23 Amazon programs repatriated from incumbent agencies in the last 18 months. Average pre-handover account health score: 47 out of 100. Average score at day 60 post-handover: 78. Average revenue change in the same window: +14%. Average ad spend change: -9%.
The repatriation playbook is not a secret. The execution discipline is. Here is the 60-day version that works.
Why agency repatriations exist in the first place
The most common version of this story is not that the incumbent agency is bad. It is that the incumbent agency is operating on the 2021 Amazon playbook in a 2026 Amazon market, and the brand finally noticed that its TACoS is creeping, its organic share is shrinking, and its agency’s quarterly business review still leads with “we hit your campaign-level ACoS targets.”
The 2021 playbook is keyword-list-driven, ACoS-targeted at the campaign level, and structurally over-weighted to manual campaigns and exact-match keywords. It worked when CPCs were lower, when competitor density was lower, and when Amazon’s auction was less reliant on first-party signals. In 2026, Amazon’s auction is signal-saturated; bid optimization has to happen at the keyword and placement level using Amazon Marketing Cloud data, and TACoS has to be managed at the brand level rather than the campaign level. Agencies that have not retooled around AMC and brand-level TACoS optimization are systematically underperforming the brands they manage. That is most agencies, frankly.
The other major repatriation trigger: ownership misalignment. The incumbent agency owns the Seller Central credentials, the campaign structure, the keyword research, the creative library, and, most damaging, the brand’s Amazon institutional memory. When the brand wants to switch agencies or in-house the program, it discovers that 18 months of work is locked inside someone else’s documentation. That alone is a six-figure switching cost in some cases.
The 60-day playbook is built around solving both of those problems on a tight clock without breaking the program in the meantime.
Days 1-15: Lock the access, audit the state, do nothing operationally
The temptation in week one is to start fixing things. Resist it for 15 days. The single most common failure mode in agency transitions is the new team making changes before they understand the current state, then attributing performance shifts to the wrong cause, then losing the trust of the brand-side stakeholder.
The first 15 days are about three things only.
First, lock the access. Get owner-level Seller Central access. Get DSP access if the brand runs DSP. Get the AMC instance. Get the brand registry credentials. Get every ad console seat the incumbent agency has and audit who else has access. We have walked into accounts where former employees of three different vendors still had API access. Clean that up before doing anything else.
Second, audit the state. Pull every active campaign, every keyword, every negative keyword list, every product targeting set, every placement modifier, every dayparting schedule, every creative asset, every A+ Content version, every brand store page. Document all of it. Most of it is not going to be touched in the first 60 days, but you need the baseline to measure against.
Third, do nothing operationally. No bid changes, no campaign restructures, no creative refreshes. The point is to establish a clean before-state. The incumbent agency is, in nearly every case, still legally engaged for 30-60 days post-notice and may still be making changes in this window. Document everything they do. It will matter.
Days 16-30: The first three changes, and only three
Day 16 is when surgical changes start. Three of them, no more.
Change one: kill the broad-match auto campaigns that are running at sub-2x ROAS with no harvesting workflow. These are the single highest-leverage cut in nearly every account we walk into. Brands with a mature ad program almost always have 8-15% of total spend running through auto campaigns that were set up in 2022, never re-audited, and are producing low-relevance traffic. Cutting them produces an immediate ad-spend reduction without measurable revenue impact. We’ve measured the cut on 23 transitions; revenue impact at day 30 averages -1.2%, ad spend impact averages -8.4%, net contribution margin lift averages +220 bps.
Change two: implement brand-level TACoS targeting in place of campaign-level ACoS targeting. This is the structural change that fixes the 2021-playbook problem. The bid optimizer, whether it is Amazon’s, an automated tool, or a human analyst, needs to be optimizing for total ad sales as a percentage of total brand sales, not for individual campaign ACoS hitting an arbitrary target. The mechanical change is moving from per-campaign ACoS goals to per-ASIN TACoS goals with a brand-level rollup constraint.
Change three: re-baseline negative keywords. Run a 90-day search term report. Identify every search term with $50+ in spend and zero conversions, and add them as exact-negative across the relevant campaigns. This is the cleanup the incumbent agency should have been doing weekly. Doing it once on day 20 produces a 5-8% TACoS reduction by day 60 in nearly every account.
Days 31-45: Catalog, content, and the things the agency was supposed to own
By day 30, the ad side is stabilized. Day 31-45 is for the operational surface area the incumbent agency was probably neglecting: catalog and content.
The audit usually surfaces five to twelve things. Listings with suppressed images. A+ Content modules that haven’t been updated since 2023. Brand store pages with broken category links. Variation families with the wrong parent. Backend search terms left at default. Title structures that don’t lead with the high-volume noun phrase. Bullets written for branding rather than ranking.
Each of these is a small fix individually. Together, on a mature account, they are typically worth 6-12% of trailing twelve-month revenue, achievable in 60-90 days from when they are fixed. The surgery in this window is sequenced for low risk: do the backend changes first (search terms, alt text, image renames) because they don’t disturb live conversion. Then do the bullets and titles in a controlled rollout. Then the A+ Content. Then the brand store. We do not refresh hero images in the first 60 days unless there is a clear revenue-blocking issue, because hero image changes have the highest CTR variance and we want to be operating from a stable conversion baseline by the time we touch them.
Days 46-60: AMC, attribution, and the institutional memory rebuild
The last 15 days are about building the analytical and operational infrastructure that the incumbent agency probably never built, the thing that makes the program defensible and self-improving from this point forward.
This means standing up Amazon Marketing Cloud queries that surface brand-level TACoS by audience cohort, placement-level performance attribution, and the new-to-brand customer rate. It means establishing a weekly operating rhythm with the brand-side stakeholder that includes both leading metrics (search rank, share of voice, glance views) and lagging metrics (revenue, contribution margin, TACoS). It means creating the documentation that is the brand’s institutional memory: every active campaign and why it exists, every keyword strategy and what it’s testing, every catalog change and what it produced.
The discipline at this stage is to build the kind of operating system that, if the brand chose to in-house the program in 12 months, the in-house team could pick up and run on day one. That is the standard. An agency relationship that creates more brand-side dependency over time is, structurally, the kind of misaligned agency the brand just left.
If the repatriation is happening alongside other channel changes, adding Walmart, adding Shopify, transitioning fulfillment, the sequencing has to compose. Our FBA-to-FBM-plus-3PL 90-day plan assumes a stable agency relationship as a precondition; running them in parallel doubles execution risk and rarely works.
The day-60 scoreboard
Average results across the 23 transitions, day 0 to day 60: revenue +14%, ad spend -9%, TACoS down 410 bps, contribution margin up 380 bps, account health score 47 → 78. Two of the 23 transitions did worse than baseline at day 60 (-2% to -5% revenue), both had compounding catalog issues that took 90+ days to resolve and recovered to flat or positive by day 120.
The discipline is the playbook. The 60-day version above is not novel. Most of these moves are obvious to any operator who has run an Amazon program. What is rare is the sequence: audit before changing, lock access early, three operational changes in the first 30 days, no hero creative changes in the first 60, AMC and operating rhythm in the back end. Sequence is the strategy.
Repatriating an Amazon program from a misaligned agency in 2026? Book a 30-minute scoping call. We’ll map the 60 days for your specific account, no SOW required to start.