Rakuten Marketing Agency or Full-Service Operations: Where the Work Splits

A quick boundary before we start, because two adjacent questions get confused with this one. Whether you need a paid agency at all when Rakuten has already assigned you an e-commerce consultant is a different article — see Rakuten's ECC versus a paid agency. Whether to run one agency across Rakuten, Amazon Japan and Yahoo! Shopping or a specialist per marketplace is a separate decision about how many platforms one vendor covers. This article assumes you are paying someone, on one platform, and asks a narrower question: within Rakuten, can the advertising function be bought separately from the operations function?

The honest starting point is that this is not a new idea on our site. Our guide to outsourced Rakuten store management already lists partial or modular outsourcing as one of three legitimate models, and notes in passing that the functions handed over first are commonly customer service and RPP ads — with the caveat that someone has to own the seams between what you keep and what you delegate. This article is that caveat, expanded, because on Rakuten the ad-versus-operations seam turns out to be unusually hard to draw.

The Cheapest Quote on Your Desk Is Usually the Ad-Only One

If you request proposals for Rakuten, the ad-management quotes will look dramatically cheaper than the full-service ones, and the gap is real but misleading. Japanese agency and comparison pages describe RPP management as typically billed at a percentage of ad spend, most commonly cited at around 20%, with some providers quoting 20–30%, a flat floor fee for very small budgets, and an initial setup or analysis fee often in the ¥30,000–100,000 range. Several also state a minimum monthly ad budget before they will take an account at all. These are second-hand figures collected from Japanese-language vendor and comparison sites, not a published rate card, so read them as market shape rather than as pricing you can hold anyone to.

Run the arithmetic and the appeal is obvious. On an illustrative ¥300,000 monthly RPP budget, a 20% management fee is ¥60,000 — against a full-service Rakuten retainer that our own Rakuten agency guide puts in a much higher band. It reads like the same job for a fraction of the money.

It is not the same job, and the reason has nothing to do with agency quality. It has to do with what the ad vendor is allowed to touch.

What Actually Decides Where Your RPP Ad Ranks

RPP is a click auction, and we have covered its mechanics — bidding, cost-per-click benchmarks, campaign types, the loop between paid placement and organic ranking — in our comparison of Japanese marketplace advertising. This section is not about the mechanics. It is about who holds the levers.

Japanese operator guides consistently describe RPP placement as determined by more than the bid: keyword relevance, the product page's click-through and conversion rate, and store review signals all enter the ranking alongside the bid amount. Rakuten does not publish the weighting, and none is asserted here. But the direction is not seriously disputed among practitioners, and it produces a specific consequence for a split engagement:

• The bid is the ad vendor's.
• The keywords derive from your product titles and copy — which the operations side writes.
• Click-through rate is a function of the thumbnail, title and price shown in the results grid — operations.
• Conversion rate is a function of the product page, its Japanese copy, shipping terms and stock status — operations. Our guide to Rakuten product page localization covers why that page behaves differently from a Western storefront page.
• Review count and store rating depend on review-reply cadence and customer-service quality — operations.

So the ad vendor is accountable for a metric whose inputs are, with one exception, held by a different company. On Amazon Japan this coupling is looser: ad performance and listing quality interact, but the advertising console is a separate system with its own published user-permission model. On Rakuten the advertising sits inside the same operating environment as everything else, and the levers that move it are mostly not advertising levers.

The Attribution Standoff Neither Vendor Can Lose

The practical failure mode is not a fight. It is a stalemate in which both parties are technically right.

The ad vendor reports that cost-per-acquisition rose because conversion rate on the landing pages fell, and asks for copy and image changes. The operations vendor reports that traffic quality declined, points at bid strategy, and puts the copy request in a queue behind the event calendar. Both statements can be true simultaneously. Neither is falsifiable with the data either party holds, because neither party can run the counterfactual — the ad vendor cannot rewrite the page to test its hypothesis, and the operations vendor cannot rebid the campaign to test its own.

