As pay-per-click (PPC) professionals, we’re often asked by clients to forecast deliverables for sign off on budgets. PPC forecasting is an uncomfortable task, I don’t know anyone that actually enjoys doing this. Working in paid search, we are used to having the data tell us what to do next. But when the data doesn’t exist yet, we essentially have to predict the future and this pushes us out of our comfort zones.
Where do you start and what are the best process to follow? We’ll give you our best tips & tricks and leave you with some actionable insight.
What is PPC forecasting?
It’s a data driven best guess
Ultimately, it’s all a guesstimate. You use as much quality data you can get, and make an educated guess as to what can be delivered. If you can be confident in your logic/approach (and you have data to back up your decisions), you can be confident in your forecast.
Understand the difference between a forecast and a target
For me, there are three terms banded around that makes this whole process even more confusing. Before we start, we have to clearly define the difference.
- Forecast = a data driven prediction with deliverables to try and achieve a goal. It’s a realistic guess.
- Target = what the client needs to achieve as a business. This will no doubt be higher than the forecast.
- Schedule = Is the final signed off plan, the numbers are probably somewhere in between a forecast and a target.
Why is PPC forecasting so stressful?
When you start forecasting in Paid Search, there’s no doubt that it can be a stressful process all in all. Here are some of the key reasons why;
- Forecasting for every account is slightly different
- True – but the process to complete it is pretty much the same
- There is no set process for delivering forecasts
- That’s because there is no one way to do it but hopefully today you will feel differently
- It is difficult if I am new to an account / haven’t worked on it for a full year
- Again the process should help with this
- I don’t understand the non seasonal peaks & troughs of the account
- There is usually someone in your team who does
- It requires a lot of thought
- Yes it does
- I don’t want to be held responsible if it is wrong
- Growth through discomfort!
What can you do to make it less stressful?
Remember UPOD (Under Promise, Over Deliver)
When you forecast, it’s always better to under promise on the predictions and over deliver on the actuals. Trust me, if you pitch in a number that’s too high and you can’t deliver on it, then you’ll have more trouble than it’s worth. Don’t over promise and under deliver.
Of course, the same can be said if you under promise and over deliver by too much
It makes the original forecast look rubbish, and this can also be troublesome. You need to find the balance.
Getting stressed just thinking about this? Don’t worry, we’ve got all the steps broken out for you below. Calm down and keep reading!
The forecasting process
Forecasting is like baking. Let me explain…
As with forecasting, baking is about following a recipe (process) to deliver an end product. But as with any recipe, you might go back and tweek it too add more/less ingredients. As you practice and as time passes, you refine your processes and what you bake is something delicious. Forecasting is the same!
The guide to PPC forecasting
Step #1 – Pull the data
All forecasts are dependent on ensuring that you have data to forecast with. Quality data ensures that your forecast has a strong enough foundation to work with. Without it, anything you forecast will be inaccurate. This is the most important step and often the most time consuming.
Here’s a map you can follow when gathering your data:
There is no point moving onto the next steps if you haven’t followed the map above and gathered enough quality data to make accurate enough decisions.
Step #2 – Build your models
Now that you have the data that you need, you can now start building up your modeling. Start off with Cost, Avg CPC, CTR, CvR and AOV (if you need Revenue). The rest of your model can then be worked through tweeking your Manual Metrics. For example, if you need to factor in an increased CPC you can change the CPC and then your other metrics will change alongside this.
Once you have built your sheet, you can start forecasting.
Step #3 – Forecast in this order
It’s important to take a bottom up approach to PPC forecasting. You need to start with your Brand, then competitor, then other (for example, Google Shopping campaigns), and then finally the Generics.
The reason you do this is because you are ensuring that you are focusing your budget on the most efficient areas first. If you maximise the potential in these areas first, the numbers will then tell you what you can do with your less efficient areas where there is more volume to achieve.
Step #4 – Do your final checks
Make sure that you check that your forecasting follows the expected trends. Ensure there are no random spikes that look particularly unusual (unless there is a specific reason for doing so). You can use Year on Year graphs and variances here to determine if there is anything you need to flatten out.
Top Tip – Borrow another pair of eyes
When you have been knee deep in an excel sheet for hours on end, you can get dazed and confused by all of the numbers. It’s good to get a fresh pair of eyes on the numbers to see if they agree with your logic. This also helps remove the uncomfortable feeling you may have about forecasting, having someone else’s reassurance can give you a confidence boost.
Step #5 – Go back to forecasts, actualize and reforecast
There’s no shame in having to go back to forecasts that you have done and changing the numbers. No forecast should be set in stone. Go back each month and actualize the numbers to see how far off your forecast was from the actual numbers you delivered. You can then use this information to re forecast the remainder of the year. In theory, your forecasting will become more accurate throughout the year.
Remember the PPC forecasting Do’s and Don’ts
- Check with members of your team to understand any unknown peaks & troughs
- Be conservative with uplift figures (UPOD)
- Use uplifts similar to those seen in previous years unless you are confident of otherwise
- Check final forecast against last years figures to ensure no crazy jumps in performance
- Be as granular as you can with your initial data set
- Forecast, actualize and reforecast if needed.
- Pretend it will go away if you ignore it!
- Feel as if the task is your burden alone… if you are struggling, ask your community for help
- Take any information from Google trends as being final, it needs to be sense checked against actual data
- Use unrealistic uplift figures unless there is a genuine reason for doing so
LinkedIn - https://uk.linkedin.com/in/siondanielroberts
Latest posts by Dan Roberts (see all)
- Microsoft Advertising keeps Average Position and introduces new metrics - July 19, 2019
- Don’t fear Artificial Intelligence & Machine Learning. Embrace it! - June 24, 2019
- Advanced Location Options in Google Ads Updated - June 13, 2019