My first post identified a sampling of the various data points that need to be considered when building your own ad revenue model to better forecast. The second post focused on current methods used by most major media, and why they don’t quite work as well as we’d like them to. This time around we’ll consider what you need to get started with your own customized ad revenue projection tool. I used my forecasting tool for 3 years, constantly keeping it up to date; even though some numbers started varying from month to month where projected and actuals were concerned, when I looked at quarterly or the annual forecast it ended up bang on with the actual yearend figures.
This forecast method is better than others for a number of reasons outlined at the end of this post. But rather than jump ahead to the answer, see next what level of granularity of data you’ll need to build your forecasting tool internally and try to figure out why this method could be better than others.
First off, you will need to forecast your website’s traffic and much more importantly its future reach. You can apply on a network, but it works much better at the granular site level as each is distinct and grows as separate speed in reach and revenues – better to perform the full exercise to get more precise indicators.
To forecast your website’s reach and traffic, you will need to have monthly numbers for this data going back as far as you can go with notes whenever relevant. Notes like the month when the website what revamped or major additions happened, when a new competitor became available and even when you ran campaigns in the past. The numbers will give you the real monthly growth, which can also be viewed year over year, and the note will qualify some of the spikes and dips along the way. My next post will be a detailed “how-to” go about setting up this information / data with formulas to start trending it for future forecasting.
In the meantime, if you want to set this up and want to get the most out of it, here is the data I recommend you start collecting (for as far back as you have access to – the more data you’ll have, the more solid your forecast will be; I worked with 5+ years of historical data for most website I managed). Also, please collect the data all from the same source – I recommend using your own internal analytics over comScore because this 3rd party had evolved its methodology in the past few years which probably has affected to varying degrees your website numbers. If you changed analytics tools in the past, make a note of when this happened to account for any discontinuities or data bumps.
– Monthly unique users of visitors
– Monthly total page views
– Monthly home page views
– Monthly section page views, for any section that’s of significant interest to you for sponsorship or integration purposes down the road.
Website reach and page view forecasting is a great tool to plan sponsorship packages for a ways down the road (like 6 months out or further). This will also help you predict when you might need to upgrade your servers down the road and avoid bad surprises.
Once you have all this data set up – which you will after you’ve collected it and applied it to next week’s “how-to” recipe – we can start looking at forecasting your website’s ad sales or ad revenues. To achieve this, you will need to collect the following data, also on a monthly basis. Where dollars are concerned please only consider sums that are net from any agency or third-party commissions – that wall they’re all on the same footing and can be forecasted without further adjustments required down the road. These data points should also go as far back as your analytics numbers (or as far as you can) – your forecasting will only be as good as your recorded history. Also, as for the analytics figures, please not any significant occurrences related to sales specifically like opening of a new sales office / team or replacement of more than just one member on a team, any major cross-market account movement (cross market can mean both geographic territories and local / national).
– Monthly actual (final and net) sales figure per sales team per website, or a clustered contextually similar bundle of sites.
– Monthly budget objective per sales team per website
Ultimately, the goal here is to generate a metric called RPMUV: Revenue per thousand unique visitors, on a monthly basis, per website and per sales team. This metric averaged across the same month from years past, coupled with your forecasted reach for that month (also averaged from the same month’s growth of years past) will give you a clear indicator of what to expect next year on that same month, assuming nothing major changes the course you’re on. Adding up every month can allow you to forecast with the same exercise reach & traffic per website along with revenue per sales team and total across multiple individual websites.
Why is this method (once you’ll have seen next week’s post on “how-to” forecast traffic and the week after that on “how-to” forecast revenue) better than others?
It is better for a number of reasons. The first and most important is that it relies on YOUR actual and historical facts, not gut feelings of people that are either biased, have a vested interest one way or the other (high or low budget growth) or don’t have the competence to estimate such things. It takes the human out of the equation’s basis. Of course once your forecast is built, you will need to adjust it on known or foreseeable future events / factors that could affect growth one way or another beyond the expected path, but that’s as far as you will include guess work in this endeavour.
Relying on your past numbers is also good because it considers growth based on your website’s current trajectory, not your desire of its future situation. It also relies on what you can really expect of various sales forces for specific sites / vertical types – again not what you think they should be able to do, just on their proven track record. If you can motivate them to surpass themselves, then all the better for you as you’ll bust your budget and have a banner year – but banner years cannot be expected, they need to be worked on to happen.
Where this method is concerned, if you either drop / loose or add new websites to your mix of responsibility / representation, the numbers from sites no longer in your portfolio will give you great insight into what you could expect of new sites (without extensive background knowledge) – same would go for a new sales team that is similar in makeup, territory and responsibility as another where you could rely on that other’s for an idea of expected performance.
Another reason why I consider this tool much more precise is that ultimately it relies on the one data that is most reliable: unique users. Ok we can bring up the argument that UVs reported by any analytics tool is actually way off the mark compared to a third party like comScore or Nielsen, however this is still the one piece of data that will be used to frequency cap campaigns, making is the most restricting piece of information you have. This, if you know a little about online advertising operations, is a primary piece of information what will more often than not affect the most any campaign you are asked to run for an advertiser. Given that advertisers and agencies are using frequency caps on a regular and growing basis, it is logical to plan you’re forecasting with their logic in mind, rather than inventory or ad impressions, which can explode and contract for any given month, with their being the same number of UVs. Impressions will come into the equation, but the UV count is the primary piece of information necessary. Ad campaigns are planned and budgeted with this piece of information.
My next post, as stated above, will focus on the precise method for forecasting your website’s traffic. If you want to be alerted as to its publishing, you should register to my RSS feed or my email alert, my Twitter account, LinkedIn, Facebook… (see the top and right side of my blog for all that.
The one following that will focus on the precise method of tying together your website’s traffic forecast to build ad revenue forecasting that’s built on a solid foundation.