This is not a hypothetical governance concern; it is the ordinary state of a two-vendor Rakuten account without a named arbiter. And it lands on you, because you are the only party with authority over both. Our guide on agencies versus freelancers arrived at the useful rule here from a different direction — split by decision, not by workload, and never let two roles share a responsibility. Rakuten advertising is a hard case for that rule, because the decision (what to bid) and the workload that determines whether the decision pays (what the page says) are not separable by contract. They are coupled by the platform.

Three Ways to Bill Ad Management, Three Different Incentives

Our guide to Japan ecommerce pricing models covers retainer-versus-commission structures generally. What it flags only in passing is the model specific to advertising — a percentage of the media budget, charged on top of the media budget — and that model deserves to be looked at directly, because it points somewhere different from the other two.

None of these is dishonest, and every agency in the market including ours works under one of them. The point is that an incentive you have not identified is one you cannot correct for. Our article on Rakuten's ECC versus a paid agency made this argument about the platform's own consultant, whose targets are Rakuten's, not yours. Percentage-of-ad-spend billing is the same misalignment relocated to the private sector — with the difference that this time you are the one paying for it.

Which produces the map worth keeping in view. On a split Rakuten account there are typically four parties with a stake in your advertising budget, and only one of them is measured on your gross margin:

• Your ECC, employed by Rakuten, who benefits when platform activity increases.
• Your ad vendor, paid on a share of what you spend.
• Your operations vendor, often paid on a share of what you sell.
• You, the only party whose interest is what remains after platform fees, points liability, affiliate commission and media spend.

This is an argument for reading recommendations with the billing model in mind. It is not an argument that any of these parties acts in bad faith.

The Event Month Is Where the Split Actually Breaks

Rakuten's volume concentrates into recurring events — Super SALE, the Okaimono Marathon, point-multiplier days. Two mechanics turn that calendar into the sharpest test of a split arrangement.

First, RPP budgeting is monthly, and a Japanese operator guide reports that Rakuten provides no native daily or weekly budget control; when the month's budget is consumed, delivery stops automatically. That source sells tooling built to work around the limitation, which is a reason to check it rather than to take it on trust. Second, a separate operator guide reports event-period click costs running roughly 1.5–2× normal levels, as every competitor bids up simultaneously. Each of these rests on a single second-hand Japanese-language source; verify both in your own RMS before planning around them.

Put those together and a Super SALE month has a predictable failure mode: the budget is consumed early in the event, delivery stops, and the highest-intent traffic of the quarter arrives with your ads switched off.

Now ask who was responsible. The ad vendor will say it flagged the pacing risk and requested a budget increase. The operations vendor will say the request arrived without a margin case during the busiest week of its year. Both may be accurate. Under a single full-service partner this is an internal escalation resolved in an afternoon; across two contracts it is a negotiation between parties with different fee bases, conducted while the event is running. It is worth noting that the discount eligibility rules that govern event participation add a second timing dependency on top of this one — we covered how a promotional calendar consumes your own price history in the one-agency-or-specialists guide.

If you split, this is the single item to settle in writing before signing anything: who may raise the monthly RPP budget mid-event, up to what ceiling, on whose approval, and within what response time.

Nobody Is Reporting the Number That Decides This

The reporting problem is quieter than the budget one and does more long-term damage. Each vendor reports competently on its own scope, and the sum of those reports does not answer the question you actually have.

That last row is the decision-relevant number, and no vendor is contracted to produce it. Rakuten's percentage costs are substantial and stack: our guide to selling on Rakuten sets out the system usage fee, the roughly 1% Rakuten Points liability, affiliate commissions and payment processing, all sitting on top of your media spend. A campaign can post a healthy ROAS in the ad vendor's report, a healthy conversion rate in the operations vendor's report, and still be margin-negative once those layers are applied — and in a split setup that fact has no owner and no reporting line.

