Straight answer: SEO traffic forecasts are usually wrong because they’re built on search volume data that’s already out of date, CTR assumptions that don’t match your specific brand or SERP, and a sales incentive to make the number look good. I’ve seen forecasts miss by 80% in both directions, and the gap almost always comes down to one thing: nobody accounted for how messy real search behaviour is.
The forecast that started this whole rant
In 2022 I worked with a mid-sized insurance broker who’d been sold a content plan by another agency before I came in. The pitch deck promised 42,000 monthly organic visits within twelve months from a set of 60 target keywords around home insurance and landlord cover. Nice round number, clean chart, upward curve, the works.
Twelve months later, actual traffic from those pages sat at just under 6,000 visits a month. Not a disaster, but a 86% miss on the headline number the client had used to justify the entire budget. When I dug into the original forecast, the problem wasn’t laziness. It was structure. The agency had pulled search volume from a keyword tool, applied a generic CTR curve based on ranking position 1 through 10, and multiplied. That’s it. No account for the fact that three of the top ten results were price comparison sites with domain ratings above 80, no account for the featured snippets and “People Also Ask” boxes eating clicks before anyone reached position one, and no account for the client’s own domain starting from essentially zero trust.
The forecast wasn’t fraudulent. It was just a spreadsheet dressed up as a prediction.
Search volume data is already stale by the time you see it
Most keyword tools, including the big ones marketers rely on daily, show you monthly search volume averaged over the past twelve months, often with a lag of one to three months before that data even gets pulled into the tool. So the number you’re forecasting from could be describing search behaviour from 15 months ago. If a topic is rising fast (say, anything to do with AI tools in the last two years) or falling fast (anything tied to a trend that’s cooled), your forecast is built on a version of demand that no longer exists.
I’ve had clients in the AI training space watch search volume for terms like “how to use ChatGPT for business” swing by more than 300% within a single quarter. No forecast built six months earlier would have caught that, because the data simply wasn’t there yet.
Click-through-rate curves are generic and your SERP isn’t
Standard CTR curves used in forecasting assume something like: position 1 gets roughly 27-30% of clicks, position 2 gets 15-18%, and so on down to single digits by position 10. Those figures come from aggregated studies across thousands of keywords, not from your keyword.
But real SERPs are cluttered. A search for “best CRM for small business” today might show a featured snippet, three sponsored ads, a “People Also Ask” box, and a comparison table before any organic listing appears. By the time a user scrolls to position one, click-through can be half what the generic curve predicts, sometimes less. I checked this for a client last year: their forecast assumed 24% CTR at position one, actual CTR once they ranked there was 9%. The SERP had a large AI overview at the top eating most of the clicks before the organic results even started.
Competitors don’t hold still while you wait for rankings
Forecasts are almost always built as a snapshot: here’s the competition today, here’s where you’ll likely land. But a 12-month content plan takes 12 months to execute, and during that time your competitors are publishing too, sometimes faster than you, sometimes with more authority behind them.
I’ve watched a client’s target keyword go from position 4 achievable to page two unreachable within eight months because a competitor with a much stronger backlink profile ran a content sprint. The original forecast never had a mechanism for that. It assumed a static competitive landscape, which doesn’t exist anywhere in SEO.
Algorithm updates aren’t a footnote, they’re the main event
Google runs thousands of algorithm changes a year, and a handful of major “core updates” that can reshuffle entire categories of results. The 2023 and 2024 helpful content updates wiped out visibility for a huge number of thin affiliate and listicle-style sites, some losing 60-90% of their organic traffic within weeks. Anyone who’d forecast growth off the old rankings for those sites was instantly, badly wrong through no fault of their content calendar.
You can’t forecast an algorithm update because nobody, including people inside Google, can tell you when the next one lands or exactly what it’ll reward. Any forecast that doesn’t build in a wide margin for this is quietly assuming the algorithm freezes for a year. It won’t.
Here’s the bit most forecasting decks won’t say out loud
Traffic forecasts are frequently a sales tool first and an analytical tool second. If you’re an agency pitching a 12-month retainer, a forecast showing modest, honest growth of maybe 15-20% doesn’t close deals as well as a forecast showing a confident curve to 40,000 visits. I’m not saying every agency does this on purpose. Plenty believe their own numbers. But the incentive structure rewards optimism, and optimism dressed up in a chart looks exactly like data.
I’ve sat in enough pitch meetings to know that a forecast which shows flat or slow growth in month one to three (which is realistic, because content needs time to be crawled, indexed, and trusted) tends to get quietly smoothed out before it reaches the client. Nobody wants to open a sales conversation with “results will be minimal for the first quarter,” even when that’s true for almost every new content programme.
A more honest way to build a forecast
None of this means forecasting is pointless. It means most forecasts skip steps that would make them more useful and less impressive. If you want a forecast that survives contact with reality, here’s roughly what I do with clients now:
- Pull search volume for your target terms, then check trend data over 24 months, not just the average, so you can see whether demand is rising, falling, or seasonal.
- Manually check the actual SERP for your top 15-20 keywords, note snippets, ads, AI overviews, and shopping results, and discount CTR expectations accordingly rather than using a generic curve.
- Audit the top 5 ranking pages for domain authority and content depth, and be honest about how long it will realistically take your domain to compete, especially if you’re starting under a domain rating of 20.
- Build three scenarios, not one: a conservative case, a realistic case, and an optimistic case, and present the conservative case as the one you’re accountable for.
- Rebuild the forecast every quarter using real ranking and traffic data instead of treating the original projection as fixed for the year.
- Separate branded from non-branded traffic in your baseline, because branded search growth (people searching your company name) often gets mistakenly credited to the content plan.
If you want the fuller mechanics of building this before you commit budget to a content plan, I wrote a longer walkthrough on how to forecast SEO traffic before you commit to a content plan, which covers the modelling side in more depth than I can fit here.
What I tell clients now instead of a single number
I stopped handing over one confident traffic number years ago. Instead I give a range and I explain the assumptions behind each end of it out loud, in the meeting, so nobody can later pretend the number was a promise. Something like: “if our content ranks in the top five for these terms within nine months and the SERP doesn’t change much, expect somewhere between 8,000 and 14,000 monthly visits. If two of these keywords stay dominated by comparison sites, closer to 5,000.”
That’s a less exciting slide. It’s also a number I can defend a year later, which matters more to me than looking impressive in month one.
Frequently asked questions
Why do SEO traffic forecasts miss so often even from experienced agencies?
Because forecasts are typically built from generic search volume and CTR data rather than the specific SERP, competitive landscape, and domain trust involved, and none of those factors stay fixed for the length of a typical campaign.
How accurate should I expect an SEO forecast to be?
Treat any single-number forecast as a guess dressed up in a chart. A range with clearly stated assumptions, rebuilt quarterly against real ranking data, is far more reliable than a fixed 12-month projection.
What causes the biggest forecasting errors: keywords, competitors, or algorithm changes?
Algorithm updates cause the biggest single swings, sometimes 60-90% traffic drops for affected sites within weeks, but stale search volume data and generic CTR assumptions cause the most consistent, everyday inaccuracy.
Should I still ask for an SEO forecast before starting a content plan?
Yes, but ask for the assumptions behind it, not just the number. A forecast that shows its working, including SERP features and competitor strength, is worth far more than one clean upward curve.
Want the complete version? Read where I break down SEO.
