In this blog post, we provide a guide to accurate forecasting for paid search campaigns, to ensure your project stays on track and achieves its goals.
We are often asked by clients to forecast the return on paid search campaigns, and here are a few questions we are regularly asked:
How many sales will I get from my signed-off budget ?
If I increase spend, what does ROI look like?
How much should I be spending to generate an X% uplift in sales?
How will seasonal fluctuations affect my performance ?
Whilst these are all fair questions, predicting the future can be daunting. However, there are a few tools which can help you form a more accurate projection. In this blog post, we are going to show you: 1. How to gather relevant data 2. How to collate and extrapolate that data into a forecast
Albert Einstein once said “If you want to know the future, look at the past”. This holds true for paid search accounts. If the advertiser has run activity in the past, this can provide the biggest indication of what to expect in future performance. The amount of information available will cover most queries thrown your way. You can view
Number of Conversions
Cost Per Acquisition
Impression Share (Lost budget)
And much more
However, there are a few things to consider before you can take this information at face value.
Has there been enough activity running and therefore enough data gathered to extrapolate? Avoid making bold predictions on minimal information.
Has the activity run all year-round? If the advertiser ran activity from October to December and you are asked to forecast June, seasonality plays a role here and past data may not be of use. More on seasonality later.
Has the business increased or decreased its capabilities? For example, a better website could drastically improve conversion rate, therefore using historical data prior to this could underestimate project figures.
Has the account been set up and optimised accurately? If somebody managed the campaigns prior to yourself, do not assume they are as proficient at managing paid search campaigns as you are. Is the tracking set up correctly? Do the ads have effective messaging? Have optimisations been made? If the essentials have not been done, then the data will not be representative of what the account could actually achieve when it is managed correctly, and you want to factor in how you feel improvements to the account could impact performance later down the line.
This is the most powerful data point in your forecasting arsenal. However, use with caution. If you do not have historical data to go on, do not worry, there are other tools to help with forecasting.
No Historical Data? No Dramas. The keyword planner tool in the google ads platform has a forecasting feature. By entering the keywords you are targeting, selecting the location and dates you require predictions for, google will give you estimates for the following
It also provides estimates for
However, we recommend validating these numbers using your performance data from the website analytics. There are some pitfalls to be aware of with this feature, if you are using low volume search terms, there can be some inaccuracies with the forecasts produced. Also worth considering is that whilst Google’s algorithms are very sophisticated, in some circumstances they need to be taken with a pinch of salt in the event of unforeseen circumstances. For example, during the coronavirus pandemic, any predictions made after this began are largely guesswork due to the drastic change in people’s demands and behaviours. As long as you take this into account, this will be a great tool and is available to anyone with a Google ads account.
We touched upon seasonality earlier and whilst this won’t impact all accounts, for some, the impact is huge. December and the Christmas holiday period are good examples of this. For example, an accountancy firm may experience a decrease in demand due to people spending their efforts searching and purchasing gifts and other things related to the holiday period rather than accountancy services. On the other hand, a craft beer company may experience a real uptake in purchases due to the product being a common gift. Whether seasonality is going to be to your company’s advantage or disadvantage, an accurate picture of the impact can be estimated through Google Trends. Google Trends is a tool that allows you to see how popular certain search terms are at various times of the year. Looking again at the 2 products mentioned above, Google Trends shows that the term “accountancy” is at its lowest search volume between 22nd-28th December, whilst Craft Beer is at its peak between 15th-21st December.
So how can you use this data? In Google Trends, you have a few options to play with, namely
Platform – web search, news search, YouTube, etc .
Time periods are given a score of 0 -100 on a time series graph. A value of 100 is the peak popularity for the term. A value of 50 means that the term is half as popular and so on and so forth.
We would recommend using google trends to support the data you gain from either your own historical campaign data or keyword planner.
What are competitors doing? – Auction Insights
Auction Insights relies on past paid search activity, so if you have no historical data, stick with Keyword Planner and Google Trends. However, if you do have some, then the Auction Insights tool in Google ads can be unbelievably valuable. Before diving into this a bit more, it is important to establish definitions of the metrics available.
Impression Share – This is the percentage of impressions that your ads receive compared to the total number of impressions that your ads could get. If you received 1000 impressions and 2000 impressions were available, the impression share is 50%.
Overlap Rate - This rate shows you how often your competitors ad appeared while your ad was displayed. If your ad had 100 impressions while your competitor’s ad had 30 Impressions on the same page as your ad, the system will show an overlap rate of 30%.
Position Above Rate - When your ad is displayed together with the competitors ad, this indicator will show you how many times your competitor ‘s ad was displayed in a better position than yours. When your ad is displayed 100 times on the fourth position and the competitor’s ad is displayed 20 times on a better position, then the position above rate is 20%.
Top of Page Rate - The top of page rate is the percentage of how many times your ad appeared above the organic search results.