Whichever model you choose, make the contribution-margin calculation someone's named deliverable. In a single-partner setup you can require it of the partner. In a split setup it almost always has to be yours.

Can You Even Grant Ad-Only Access to Rakuten?

A question worth asking before the commercial debate, because on Rakuten it may settle it: can you technically give a vendor advertising access and nothing else?

Rakuten staff accounts are issued through R-Login with a permission class attached to each user. The classes described in Japanese operator documentation are module-level — full RMS, R-Storefront only, R-Storefront plus R-Mail, everything except R-Backoffice, everything except R-Datatool, and R-Datatool only. None of them is an advertising-only scope. We were unable to verify this against a first-party Rakuten page, because the authoritative permission documentation sits behind the merchant login; the classes above come from Japanese operator guides, so check the current options in your own R-Login screen rather than treating this as settled.

If it holds for your account, the implication is straightforward. Amazon publishes a user-permission model for its advertising console in which users are invited at the manager-account or advertiser level and assigned standard roles — admin, editor, viewer — or custom, application-level permissions scoped to a single function, with a separate tab for reviewing and revoking third-party application access. Rakuten's permissions, on the evidence above, cannot be sliced that finely — your ad vendor will hold credentials reaching well beyond the ads. "Just advertising" then becomes a contractual boundary policed by trust and audit, not a technical boundary enforced by the platform. That does not make splitting wrong. It does mean the written scope — what this vendor may change without asking — has to be far more specific than it would need to be on Amazon.

The Switching Cost Runs the Opposite Way From Intuition

Most brands assume the risky thing to outsource is the money-spending function, and so they hold advertising close and hand out operations. On Rakuten the reversibility argument runs the other way.

As our guide to switching your Japan EC agency sets out, Rakuten's advertising configuration and reporting live inside RMS and stay with the store account. Replace the ad vendor and the bid history, keyword-level performance and campaign structure remain where they were. The handover is genuinely light.

Operations is the opposite. The Japanese product copy and the rights to it, the review-reply voice, customer-service history and macros, the event playbook, the working relationships with logistics and suppliers — the accumulated value of that work is not all inside the platform, and some of it leaves with the vendor unless your contract says otherwise.

Which inverts the usual conclusion: the advertising relationship is the one that is safe to keep short, benchmarked and competitively re-tendered, because switching costs are low. The operations relationship is the one worth depth, tenure and a carefully drafted exit clause. If you are going to run a split at all, this is the orientation that makes it survivable.

A Decision Rule for the Split

Splitting Rakuten advertising from Rakuten operations is defensible when all four of these are true. If any one fails, the split will consume more of your time than it saves in fees.

1. The storefront is already run competently, whether in-house or by a capable vendor. A specialist ad manager cannot compensate for weak Japanese copy, poor images or slow review replies — those are the inputs to the number it is being judged on.
2. Someone on your side can arbitrate, with authority to overrule either vendor on budget, pricing and page changes. Not a coordinator relaying messages — a decision-maker.
3. Copy and image change requests have a contractual response time. The ad vendor's requests are inputs to its own KPI. If they can be queued indefinitely, its accountability is fictional.
4. Ad spend is large enough that specialist skill outweighs the coordination cost. Below the point where a percentage fee buys meaningful senior attention, you are paying two vendors to disagree about a small budget.

Condition two is the one that fails most often, and it fails silently. Where you have decided you do not want to build that internal arbitration capacity, the argument for a single partner is not that one vendor is better at advertising — it is that the arbitration happens inside their organisation instead of landing on your calendar.

What LAUNOVA Does — and Where We Are Not Neutral

We should state the conflict plainly. LAUNOVA runs Rakuten operations for overseas brands, so an article concluding that ads and operations are hard to separate on Rakuten is an article that points toward what we sell. Treat conditions one and two above as the check on that: if your storefront is already well run and you have a real arbiter internally, splitting is a reasonable decision and we would say so.

