⚡ THE ESSENTIALS IN 20 SECONDS
- In-house: your team runs the campaigns. Maximum control and brand knowledge, but slow to launch and hard to scale.
- Managed services: a specialized external team runs your account with proven technology and processes. Fast to launch and scalable, but more expensive, and learning stays outside unless you demand it.
- Hybrid: your team sets strategy and goals; the external team executes and optimizes. In my experience, it best balances results and control.
- Below is my 1-to-5 star rating across 10 criteria.
IN THIS ARTICLE
I have spent years running retail media accounts with Pacvue, and I have worked directly with Pacvue’s managed services model on US brands investing more than US$3 million a year. I have also seen brands that do everything in-house and brands that combine both. This comparison comes from that experience: here is my honest rating of each model and when each one fits.
Managed services in retail media: a model in which an external team, from an agency or a technology provider, runs a brand’s campaigns day to day: strategy, launch, optimization, reporting and testing, usually backed by an automation platform.
The three models
| Model | Who does what | Best for |
|---|---|---|
| In-house only | Your team handles strategy, operations and reporting | Brands with specialized talent and one or two platforms |
| Managed services | An external team runs everything with its technology and processes | Brands that need scale or fast results |
| Hybrid | Your team sets goals and strategy; the external team executes and optimizes | Brands that want control without carrying all the operations |
There are lighter options too, such as paying for an audit or hourly advisory to review your account without handing over operations.
The rated comparison
My 1-to-5 star rating, based on what I have seen running accounts. It is an experience-based opinion, not a rule: your case may differ.
| Criterion | In-house only | Managed services | Hybrid |
|---|---|---|---|
| Monthly cost (lower is better) | ★★★★★ 4/5 | ★★★★★ 2/5 | ★★★★★ 3/5 |
| Control over the account | ★★★★★ 5/5 | ★★★★★ 3/5 | ★★★★★ 4/5 |
| Speed to launch | ★★★★★ 2/5 | ★★★★★ 5/5 | ★★★★★ 4/5 |
| Expertise and best practices | ★★★★★ 2/5 | ★★★★★ 5/5 | ★★★★★ 4/5 |
| Use of automation and tools | ★★★★★ 2/5 | ★★★★★ 5/5 | ★★★★★ 4/5 |
| Deep brand knowledge | ★★★★★ 5/5 | ★★★★★ 3/5 | ★★★★★ 4/5 |
| Test-and-learn discipline | ★★★★★ 2/5 | ★★★★★ 4/5 | ★★★★★ 4/5 |
| Measurement and incrementality | ★★★★★ 2/5 | ★★★★★ 4/5 | ★★★★★ 4/5 |
| Multi-retailer scale (Amazon, Walmart, Instacart…) | ★★★★★ 2/5 | ★★★★★ 5/5 | ★★★★★ 4/5 |
| Learning that stays in your company | ★★★★★ 5/5 | ★★★★★ 2/5 | ★★★★★ 4/5 |
I have worked with this model on US brands investing more than US$3M a year in retail media, and honestly the results have been very good: organized launches, far more testing discipline and a level of automation that an in-house team rarely reaches on its own. That said, it works best when the brand stays involved in goals and strategy.
Where in-house wins
- Brand knowledge: nobody knows your product, margins and priorities like your own team.
- Control: immediate decisions, no intermediaries.
- Owned learning: everything learned stays in the company.
Its weak spot: daily operations consume the team, leaving little time to test, measure incrementality and scale to other retailers. Experience with advanced tools is also often missing.
Where managed services win
- Speed: they arrive with processes, templates and a proven launch plan.
- Technology: they make full use of bid, budget and dayparting automation, something few brands exploit alone.
- Accumulated experience: they have seen dozens of accounts and know what works.
- Scale: they run Amazon, Walmart, Instacart and other networks with the same method.
Its weak spot: it costs more, it can take time to understand your business and, unless you ask, learning stays on the provider’s side.
Automation: big swings or fine-tuning
One of the biggest differences I have seen between teams is how they make bid decisions. Many in-house accounts make big changes every so often: “let’s raise everything 20%” or “cut everything because ACOS spiked”. It works short term, but it punishes good and bad keywords alike.
With Pacvue Managed Services the logic is different: many small corrections, by keyword, by product and by hour of the day, backed by rules reviewed against data. The result is a more stable account, where what works is protected and what does not is fixed without hurting the rest.
Whoever runs your account, ask them to explain their rules in plain language. More context in Amazon Ads pacing.
The first 90 days: what to demand
| Period | What you should see | Red flag |
|---|---|---|
| Days 1–30 | Account diagnosis, business goals agreed in writing, first quick wins (negatives, budget, inventory) | Big changes without a diagnosis |
| Days 31–60 | Account reorganized by goal, clear budgets, reports in business language | Reports full of metrics with no conclusions |
| Days 61–90 | Fine-tuned automation, first incrementality test and next-quarter plan | No test proposed |
In my experience, a good start shows in order rather than miracles: by month three you know exactly where every dollar goes and why.
Organize the account by goal, not by habit
Something I apply in any model is splitting investment by what each campaign must achieve. That way every dollar has a measurable purpose:
| Block | What it is for | How I measure it |
|---|---|---|
| Brand defense | Nobody takes your own searches | Cost per sale and presence on your brand |
| Generic growth | Win the category’s searches | Total sales and share |
| Conquest | Win shoppers from other brands | New customers and cost per new customer |
| Discovery | Find new searches and products | New keywords that graduate to fixed campaigns |
Mixing these goals in the same campaigns is the most common mistake I find in audits. More in auto vs. manual campaigns and what is iROAS.
What to ask a managed services provider
- What experience does the team running my account have, not just the company?
- How many accounts does each manager handle?
- What does the first 90-day plan look like?
- Which automation rules do you use and how do you explain them?
- How do you measure incrementality?
- Are the data and campaigns mine if we end the relationship?
- How will you transfer learning to my team?
A typical case (fictional)
A home care brand ran everything in-house with one person. Campaigns worked, but there was no time for Walmart or for tests. It moved to a hybrid model: its team set quarterly goals and the provider executed with automation and a test plan. Within six months it entered Walmart in an orderly way, cut purposeless campaigns and its team learned to read incrementality.
It is not about who operates, but about someone having the time, method and technology to do it well.
When to choose each one
| Your situation | Recommended model |
|---|---|
| One platform, small budget, someone experienced | In-house |
| Several platforms, aggressive goals, small team | Managed services |
| A team with judgment but no time to operate | Hybrid |
| You want to validate before deciding | Audit or one-off consulting |
Agency secrets about in-house vs. managed services
What I have learned from both sides:
Campaigns, history and reports must stay with you if you leave.
A good external team proposes tests from the first quarter.
Total sales, share and incrementality, not just ROAS.
The best model combines your brand knowledge with the discipline and technology of someone who operates every day.
I’m Alonso Jiménez, a Retail Media account director. I manage campaigns on Amazon, Walmart, Target, Instacart and Mercado Libre every day, for brands in the US, Mexico and Latin America.
Frequently asked questions
What are retail media managed services?
An external team that runs your retail media campaigns day to day with its own technology and processes.
Is it better to run Amazon Ads in-house or externally?
It depends on your team, platforms and goals; many brands get the best results with a hybrid model.
How long do managed services take to show results?
A good onboarding takes 6 to 12 weeks across audit, new structure and optimization.
What should I require from an external provider?
Transparency on rules, a test plan, incrementality measurement and ownership of your data.