Reflections on Forecasting for SEO
Forecasting is an area I'd managed to avoid for at least half of my career, but for the last few years I found the need to embrace and engage with this area more and more.
As an area this is notioriously problematic with many people simply refusing to take part, I'd argue that you couldn't & shouldn't avoid it. Forecasting - when pitched right - can help prove the potential of SEO, but you must realise you are not predicting the future. All parties involved need to be 100% on this one.
This is the first of a series of posts where I want to share some of my thoughts around different SEO aspects. It's not quite a podcast, nor is it a blog - but there's a recording and a transcript. You can listen to the recording on Soundcloud or keep scrolling for an (edited) transcript.
Forecasting For SEO Transcript
I've been an SEO for a while and I think for the vast majority of time, I've been an SEO. I very much disliked forecasting, for a number of reasons.
Forecasting makes me very nervous and most of my reasons behind why I dislike forecasting will also be met with annoyance and frustration by most C-suite. Predicting the future isn't possible, lack of results is hard to navigate and making the numbers meaningful is challenging. That said, forecasting has a certain necessity behind it, we all know that forecasting is imperfect - so we have to deal with it.
"So just please stop complaining and do a forecast."
This may or may not have been said to me in the past, I may or may not have said that to other people in the past, but you catch my drift - It's not an easy area.
From an SEO perspective, the main factor that makes forecasting uncomfortable is because of algorithm updates. We are talking about our ability to forecast the amount of people that will come to our website from a search engine of which we know worryingly little.
There's a lot around Google and SEO that is written and broadly acknowledged, but I think we gain a lot more confidence out of the shred of knowledge we actually have versus the deep complexities and machinations of it.
What's our ability to predict the future?
When we're talking about a forecast and we say "well, what's our ability to predict the future?" Typically a forecast will - whatever method we use - look at all the data that has come before and we'll generally approach with "based on this data, here's the best guess of what will happen."
Now that's a perfectly reasonable kind of statement and a perfectly reasonable way to position a forecast. For most people this should be unproblematic, but I've seen so many forecasts that have been ruined by algorithm updates.
"Ruined" can go either way by the way. "Ruined" means a forecast that is not correct or accurate. I don't mean, ruining is, "it's only underperformed versus forecasts" if you're over performing the forecast it means you've done really well or your forecast wasn't that accurate. Either way, we need talk about that.
Because of the variance of Google algoritms and the ability that Google has had to completely shake up the fortunes of businesses based on rankings, people are are very unsure about what to do. But another issue here are the implications of a forecast in SEO and how SEO is viewed as a source of traffic or source of revenue.
This is so often also makes us pretty reluctant to give a forecast.
I have many times in my career been told "if you do not hit the numbers, you say you will, from an SEO point of view, you can help decide who I fire because I've only employed that person because of what you've said, you'll do from an SEO point of view."
Now that conversation is exceedingly uncomfortable, it is a misuse of a forecast and the spirit in which a forecast was intended. That is the grim reality of it, if any marketing channel needs to deliver a certain amount of revenue and doesn't, then other parts of the business may get lost.
"...if you do not hit the numbers, you say you will, from an SEO point of view, you can help decide who I fire..."
Part of the issue with SEO, is how forecasts are often deployed. SEO needs to provide is "how much more traffic, revenue (or both) are search engines going to provide to me?" That's a hard question. You can start to build on it, you can make some assumptions, you can look at what's happened, but the issue at the very heart of that question is the assumption that everything that SEO is, brings benefit in palpable traffic or a revenue point of view.
This undersells and downplays all of the functions that SEO have in an organisation - without even discussing brand benefits of content strategies & digital PR.
The KPIs of traffic & revenue aren't kind of deep or full enough to express the full benefits of the channel.
Isn't SEO more than just organic traffic?
Technical SEO focuses on a website and how the website functions. What are the websites features? How do we deploy it to the web? How to search engine receive it?
