The Impact Report · Issue 2 · November 2026
What moves budget, and what doesn't
A monthly note on what The Impact Lab is learning about measurement for brands that convert off Amazon.
What moves budget, and what doesn't
We delivered an AMC analysis that was SPOT ON.....but it changed nothing for the brand. It covered tens of millions in streaming and display spend across 18 channels. It was accurate. But it went nowhere. The study answered the question the media team asked, not the one the budget owner had.
2027 budgets are being set now. For a non‑endemic brand, Amazon stays on the plan if someone can show what it did in the budget owner's terms. We've run AMC readouts for a lot of advertisers, and the ones that moved money had five things in common:
- 01
The budget owner named the number before anyone ran a query.
Example: An insurer agreed to move part of its linear TV budget into a streaming test, on one condition: four numbers every day and every week. How many people the campaign reached, how often each one saw it, and impressions and conversions by local market. Those are the measures its TV buyers already used, so the test could be judged against TV on TV’s terms. The numbers arrived on schedule, and the test went ahead.
- 02
The outcome was their own sales data, not Amazon’s attributed conversions.
Example: A card issuer couldn’t see its card sales next to its Amazon advertising, because a card application doesn’t fire a website pixel. Before it would sign the next year’s plan, it asked for one thing: its own transaction records, matched to who saw the ads and compared against its own control group. The lift showed up in the issuer’s own sales data, not in Amazon’s reporting. The issuer kept investing and put more of its own analysts on the work.
- 03
It led with what a customer was worth, then what one cost.
Example: Our first sportsbook readout opened with cost per first deposit, the number the acquisition team is measured on. The rebuilt version opened with what happened next. Over the following months, the customers in the audience we tested bet more often and put more of it into the sportsbook’s most profitable products than similar customers outside it. We set the very largest players aside so they couldn’t skew the average. Cost per first deposit came second, as the price of a customer whose value was already on the page.
- 04
It was written for the person who owns the budget, and the analyst signed off too.
Example: A consumer brand’s analytics team ran the work for months without involving the brand managers. When results arrived, nobody with budget had asked for them, and nothing changed.
- 05
It reported what changed in the budget, and the misses alongside the wins.
Example: One quarterly review laid out every audience test the same way, win or lose: the goal, what was tried, the result and the next step. A lookalike test with zero conversions sat in the same table as tests that beat their benchmarks, and its next step was to narrow the starting audience. A few months later, the narrowed version brought in new to brand customers at about twice the benchmark rate. Because the miss was in the table, the win was easier to believe.
1P Activator is our agentic platform for Amazon Marketing Cloud
An agent that knows AMC and knows your business
Amazon Marketing Cloud (AMC) holds the most detailed view of how your Amazon ads perform. Your DSP reports tell you how each campaign performed. AMC goes further. It shows how your ads worked together: which mix of streaming TV, display and sponsored ads reached your customers. It shows how often people saw your ads before they acted, and where more ads stopped helping. It shows which audiences responded, and the paths people took before they bought or signed up. Few non‑endemic brands use AMC. Getting answers takes AMC’s own SQL, plus the Amazon Ads knowledge to read the results. That knowledge has a steep learning curve.
1P Activator makes AMC easy to use. It converts your data into decisions. Here’s how:
Anyone on your team can get answers. Type a question in plain business terms. 1P Activator writes the AMC query, runs it and explains the result. You get an answer in minutes. Along with a visual chart of the data and recommendation(s) you can act on. No SQL. No data interpretation and aggregation. No waiting for an analyst to do the work for you.
1P Activator knows your business. Every deployment starts with your goals, KPIs and economics: what a customer is worth and what one should cost. We build in your business rules too. For a lender, that means knowing when it changed its rates, so a drop in applications reads as a pricing move, not a bad campaign. 1P Activator also learns from every question your team asks, so it gets sharper about your business over time.
