can spot the difference. For SEOs the webinar we be an open discussion format with a Q&A portion for comments, questions, and thoughts.
We encourage you to come to the live forecasting panel event to listen, share, or just learn more about better ways to forecast. The panel webinar is Wednesday, April 27 at 10 AM PT, register here.