So, what can you do with this data? When you bring the Auction Insights page up, you will see your metrics compared to your competitors. You can choose this feature at campaign, ad group or keyword level. Focus solely on the largest competitors as it is their movements that will have the biggest impact on your forecast. Also focus on those who you have the biggest overlap rate with, these are the ones that are operating in the same space as you. The most effective way to use this data is segment it by a time period of your choice. This will allow you to monitor how your competitors have behaved over time. Have they increased activity recently? If so, this could increase your cost per click, reduce your impression share and therefore impressions and ultimately your conversions. It might show the opposite in which case you experience the reverse impacts. Maybe your competitors’ activity changes with the season. Whatever you find, be sure to sense check against your own activity as sometimes what appears to be their movements could actually be a result of yours. For instance, you might notice your competitor’s impression share and average position decreased at a certain point in time and whilst this may be due to their activity, it could be down to an increase in spend on your part.
What are competitors doing? – Third party competitor research tools
Whilst the Google ads auction insights tool is valuable, what it can’t show is competitors behaviour on keywords you are not bidding on, the tool also does not give any indication in terms of what ad messaging your competitors are using.
Tools such as SEMrush, SE Ranking, and many more, will allow you to input a competitor domains and review the keywords they are bidding on, estimated spend (Although take this as a pinch of salt), estimated CPC’s and volume. If you find your competitors gain a larger volume of traffic from a set of keywords, it is likely these keywords are their top performers which you may also want to bid on. The tool will also help you understand groups of keywords where there is less competition which you can capitalise on. You can really start to understand your competitor’s strategy including keyword portfolio, ad messaging and approach to seasonality.
Ready to create your forecast?
You are now in a position where you can formulate a prediction that will answer those questions we highlighted at the start of the blog. In most cases, the more information you use, the more accurate you will be. However, avoid making over complicated calculations. Unfortunately, there is not a cookie cutter method to formulating a forecast, here are a few scenarios based guides.
How many sales will I get from my signed off budget? If you have a signed off budget and want to understand how many sales you are likely to generate in the next month we would suggest the following. a. Where historical campaign data exists, if you have spent the same level of budget previously then use your campaign data, take the last 30 days CPC, conversion rate and CTR averages to forecast out your results. Check there have not been any significant changes in those metrics in the last 7 days which you may need to factor in and sensor check with Google’s keyword planner to check whether you should expect a dip in search volume in the upcoming month. b. Again, if historical campaign data exists but this time you haven't spent that level of budget before, I would suggest you use the same approach as above, but this time add an extra step. Use the impression share Lost to budget column in the interface to understand how much impressions you would deliver if you were to maximise your impression share. For example, right now you may be losing 30% impression share due to budget. If you can work out how many impressions 100% is you can work out a total impression volume, you can then estimate the volume of clicks based on your avg. CTR, and your estimated spend based on your average CPC. Finally apply your conversion rate to understand how many sales that would generate. You can then apply your AOV to understand the revenue value and then ROI. If you still find your still below the signed off budget, then you will want to be following the steps in point C below. If you have no historical data, also start with point C below.
If I increase spend, what does ROI look like? If an uplift in budget is available, you will want to start with steps A and B. If you find you're still below the signed off budget, then you will want to be following the steps below. c. If you have a third-party tool available to you i.e. SEMrush you can start to build a keyword list (if it is a new account) by understanding the themes of keywords your competitors are bidding on. Here you can make some estimates in terms of Volume. You can then plug these terms into the Google ads keyword planner to understand volumes per month, estimated CPC’s, CTR etc. You can then make a judgement on conversion rate and AOV via your Google Analytics profile. If you have an existing account, you can carry out a keyword gap analysis via tools such as SEMrush to understand where you may have gaps in your current keyword portfolio. Again, whilst you can get impression volume, CPC’s and CTR estimates via Google Ads keyword planner, this time you will have your own campaign data to apply a conversion rate and AOV to forecast out. If you do have an existing account, you can also check out the opportunities tab in Google ads which will have suggestions on keywords, you can also check if you can identify any themes you may not already be bidding on based on your search query data.
How much should I be spending to generate an X% uplift in sales? First, what does X% more sales look like? Can this be achieved by following the steps in point B? If not, then try testing new keywords which were discussed in point C.
There is a bank holiday coming up, is that going to affect our performance? As mentioned previously in the blog post, if you have historical data use that first. What happened over the last bank holiday? Although try and compare like for like i.e. Christmas will be different from Easter. You may find you pulled back the budget last year but analysis after suggested performance was strong and therefore this year you would increase spend significantly. You may find performance dropped two days prior to the bank holiday so you need to pull back slightly earlier this time around? What did your competitors do? If you do not have any past data you can use either the planning tools in Google ads, Google trends or use third party competitor tools to understand how your competitors behave over these periods which will give you an indication of the opportunity or a sense of what direction you should take. It is also important to note that whilst you may or may not have PPC data specifically, behaviour of your organic site traffic will give you a clearer view of user behaviour as this traffic is not influenced by budget changes.
Whilst there are other scenarios which may come into play, we hope the above helps bring together some of the key ideas and data sources you can utilize when posed with the task of forecasting for clients and being able to form a strong foundation to base these off. As mentioned the key is not to overcomplicate things, whilst we are all aware there are a number of influencing factors on demand we can’t necessarily predict the future with complete accuracy, but the more data you have over time, the better your predictions will be.