1P Activator knows AMC. It’s built on AMC rules and logic The Impact Lab has spent the past year developing. It also draws on our founder’s decade at Amazon building AMC and AWS solutions for customers. That context matters. In our testing, a general AI model pointed at AMC gave answers that looked right but weren’t. It used the wrong sales definition. It recommended scaling a campaign with six purchases. It gave a confident number for a question AMC can’t answer. 1P Activator uses the right tables, definitions and dates. It checks every query before it runs. You get the answer an experienced AMC analyst would give, not the one that only looks right. You can defend the numbers to your leadership team.
1P Activator brings your customers into AMC. On its own, Amazon’s data shows who saw your ads. If your customers sign up on your own site, app or store, Amazon can’t see who became a customer. Upload your customer list in the app. 1P Activator hashes the personal details inside your own AWS account, then loads the list into AMC. No engineers. No APIs. No code. Now your answers cover who became a customer, not just who saw an ad.
Built to be trusted.
- Runs in your own AWS account. Your data stays with you. We can set it up and manage it for you.
- Nothing you share trains the model or informs another brand’s answers. Everything stays local to you.
- Validated by AWS on its Software Path. We completed a successful Foundational Technical Review of how it’s built, secured and run.
See what your team could answer from AMC in your own data. Start with a free 30 day pilot.
Highlights by Vertical
What we're working on this month. Brands are not named, and some examples combine more than one conversation. Where the work is still in progress, we say what we expect, not what we've proven.
Automotive
One view of reach and frequency across national, regional and dealer ads
The situation. An automaker runs national campaigns on Amazon. Its regional dealer associations run their own, and so do hundreds of individual dealers through their own agencies. In between sits the automaker’s program management office, the firm that enrolls dealers in co op programs, approves their vendors and agencies, and tracks what each dealer sells. Each is measured on its own. A household can see the national spot, the regional offer and the dealer ad, and nobody knows the total. One automaker put the question plainly: are we reaching someone 10 times, or 100? With budgets tighter going into 2027, the first ask is to reach more people with the same money.
What we’re building with automakers:
- 01One view in AMC, with everyone’s agreement. National, regional and dealer media brought together in the automaker’s AMC instance. The program office is the natural route to dealer sign off, since it already manages dealer enrollment and approved vendors.
- 02Protect what each party won’t share. Dealer agencies don’t share their rates or fees, and they shouldn’t have to. Results come back aggregated, with media costs removed, so nobody can work backward to another party’s pricing.
- 03Start with reach and frequency across all three. How many households the combined spend reached, how often, and where the ads overlap. This is the easiest win and needs no sales data.
- 04Then connect to vehicle sales. Using the registration data many automakers already license, loaded into AMC as their own data.
Why it matters. We would expect the same budget to reach more households once overlap is visible. For the automaker, it’s the first time its whole Amazon footprint can be seen in one place. The hard part is getting the approvals, not the technology.
Financial Services
A lender that counts applications but not funded loans
The situation. A digital lender new to AMC buys Prime Video and display. It can see applications started on its site. It can’t see which of those became funded loans, and a funded loan is the number its finance team plans around.
What we’re scoping:
- 01Answer one question on Amazon’s data alone. Which audiences and formats reached people who went on to apply. No data sharing needed to start.
- 02Upload the funded customer list. A file of funded customers, hashed and loaded into AMC, so the read moves from applications to loans.
- 03Read it out in the lender’s terms. Cost per funded customer from Amazon, next to what a funded customer is worth and what other channels cost.
Why it matters. Applications flatter every channel. Funded loans are what the lender’s budget is set on, and that is the number that would keep Amazon on the 2027 plan.
Sports Betting
Which product brings a customer in, and what they become
The situation. A sportsbook measures Amazon on first time deposits, like every other channel. But a depositor’s value depends on what happens next: which products they move to, how long they stay, and what they’re worth over time. A channel that brings in fewer depositors who stay longer can be the better buy.
What we look at:
- 01The entry product. Which product each customer started with: sports, casino or prediction markets.
- 02Movement across products. Whether customers who saw the Amazon ads moved from one product to another more often than similar customers who didn’t. Without a holdout test this is a pattern, not proof, and we say so.
- 03Value over the first year. What each group was worth over its first year, with a handful of very high value customers measured separately so they don’t skew the average. AMC looks back about a year, so this is first year value, not lifetime.