A lot of them these points are beneficial to users - as well as search engines - and they should be. If you're doing it right, everything that benefits a website from an SEO point of view should benefit a user at some point in the funnel.
But when you're talking about forecasting from an SEO point of view, the satisfaction of a user is not something you're actually measuring or forecasting.
[just traffic or revenue] undersells and downplays all of the functions that SEO have in an organisation. These KPIs aren't kind of deep or full enough to express the full benefits of the channel
You could argue the more satisfied the user is, the more traffic and rankings you will get eventually, but these are fundamentally different things.
Forecasting often misreports on all of the things that SEO can do or because those other benefits are not included in the forecast.
When we're just talking return on investment (ROI). We just talk about it in monetary terms because that's easy, but we then miss it all the value proposition and you can get yourself in a vicious circle.
SEO is not showing an obvious return (or we can't forecast it all), therefore we're not going to be investing any more into it, and performance gets worse. The cycle continues.
when you're talking about forecasting from an SEO point of view, the satisfaction of a user is not something you're actually measuring or forecasting.
I'd argue this issue is more than that because that's an education problem. If you're being asked to deliver a forecast and you think that forecast doesn't show the full benefit of the efforts of your channel, that's up to you to educate for that.
You can do it, but if someone is asking you for a forecast, you need to deliver one. This is where I think forecasting has a bit of a PR problem. There's a lot of individual impressions about forecasting, what it can do & what it shows - that somehow it does predict the future.
The "nothing's different" model
I think calling it a forecast is part of the problem. It can't ever credible because you can't be aware of some of the most significant variables (algorithm changes).
You can create a scenario, however. You can build a forecast based on historic data to show you if nothing else changes, this is our "nothing's different" model, our baseline. Where we hope the future will go and then SEO uplift is the forecasting element that's been layered on top. We're not saying "this will happen", just that in this version of the future, this is how our benefits will be felt.
Very often SEO forecasting conversations are also at the start of the, "I need more budget. I need more resources conversations", using a scenario like this to show you roughly how much extra we can expect to get is key. There's a bit of a nuance difference - whilst you could argue that still is a forecast - I would say that you're just offering someone a scenario and illustrating your ability to change that.
We don't know what the algorithm is going to do. We don't know how that's going to change anything. And obviously that could ruin your forecast within days or weeks or months of it being produced. But at least your predictions of the levels of uplift you may see can still hold true.
How exactly you put numbers to that, how you work it out is all dependent on you, your scenario, and what you're trying to forecast on.
For everyone who's out there that refuses to or doesn't want to produce a forecast, I think you're going to struggle. A forecast are the table stakes for most meaningful conversations, whether you're trying to convince your board to change, whether you're pitching your business into someone else's, you need a forecast to weigh up what you'll get back from it.
If you're refusing to provide a forecast because you don't think it has the value, then good luck having that conversation to convince them that they're wrong for asking you for a forecast.
It's possible, if you feel that strongly, that you can educate them, then this is your opportunity to do so, but there's the "yes, and" approach to it. "Here's your forecast, and here's where we need to supplement that". Don't say, "no, you're not having your forecast because it's not gonna work that way".
That's my, take on it, I think there's lots of tools out there for forecasting that I'll leave links to them below. I don't think we should be avoiding it. Here's an area we need to tread really carefully on, but just think about using the forecast to create a version of the future if nothing were to change and then ask people, are they happy with that? And then are they happy with the level of change you think you could bring to that and maybe approach from there see how it goes.
As usual, any comments, any feedback, any questions, drop me a message. Leave me a comment. And hopefully we'll speak again soon.
Useful Links on Forecasting
In the transcript I allued to a few resources to help get on top of forecasting for SEO
The Anvil - https://www.forecastforge.com/anvil/
Learning SEO - "SEO Forecasting Section here - https://learningseo.io/deepen-knowledge/
Facebook Prophet - https://facebook.github.io/prophet/
Andrew Charlton, Forecasting for SEO - https://bertiecharlton.gumroad.com/l/forecasting-for-seo