THE AMAZON
AGENCY BRIEF

Same data. Different decision.

Why an agency’s most useful information may sit outside the Amazon account.

A price falls. Should advertising pause, or should spending increase?

Adam Stamper, founder of The Automation Engine, offers two possible responses. A client protecting profitability may want advertising paused below an agreed price floor. A client prioritising promotional sell-through may want advertising activated to support the lower price.

The signal has not changed. The objective has.

That distinction explains why the conversation about agency relevance needs to reach beyond access to performance data. An account-level recommendation cannot, on its own, resolve the brand’s competing demands on cash, stock and margin.

Ed Sweet, co-founder and managing director of Unicorn Orange, describes the work of building systems that continuously capture and update context. The useful outcome is an explanation the team can investigate: whether a sales decline involves availability, competitor pricing, creative or a combination. An apparent pattern still needs scrutiny before anyone treats it as a cause.

James Northcott, CEO and founder of Toucan, applies the same discipline to campaign measurement. Strong attributed performance does not settle whether spending created additional demand. The agency needs to question the result against the brand’s economics and the evidence available.

Matt Anderson, former CEO of Marketplace Amp, challenges agencies to look further still. A product with persistently worsening advertising economics may have a sourcing, packaging, margin or range problem. More campaign history is useful only if someone uses it to confront the underlying issue.

The agency opportunity is access to relevant knowledge Amazon does not have, combined with the judgement and authority to make that knowledge count. As clients connect more systems, the boundary of what a platform knows will move. The agency has to keep earning its role.

There is also a limit to the argument. A straightforward execution assignment may need only a narrow set of inputs. Collecting more information is worthwhile when it changes a decision, rather than simply making the agency’s records more extensive.

The difficult work is knowing which context matters and noticing when it has changed.

Better tactics. Wrong strategy.

The most valuable intervention may happen before the work is commissioned.

A client asks for improved advertising performance. The team gets to work.

But what if the product cannot support the spend? Or if the visible problem is only a symptom of something happening elsewhere?

Ant Finch, founder of Vendor Rocket, has seen the uncomfortable version of successful delivery: an agency executes the brief well while leaving the business worse off. Pricing, availability, range and vendor terms can create problems advertising is poorly placed to solve.

Sabya Mukherjee, Amazon Lead at Cobalt Consulting, supplies a practical example. A brand faced falling Amazon prices as third-party sellers undercut its position. A tighter distribution approach was one possible response, but it carried costs and potential consequences for the much larger business outside Amazon.

His team investigated a different question: how could the sellers afford those prices?

Sabya reports that test purchases uncovered parcels sent through standard consumer delivery services without the hazard markings or declarations required by the carriers concerned. The investigation pointed towards avoided compliance costs.

The team gathered evidence and pursued the compliance concerns. Sabya reports that sellers were subsequently removed, prices stabilised and margins recovered. The basis for the intervention was evidenced safety and compliance concerns; a low price alone would not have justified it.

The initial problem had suggested one kind of assignment. The investigation produced another.

For owners, this raises a question about how work enters the agency. If proposals fix the delivery plan before anyone has tested the diagnosis, challenging the brief becomes an exception rather than part of the service.

That does not make every execution brief inadequate. A client may have completed the diagnosis internally and want a specialist to deliver it. The distinction is whether the agency understands the basis of the brief and recognises the evidence that would justify reopening it.

In Sabya’s example, the value came from changing the investigation. In another assignment, the useful judgement might be to proceed with the work already agreed.

Read the full weekly feature.

AVNs: clarity before complexity

Faster preparation is useful. An agreed negotiating position is essential.

Annual Vendor Negotiations are a sharp test of commercial advice.

An additional funding request may look modest as a percentage. Across the affected sales base, it can create a substantial commitment, particularly if the growth expected in return fails to arrive.

Matt Redfearn, a former Amazonian with 15 years’ experience negotiating with suppliers and preparing teams and vendor managers for negotiations, warns against letting the prospect of more sophisticated automation lead to unnecessarily elaborate preparation. Teams need to understand the outcome they want, the trade-offs they can accept and the decisions that follow if agreement is not possible.

A complex model offers limited help if the negotiator cannot explain its assumptions or recognise when circumstances have changed.