- 04Cost against value. What a customer from Amazon cost to acquire, set against what that customer became.
Why it matters. Two channels can bring in the same number of depositors at the same cost and be worth very different amounts. The one that brings in customers who stay and grow keeps its budget.
Agencies and Consultancies
More brands sending their outcomes to Amazon
The situation. Large consulting firms are onboarding brands to Amazon’s Conversions API, a direct feed of a brand’s sign ups or purchases to Amazon. Independent agencies want AMC answers for their clients without hiring AMC analysts.
Where we fit:
- 01A specialist bench for Conversions API work. We work alongside large firms on the Amazon specific parts of onboarding: which events to send, how to match them, and how to read the results in AMC.
- 02One 1P Activator per client, run by the agency. Each client gets its own stack in its own AWS account, managed by the agency, so client data never mixes.
Why it matters. Every brand that sends its real outcomes to Amazon can be measured on customers, not clicks. That is good for the brand and good for the Amazon team covering it.
What we're reading: Amazon and Non‑Endemic news
Non‑endemic news from the past few weeks that's useful in a client conversation.
Digiday · Nov 17
CFOs take a bigger role in judging marketing spend as budgets stay flat ↗With marketing budgets flat, a Gartner analyst says CFOs are now a critical piece of new business, and Deloitte Digital is pitching its marketing operating system straight to CEOs.
For your non‑endemic accounts: When the CFO signs off, the readout has to be in the CFO’s terms: customers and what they are worth, not impressions.
PPC Land · Nov 5
Off Amazon purchase audiences come to DVA+ ↗Attain says audiences built from opted in purchase data on more than 14 million US consumers are now live in DVA+, and it names QSR, travel, financial services and automotive brands.
For your non‑endemic accounts: A new targeting option built for the categories that don’t sell on Amazon. Worth raising with QSR and auto accounts.
Digiday · Nov 5
Financial services jumps to 13% of ChatGPT ad spend ↗Per Sensor Tower, financial services rose from 2% of US ChatGPT ad spend in April to 13% in September. A working paper finds paid ChatGPT visits convert at about a third of Google’s rate.
For your non‑endemic accounts: Financial brands are testing new channels before they can prove them. The one that shows customers, not clicks, has the stronger case in 2027 planning.
MediaAlpha · Nov 10
Home insurers raised ad spend 84% this year ↗Insurers on MediaAlpha’s marketplace lifted home insurance ad spending 84% through September and now advertise in about twice as many states each month as in 2023.
For your non‑endemic accounts: Insurers are spending again. A conversation about bound policies, not quotes, is the way into that budget.
PPC Land · Nov 12
Amazon puts prediction markets under its gambling ad rules ↗Amazon’s revised Gambling and Games policy requires verification for every real money gambling advertiser, prediction markets included, with rules for 22 countries. In the UK, gambling ads are limited to 9pm to 6am.
For your non‑endemic accounts: Sportsbooks and prediction markets now pass the same gate and face tighter placement rules. With fewer impressions to buy, proving which ones brought in a first time depositor matters more.
PPC Land · Nov 16
Amazon Ad Tag now sets a 30 day first party cookie by default ↗The Amazon Ad Tag on advertiser websites in North America now links a visitor’s page views, sign ups and purchases for 30 days, subject to consent. Separately, AMC’s new to brand columns began including modeled conversions on Nov 2.
For your non‑endemic accounts: Brands that don’t sell on Amazon measure it through events on their own sites, so their conversion counts can change with no action on their part. Compare reports before and after mid November with that in mind.
Digiday · Nov 3
Buyers move 10% to 30% of Google search budgets elsewhere ↗With AI Overviews cutting clicks and costs rising, some advertisers are shifting search money to other channels.
For your non‑endemic accounts: Lead generation budgets in finance and insurance are in motion. That is an opening for Amazon, if the brand can measure applications and policies.
When to call us
If you cover a non‑endemic brand that would spend more on Amazon but can't connect that spend to customers, bring us in. We start with a 30 day pilot of 1P Activator, at no cost, and a readout written for the person who sets the budget.
Next month: how to read AMC results when the outcome takes weeks to land.