Good morning, まさきん here.
There’s a line that’s been well known in the advertising world for a long time. It’s often attributed to the department store magnate John Wanamaker: “Half the money I spend on advertising is wasted.” The idea being, half of every ad budget goes to waste — the catch is, nobody knows which half.
Marketing Mix Modeling, or MMM, is one attempt to prove that line wrong. It feels like it’s been getting renewed attention lately, so I wanted to lay out why.
Lately it’s become common to see services that used to live entirely online start putting money back into offline advertising, like TV commercials. Once digital-first companies start flowing back into mass advertising, measuring performance with online metrics alone stops being enough. I think that’s part of what’s driving the growing demand for a method that can evaluate offline and online efforts together, as one picture.
What Marketing Mix Modeling is
It’s a technique for statistically modeling how various marketing factors — ad placements, promotions, competitor moves, seasonality — affect sales.
What sets it apart is that it can handle not just clean, trackable numbers like clicks and conversions, but also factors that are hard to measure directly, like TV commercials or in-store promotions. I’ve touched on a similar risk before — judging things purely by whichever metric happens to be easiest to track — when I wrote about what CRM is really about.
It used to be something only a handful of companies could pull off
When I first came across this idea, my impression was that the barrier to actually putting it into practice was pretty high.
You need to gather not just clean numbers like clicks and conversions, but also fuzzy, hard-to-quantify factors. On top of that, working out how to allocate costs across each media channel and ad slot gets complicated fast. Honestly, it was a lot of grinding work.
My sense was that adoption was limited to a handful of cutting-edge companies — the ones with especially large mass-advertising budgets, or with former ad-agency people already on staff. There’s also a quirk specific to the advertising industry: the more advertisers understand modeling, the harder it becomes for agencies to sell them their services. So agencies were never exactly rushing to evangelize the method themselves.
I’ll admit I have a similar streak myself. Whether it’s the ad budget at work or the fixed costs at home, I tend to decide things by feel. My phone plan is a good example — I signed up once and never got around to comparing it against anything else. Maybe the habit of reviewing things by the numbers pays off in both places more than we’d expect.
Clicking this opens the Rakuten login page. Once you log in, you’ll see the campaign details.
Why it’s back in the spotlight in 2026
I feel like the landscape has shifted over the past few years. That said, I’ve recently come around to thinking it’s a bit misleading to chalk the reason up to just “cookie restrictions made click tracking harder.”
In reality, Google walked back its plan to phase out third-party cookies in Chrome, and as of October 2025, it has shut down most of the Privacy Sandbox APIs that were the centerpiece of that plan — only a handful of pieces, like CHIPS, FedCM, and Private State Tokens, are still standing. In other words, tighter restrictions on the Chrome side haven’t progressed nearly as far as originally expected.
Even so, browsers like Safari, Firefox, and Brave have blocked third-party cookies by default all along. Together they’re said to account for a bit under 20% of traffic, and some estimates put the effectively cookieless share of all traffic at somewhere around 15 to 35%. In other words, a situation where last-click alone can’t fully explain “which ad actually worked” hasn’t gone away, even after Chrome’s about-face.
There’s another concrete development that backs up the renewed interest in MMM. In March 2024, Google itself released an open-source MMM tool called Meridian. It’s the successor to the earlier Lightweight MMM, built for engineers and data scientists, and it supports things like reach-and-frequency modeling and folding domain knowledge into ROI parameters. To me, the fact that the measurement platform itself is investing in MMM is far more convincing evidence than any amount of talk about cookie restrictions.
The growing availability of more approachable analysis tools and data is helping too, I think. I wouldn’t say the difficulty has dropped exactly, but it does feel more within reach than it used to. I felt a similar shift — the sense that the bar had dropped from where it used to be — when I wrote about the 3 elements of personalization. For anyone working hands-on with advertising, Nielsen has published a free rundown of the modeling concept along with a practical checklist, which should be a good starting point if you just want the big picture first.
Companies where adoption is moving forward seem to share a few traits. One is simply having a mass-advertising budget large enough that the cost of modeling pays for itself. The other is having agency alumni or data scientists in-house, so they’re not entirely dependent on an outside black box and can check the model’s validity themselves. Put the other way around, if your ad budget is still fairly small, it might be more realistic to hold off on a full rollout and start instead by taking stock of how much of your own advertising data you actually have, using something like Nielsen’s checklist.
The difficulty that still remains
That said, the difficulty of translating offline impact and branding effects into numbers hasn’t really changed much.
Force a factor that can’t really be quantified into a number, and it can actually undermine how convincing the model is. Actually building a model out tends to be a long-haul project, one where you end up working closely with consultants or data scientists over an extended stretch. On top of that, you need buy-in from the media outlets and agencies you advertise through, as well as from the relevant teams inside your own company — and getting everyone aligned on that is, I’d say, just as hard as collecting the data and building the model, if not harder. Even as the tools keep improving, this kind of balancing act still seems to come down to human experience in the end.
Wrapping up
Not relying purely on gut and experience, and not just chasing click counts either — as a method that lands somewhere in between, I think Marketing Mix Modeling is only going to matter more from here.
Whether it’s how a company allocates its budget or the fixed costs at home, it’s worth taking a number-driven look every once in a while. Your phone plan might well be one of those things too.
Clicking this opens the Rakuten login page. Once you log in, you’ll see the campaign details.
This article contains affiliate advertising. If you sign up for a product or service through a link on this site, we may receive compensation from the partner company. The site operator is also an employee of Rakuten Group and may receive compensation through an employee referral program. The content and opinions here are based on the operator’s own experience and research regardless of any advertising relationship, but please read with the above in mind. See the Disclaimer & Affiliate Disclosure for details. Disclaimer & Affiliate Disclosure
This article includes machine-translated content. Please check the official Rakuten Mobile multilingual page for exact terms.
If you have concerns about Rakuten Mobile coverage or signal quality, you can consult through the official signal improvement request form .
