Marketing can't fix a broken business.

What a Tokyo conference reminded me about ROAS, feeds and the rest of the system.

The third conference I went to in Tokyo last month was on supply chain analytics for digital markets. It might sound far from performance marketing, but it ended up being one of the talks I thought about most after coming home.

The reason is simple: most of the time when ROAS drops, it's not really the campaigns' fault.

The reflex of blaming the campaigns

In performance marketing, there's a reflex that kicks in when results go down. We open the dashboards. We look at the campaign. Did the bid change? Did the auction get tougher? Is there a creative issue? Is the audience saturated?

These are real questions. Sometimes the answer is yes.

But often, the campaigns are doing exactly what they were doing last month - and the numbers are still off. That's when I start looking somewhere else.

Things that change ROAS that aren't your campaign

A few things I see often:

• The best-selling product is out of stock. The algorithm keeps showing it because it was strong, but conversions stop.
• The product feed has issues. Missing GTINs, wrong prices, outdated availability. PMax especially is allergic to messy feeds.
• The margin shifted. Sales went up, but profit went down. ROAS looks fine on screen - but the business is losing money.
• Delivery times changed. Customers add to cart, then drop off at checkout because shipping went from 2 days to 7.
• A competitor undercut pricing. The campaigns can't compete on what's on the landing page.
• The product page was updated, and something broke.

None of these are "campaign problems". They're business problems that show up in marketing data.

This is what the Tokyo conference reminded me of. The campaigns sit at the end of a long chain. If anything earlier in the chain breaks, the campaigns can't fix it.

Marketing and business data should talk to each other

For a while, I treated marketing dashboards and business dashboards as two separate things. The marketing one had Google Ads, Meta, GA4. The business one had stock, margin, fulfilment, returns.

I don't do that anymore.

The most useful dashboard I've built recently for an e-commerce client has both. ROAS sits next to stock levels. CPA sits next to current margin. Each campaign is tagged with the products it pushes, and each product is tagged with its current availability.

When something drops, it's much faster to figure out why.

I don't have the perfect setup. Most clients still don't have everything connected. But even a partial view changes the conversation.

When I can show a client that ROAS dropped at the same moment the best product went out of stock, the conversation changes from "fix the campaigns" to "fix the stock". That's usually the cheaper fix.

PMax specifically

For PMax campaigns, this is even more obvious.

PMax is opaque on purpose. You don't choose which products to push. You don't choose the placements. You give it a feed and it decides.

So if the feed is wrong, PMax is wrong. If the feed prioritises low-margin products, PMax burns money on them. If the feed has stale availability, PMax sends people to "out of stock" pages.

I've helped a few clients audit their feeds. Most of them found dozens of small issues. Fixing those - not the bidding strategy - often moved the numbers more than anything I could change inside the campaign.

The bigger point

Honestly, what stayed with me from the conference is pretty simple: marketing performance is a symptom, not the cause.

If the business has a problem - stock, pricing, delivery, product quality - marketing will surface it as bad numbers. We can try to compensate by spending more, but that usually just makes the problem more expensive.

The better question is not "how do we improve ROAS?". It's "what's actually breaking, and where?".

When I'm working with a client now, the first thing I want to see is not just the campaign data. It's the stock, the margin, the delivery times, the returns, the support tickets. If any of those look off, I want to fix that first.

Takeaway

The Tokyo conference made me more careful about one thing: assuming marketing problems have marketing solutions.

Sometimes they do. Often they don't.

If your numbers drop and the campaigns haven't changed, look at the business behind them. The feed. The stock. The margin. The product page. The delivery promise.

Better marketing isn't always about spending more. Sometimes it's about fixing what's behind the campaigns first.

That's not a glamorous insight. But it's the one I keep coming back to.