What we actually run is the Japanese-language operating layer — storefront, content, customer service and the event calendar — with advertising held inside the same accountability line rather than across a contractual seam. We scope and price that to the modules you hand over rather than to a fixed package, so there is no rate card to quote here; the honest answer depends on what you keep. If you are earlier than that and want the storefront assessed before deciding anything about vendors, our Rakuten setup review is the narrower engagement.

What we do not do is take pricing authority you have not delegated in writing, guarantee an advertising outcome, or claim that consolidation removes concentration risk — it relocates it, and the mitigation is contractual, which is why the exit terms matter as much as the scope. We are also not a law firm; the promotional-display rules that shape the event calendar are a matter for Japanese counsel when a specific display is in question.

If you are weighing this now, the most useful things to send are your current monthly RPP spend, who writes your Japanese product copy today, and what happened the last time an ad change needed a page change. Get in touch and we will tell you where the seam is landing — including when the answer is that your current split is working and does not need us.

Related articles

Outsourced Rakuten Store Management

The full, partial and in-house models — and the modules that make up ongoing Rakuten operations.

Rakuten's ECC vs a Paid Agency

The same incentive question one level up: what Rakuten's own consultant is measured on, and what they will not do.

One Agency or Platform Specialists?

The same split decision drawn across platforms rather than across functions within one platform.

Sources

• • RPP ad management fee structure — percentage of ad spend commonly cited at around 20% (some providers 20–30%), a flat floor fee for small budgets, setup or initial analysis fees roughly ¥30,000–100,000, and stated minimum monthly ad budgets. Second-hand, aggregated from Japanese-language agency and comparison pages (pureflat.co.jp, finner.co.jp, grill.co.jp, ec-benriya.net); Rakuten publishes no agency rate card and neither does the agency market. Treat as market shape, not as a quotation.
• • RPP placement inputs — bid amount together with keyword relevance, product-page click-through and conversion rate, and store review signals. Second-hand, consistent across Japanese operator guides (nint.jp, jagoo.co.jp, pureflat.co.jp, bottleship-consulting.com). Rakuten does not publish the ranking weighting and none is asserted here.
• • RPP budget control — monthly budget only, no native daily or weekly budget setting, delivery stops automatically once the monthly budget is consumed. Second-hand, single source (empowerment-town.com, which sells tooling to work around this limitation — a source with a commercial interest in the gap existing, noted here rather than hidden). We did not verify it against a second independent source or a Rakuten-published page. Verify current behaviour in your own RMS.
• • Event-period click costs roughly 1.5–2× normal levels during Rakuten Super SALE and comparable events. Second-hand, single source (tatap.jp); no Rakuten-published figure exists and we did not corroborate the multiple independently.
• • R-Login / RMS staff permission classes — full RMS, R-Storefront only, R-Storefront plus R-Mail, excluding R-Backoffice, excluding R-Datatool, R-Datatool only; no advertising-only class among them. Second-hand (5springs.co.jp, allweb-consulting.co.jp). Not verified against a first-party Rakuten source — the authoritative permission documentation sits behind the merchant login and was not accessible for this article. Confirm the current classes in your own R-Login screen before relying on this.
• • Amazon Ads user-permission model — invitation at manager-account or advertiser level, standard admin / editor / viewer roles, custom application-level permissions scoped to a single function (billing, Stores builder, Posts), and a Third Party Applications tab for reviewing and revoking application access — Amazon Ads support and product documentation (advertising.amazon.com). First-party. Used here only as the contrast case; we did not verify how each control behaves on an amazon.co.jp account specifically.
• • Rakuten percentage cost stack (system usage fee, ~1% points liability, Rakuten Super Affiliate commission, payment processing) and full-service retainer ranges are carried over from our existing published guides rather than re-sourced here; see the linked articles for their underlying sources and caveats.
• • The ¥300,000 monthly budget used in the fee arithmetic is an illustrative figure chosen to make the calculation concrete. It is not a LAUNOVA recommendation, a client figure or an industry average.