Google Shopping is the one paid channel where most of the work is not in the ads platform. It is in the product feed, and the feed belongs to whoever owns your product data. That single fact reorganises the in-house versus agency question into something more useful than a cost comparison.
The generic version of this decision — agency, freelancer or in-house — applies here too. But Shopping has an unusual shape, and getting the split wrong is the most common way the channel underperforms for reasons nobody can locate.
The four jobs inside “Shopping”
1. Feed and Merchant Center health. Continuous, unglamorous, and the highest-leverage work in the channel. Required attributes present and correct. GTINs where the product has one. Disapprovals cleared — policy flags, price mismatches between the feed and the landing page, image problems. Titles structured so they match how people search, because Shopping has no keywords and the title is most of what determines which queries you appear for. Supplemental feeds and feed rules to fix things you cannot change in the source. Custom labels so campaigns can be segmented by margin, season or stock position rather than by category alone.
This never stops. Prices change, stock moves, products get added, and every one of those events can break something.
2. Campaign structure and bidding. Less work than most people expect now that Performance Max absorbs much of it. Still real: deciding what belongs in Performance Max versus Standard Shopping, segmenting by margin instead of revenue, negative keyword work as the primary targeting control, and knowing when Performance Max is the right container at all. Episodic — intense for a week, then light for a month.
3. Price and promotion competitiveness. Merchant Center will show you where your price sits against other sellers on the same product. If you are consistently above the market, Shopping will tell you by quietly not converting, and no campaign change will fix it. This is a merchandising decision, not a marketing one, and it has to sit with someone who can actually change prices.
4. Measurement. Getting margin and returns into the conversion value rather than optimising against gross revenue, which is how Shopping accounts end up profitably scaling their worst products. Related to how you calculate ROAS in the first place, and more consequential in Shopping than almost anywhere else because the product mix shifts under you.
Which of those outsource well
The split follows the rhythm of the work, not the org chart.
| Job | Rhythm | Better home |
|---|---|---|
| Feed and Merchant Center | Daily, reactive | In-house (or internally owned) |
| Campaign structure and bidding | Episodic, deep | Outsourceable |
| Price competitiveness | Continuous, commercial | In-house, necessarily |
| Measurement setup | Project, then stable | Outsourceable |
Feed work wants to be internal because it is coupled to systems you control and because response time matters. A disapproval on your best-selling product the night before a sale needs someone with access to the PIM or the ecommerce platform right now, not a ticket in somebody’s queue. It is also closer to data engineering than to media buying, which means the person who is best at it may not be in your marketing team at all.
Strategy and measurement outsource well because they benefit from pattern recognition across many accounts, and because they are bursty. You do not need forty hours a week of campaign restructuring. You need someone good for a few days a month, who has seen what happens at your SKU count.
The threshold question
People ask about this in terms of spend. SKU count predicts it better.
- A few hundred stable SKUs, at almost any budget: maintainable alongside other duties. The feed barely moves. Internal ownership plus an outside specialist a few days a month is normally the best value available.
- A few thousand SKUs with regular price and stock movement: the feed becomes someone’s actual job, part-time. Strategy still outsources cleanly.
- Tens of thousands of SKUs: a full-time data job regardless of spend, and often an engineering one. At this scale the feed is the channel, and an agency that does not touch it is managing the visible 20%.
Spend matters mainly for whether you can afford the senior time — the in-house versus external cost comparison holds here. But a $100k-a-month account with 200 SKUs is a much smaller operational job than a $20k-a-month account with 40,000, and budget alone will mislead you about which one needs a hire.
What to check before you outsource Shopping specifically
Beyond the usual checks on any outsourcing arrangement, three questions are specific to this channel:
Do they touch the feed, or only the campaigns? Many agencies quote Shopping management and mean campaign management. That is a legitimate scope, but if nobody owns the feed, the most important work in the channel is unassigned. Ask explicitly, and get the answer in the scope document.
Who clears a disapproval at 11pm before Black Friday? The answer defines the arrangement more than the retainer does. If it is you, you need internal capability regardless of what you are paying them.
What do they optimise to? If the answer is revenue or ROAS without margin, they will scale your highest-revenue products rather than your most profitable ones, and the account will look excellent while contribution falls. This is the most common quiet failure in outsourced Shopping.
The arrangement that usually works
Internal ownership of the feed and Merchant Center, including someone with access to the product data and authority to change it. An outside specialist for structure, bidding, query analysis and measurement, working a few days a month against an agreed scope. Margin in the conversion value, so both sides are optimising to the same number.
That split is cheaper than a full-time hire for most catalogues, and it puts each job where its response time and its skill requirement actually point. The failure mode it avoids — handing the whole channel out while keeping no internal feed ownership — is the one I see most often, and it is usually diagnosed as a campaign problem for several months before anyone opens Merchant Center.
If Shopping is collecting impressions and not converting, the diagnostic order starts with price competitiveness and tracking, not campaign structure.
If you want someone to tell you which of these four jobs is actually broken in your account, that is what an audit is for.
