Your CFO wants a number. Google wants you to spend more on Google. Your agency has its own agenda. However, if your eCommerce forecasting starts with any of those conversations, you’re already building on sand.
Watch and listen to the episode
John Readman has spent decades watching brands make expensive decisions based on top-down targets, conflicted media recommendations, and ROAS numbers that tell a comfortingly simple story in an increasingly messy world.
John is CEO and founder of Modo25, a digital marketing business that built ASK BOSCO, a platform that connects all of a brand’s data and uses it to predict where they should be spending their money.
Moreover, he’s been in and around eCommerce since before eCommerce was eCommerce: starting with postcode lookups, moving into email, then search, then helping brands make data-led decisions rather than gut-feel ones.
Your eCommerce forecast is built on someone else’s agenda
Most forecasts start with a target handed down from above (grow by 10%) and everyone below works backwards from there. Nevertheless, nobody stops to ask whether 10% is actually achievable, or even the right number.
John calls it “a love triangle of distrust” between brands, agencies, and media owners. Everyone in the room is conflicted. “If you go for a meeting with Google, the answer is normally to spend more money on Google.”
As a result, you end up with a plan shaped by competing agendas rather than an honest analysis of the opportunity. The fix: start bottom-up. Not “we want 10% growth, build a plan”, but “how much could we grow, and what would we need to spend to get there?” Different question. Completely different conversation.
ROAS got everyone addicted to the wrong thing in eCommerce forecasting
For a decade, the playbook was simple: spend £1 on Google Shopping, make £10, repeat. So everyone piled in. Gradually, the whole industry got hooked on the bottom of the funnel – buying intent, chasing last clicks, optimising ROAS as though it were the only number that mattered.
However, two things have broken that model. First, Google raised auction costs to fund its AI ambitions. Second, tracking stopped working properly due to privacy changes, cross-device behaviour, and browser rules. The clean attribution picture that CFOs got comfortable with is increasingly fiction.
Meanwhile, brands forgot how to do anything else. Brand building. Upper-funnel investment. Proper creative. Those muscles have atrophied. In 2026, that’s going to hurt.
Most brands have their traffic mix completely backwards for accurate forecasting
Best practice: 60% organic or direct traffic, 40% paid. Most eCommerce brands are the other way around. In fact, John’s team has analysed billions of rows of data, and the over-reliance on PPC (particularly Google Shopping) is systemic.
Furthermore, when you account for agency time and content investment, the cost-per-visit case for SEO is compelling. Brands just stopped believing in it after years of algorithm anxiety.
Here’s why this is urgent now: Google and Shopify’s Universal Commerce Protocol means people will soon research in Gemini or ChatGPT, see product recommendations, and buy with a single click inside the LLM. Consequently, you’ll see zero impressions, zero clicks, zero cost in your analytics.
If you’re not appearing in LLM-generated results today, you won’t be appearing in those buy moments either. The diagnostic is simple: find your best PPC keywords and check your organic rankings.
Not on page one? Not in any LLM citations? That’s the gap.
What predictive analytics actually does for eCommerce forecasting
Think of it as a roulette table. Instead of guessing where to put the chips, you use data to understand where the best odds are across all your campaigns and channels at once.
You define the win:
- maximum revenue
- maximum profit
- a specific cost-per-new-customer
And it tells you how to mix the budget to get there. Alternatively, you can reverse-engineer it: here’s the sales number, here’s what your media mix should look like.
However, John isn’t ready to put it on full automatic yet and is honest about why. “I’m not comfortable putting it in automatic mode and just letting it go off and spend the money.”
It’s the driverless car problem. The data may already show the machine makes better decisions, but trust takes time. Human in the loop for now.
But the other thing this gives you is something concrete to show the CFO: not “we want to try this,” but “based on two years of data, here’s what we expect to happen.”
What to do right now
Pull your best-performing PPC keywords and check where you rank organically. Not on page one? Not appearing in LLM results? You’ve found your priority.
It needs two teams (or two agencies) in the same room, which is why it doesn’t happen. Nevertheless, that’s also why it’s an opportunity.
Blatant plugs
Connect with John Readman on LinkedIn.
ASKBOSCO – AI analytics for better reporting, forecasting, benchmarking, decision-making, and reporting
Modo25 – Digital Marketing Agency & In-housing Services
The gist
- Many eCommerce forecasts rely on top-down targets that create distrust among brands, agencies, and media owners.
- Companies often misuse ROAS metrics, which leads to poor forecasting as they ignore upper-funnel investments and brand building.
- Most brands over-rely on paid traffic, hindering their forecasting accuracy; best practice suggests a 60% organic mix is ideal.
- Implement predictive analytics to understand budget allocation and mix for desired outcomes, ensuring a human presence in decision-making.
- Check your PPC keywords and organic rankings to identify gaps; stay visible in LLM-generated results for future success.
