Many enterprise communities are experiencing a significant decline in search traffic.
Or, as it’s known in the search industry, an alligator graph. The number of impressions increases (or stays approximately the same), while the number of clicks plummets. You can see an example below:
This is because Google increasingly shows the answer within search without people having to click the link to visit the community.
You can see what this looks like here:
On the surface, this seems like a bad thing. Fewer people are arriving at our enterprise communities.
And in some ways it is.
When search traffic represents 70% to 90% of all traffic for most enterprise communities, this means fewer visitors, fewer people browsing, and fewer people likely to ask a question in the community when they don’t find an answer.
Yet, for established communities, members are still getting answers to their questions from the community. The value the community provides hasn’t changed much.
But now they get these answers without having to visit the community.
And this presents a problem.
Many brand communities are measured by cost savings that accrue from members getting answers to their questions instead of contacting costly support channels.
You can see the methodology here:
In short, you estimate the number of people who visited an answered question, use a pop-up poll to estimate what % got the solution to their problem, and if they would have contacted support if they hadn’t, eliminate those who are out of warranty (often optional), and what’s left is the number of calls deflected.
You then multiply this by the average cost of a call/ticket to get an answer.
The problem with this is that search traffic is the critical variable. It only counts people who got the answer when they visit the site. When search traffic goes up or down, so does the number of calls deflected (and thus the value of the community).
Now that Google is keeping more traffic for itself, the value of many communities using this formula is plummeting.
What should we do about that?
It’s Not A Traffic Problem, It’s A Measurement Problem
There’s an important thing to remember (and keep telling your colleagues here).
Your customers are still getting answers from the community.
They’re not just getting those answers in the community.
The community is still delivering the same amount of value, it’s just doing it through new mediums.
In short, your community doesn’t have a search traffic problem. Your community has a measurement problem. And a measurement problem is, thankfully, a much easier problem to solve.
A few people have mistakenly attempted to address this issue by implementing various optimisations aimed at increasing traffic to a community.
But sadly for most mature communities, that ship has sailed. The new, lower traffic era is here to stay. Trying to reverse a traffic decline is like trying to change the tide – you’re not going to win.
As community strategists, we always recommend embracing trends rather than fighting against them.
So the challenge is, how do we measure the value of ‘ghost deflections’?
We know it’s happening, we just can’t see it.
Measuring Ghost Deflections
My former colleague Pawel used to say the challenge with measuring deflection is you’re trying to measure something that didn’t happen.
That was hard enough when we had clean community data to begin with; it’s significantly harder when we don’t even have that. But it’s always a fascinating challenge to solve.
In a perfect world, we would all run tests where we shut down a community for a few months and see the result. The data is hard to refute.
The reality is that’s not always possible. Not many people like to run extreme experiments to prove a point (sadly).
So instead, we must develop a new formula that estimates how many people would previously have visited the community
You Need To Estimate How Much Traffic Is Being Siphoned By Search Engines
This is the key question.
How many people would have visited the community if search engines hadn’t begun showing complete answers in search results?
For months, I was stumped on this until my friend Dennis Pollet from Microsoft Answers, shared his exact methodology that Microsoft uses to determine how much traffic they lose.
I’ve tried to simplify the key steps, but Dennis and his team deserve all the credit.
In short, you need to:
- Find out how often your community is appearing in search results (from Google Search Console)
- Use SEO tools to estimate how frequently those results appear in the AI overview and featured snippets.
- Remove the instances where it didn’t appear in the top 3 results (and weight the times it did*)
- Remove the people who clicked to visit the site anyway.
And what’s left is the traffic you would have gotten.
This is the number you can combine with visits to an answered question above to get the actual answer to the question. What’s fascinating is that when Microsoft Answers performs this calculation, it almost entirely counterbalances the traffic decline over the past year.
This is why you need to measure this!
It proves the community is still delivering the same amount of value as before.
Now it’s not perfect. There are plenty of assumptions baked into this. It also doesn’t include traffic lost to LLMs like ChatGPT and doesn’t cover other search engines.
However, right now, it’s the best way to measure the full deflection value of an enterprise community.
I strongly recommend you watch the full webinar with Dennis to dive into the complexity of this. It’s not the easiest methodology to grasp, but once you do, it will give you the data and the processes to explain the value of your community internally.
And if you want help setting up this methodology, contact FeverBee.
P.S. Also, check out our first webinar with Microsoft, where Dennis shares how they estimate $80m+ in call deflection.
