My Agency Handles Our Feeds. What Should I Be Asking Them?
Your agency keeps the feeds running. But who owns the meaning inside them? Five questions that move the conversation from format to data.
If an agency manages your channels, your feeds are probably in good operational hands: formats validated, rejections handled, campaigns fed. That work is real, and this article is not about second-guessing it.
It's about a conversation that tends not to happen — because it falls between jobs. Your agency inherited your product data; you assume they're optimizing it; they assume it's yours to fix. Meanwhile, every insight from this series applies to your catalog too: platforms now interpret what they receive, and interpretation depends on the data, not the delivery.
For agencies themselves, we've seen this from the other side: the recurring cost of feed work stopped being formatting years ago — tools solved that. What remains is the data-quality conversation they have with every client, every quarter. These five questions are that conversation, started by you.
The five questions
1. "Where do our categories come from — and do they resolve?" Not "are they filled in," but: do our products carry structured, resolvable categories, or generic labels the platform has to guess around? This is the single highest-leverage question — in our research, category resolution consistently emerges as one of the strongest signals of whether a product can be interpreted at all. If the answer is "we map whatever the store has," the store's ambiguity is traveling untouched.
2. "When products get disapproved or misclassified, is it format or data?" A good agency fixes format issues quickly. Ask what share of recurring problems are actually data problems — valid values that don't carry enough meaning. If nobody has ever split the problem that way, the second category is invisible by definition (Article 2 of this series is the longer version).
3. "Does the same SKU tell the same story on every channel?" Same category, same attributes, same description everywhere? Or has each channel's feed drifted on its own schedule? Cross-channel inconsistency quietly erodes credibility with the platforms — and it's architectural, not a diligence failure (Article 3).
4. "When a field is missing, what fills it?" The line that matters most: is enrichment structuring what we published, or estimating what's probably true? If any tool in the chain fills gaps with model-generated guesses, understand that those guesses ship under your name, as statements of record about your products (Article 4 — we're absolutists here: verified, never invented).
5. "Can you show me one of our products the way the machine sees it?" Not the dashboard — the reader's view. Which fields resolve, which are silent. If that view doesn't exist anywhere in the reporting, then everyone is managing the human performance and nobody is watching the audience that increasingly decides visibility (Article 5).
What good answers sound like
You're not looking for perfection — you're looking for whether the data layer has an owner. Good answers name specifics: how categories get resolved, where enrichment comes from, what's verified versus inferred. Vague answers ("the tool handles that") mean the meaning of your catalog is nobody's job.
And there's a genuinely good outcome available here: an agency that delivers feeds built on enriched, verified data changes what it's selling — from pipe maintenance to catalog intelligence — without changing how your store is built. Some of the best conversations this series can start are between a merchant holding these questions and an agency ready to answer them.
One neutral way to ground that conversation: run the free Agentic Catalog Readiness Audit™ together. One real product, seen the way machines see it, before and after. Ten minutes, and questions one through five suddenly have a shared exhibit.
The feeds are running. Make sure someone owns what's inside them.
Find out where your catalog stands
→ Run the free Agentic Catalog Readiness Audit™ — one real product, machine's-eye view, before and after. Bring it to your next agency call.
→ Read the complete Pillar — AI-Ready Commerce Feeds™ — the full picture this series walked through. (Series complete.)
Continue the series
← Previous: Why Does the Price Follow My Catalog, Not My Channels? · ⌂ AI-Ready Commerce Feeds™ (Pillar) · → Next: Read the complete Pillar → (series complete)
Series: AI-Ready Commerce Feeds™ (PI-PL002) · Knowledge Domain: Product Intelligence
NSOLVIA Intelligence — Products generate knowledge. Knowledge generates authority.