Ant Finch, founder of Vendor Rocket, drawing on 16 years of negotiating with Amazon from brand and advisory roles, brings the discussion back to product economics. A blended account margin can hide products unable to support another concession. Commercial, finance and leadership need to agree what they will accept before the live discussion creates pressure to decide.

Preparation also needs memory. An exception can become an expectation in a later negotiation. Recording what was offered, on which conditions, in exchange for what and with whose approval helps the next team assess whether the precedent remains relevant.

Brands can organise this work internally, with external support or through a combination. What matters is that the analysis, negotiating position and authority to make concessions are connected. An adviser with incomplete economics faces a constraint; so does an internal team whose finance and commercial leaders have not agreed the objective.

The relevant test is whether the people at the table understand the position well enough to hold it, explain it and adapt it.

Your agency. Your software?

Build around a reason, then budget for the responsibility.

The ability to produce an internal tool makes replacing a subscription look tempting. A prototype shows what is possible. It rarely shows the full cost of operating the alternative.

James Northcott, CEO and founder of Toucan, describes the agency’s motivation as embedding its methodologies, processes and judgement into technology. Specialist tools remain part of the approach. The purpose of building is to support how the agency makes decisions.

Rich Morris, co-founder of Push-Pull, describes a deliberately narrow build. The agency owns one layer: client-authorised Amazon data in a single warehouse alongside its forecasts, listing content and tasks. It buys project management, language models and specialist tools, and chose not to build a bid-optimisation black box or anything that duplicates Amazon’s own reporting.

One practical benefit: the advertising budget an account manager sees is derived from the same forecast the automated spending rules spend to. Every automated change is dry-run first and its receipt read back before the client is told it is live.

The harder part, he says, was trust rather than plumbing. Client-facing numbers have to match Amazon for the settled window, so more effort goes on caveats, freshness and approvals than on charts.

Andrew Hamada, CEO and co-founder of Reason Automation, adds the infrastructure perspective. Reason Automation initially wanted to build insight and management tools, but obtaining reliable Vendor Central data became a more immediate job. Its pipelines became the primary product for years.

The inputs need as much attention as the interface: detail, history, refresh frequency, storage, security and access.

Scott Needham highlights where buying retains an advantage. Competitor and market intelligence can involve data that costs money to acquire. Recreating a screen does not recreate its underlying information, and every tool still needs maintenance.

For owners assessing a build, there is another distinction worth making: time released is not automatically money saved. If automation creates capacity, the business needs a plan for using it. Serving additional accounts, avoiding a hire and improving client advice are different benefits with different evidence.

There is no necessary prestige in either choice. Bought software can contain expertise an agency would struggle to reproduce. A targeted internal build can solve a problem no commercial product addresses well. Either can become expensive when poorly adopted.

The stronger comparison includes the work required after launch and the evidence that the proposed capability matters to the client or the agency’s economics.

The next question is who benefits when delivery becomes cheaper.

Do clients receive lower fees? Do agencies retain the saving, add capacity or offer work that was previously uneconomic? Does the time released produce better advice, or simply more activity?

Those outcomes need evidence. Over the coming editions, I would like to hear from owners and clients who can describe what has actually changed: the price, the scope, the staffing or the result.

There is a leadership question alongside it. As teams and tools take on more execution, the founder’s role may change. That could create room for better decisions, or leave more exceptions arriving at the same desk.

For now, the question I am left with is:

When your agency becomes more efficient, who gets the benefit?

Simon

Published by ELEVATE. Edited by Simon Ellicott.

If this edition was forwarded to you, subscribe to the monthly email here. The weekly LinkedIn briefing is a separate subscription.

About this briefing

The Amazon Agency Brief is published by ELEVATE and edited by Simon Ellicott. The opening editorial and analysis express the editor’s views; contributed perspectives are identified separately. Contributors’ affiliations are included so readers can assess their interests, including where they build or sell the technology discussed. Participation does not constitute an endorsement of a product or service.

This edition draws mainly on agency, software-provider and adviser perspectives. It does not establish what brand clients generally want or are willing to pay. We welcome evidence and contrasting experience from brand teams, including cases where specialist execution is more useful than a broader advisory relationship.

If you spot an error or have evidence that challenges an argument here, contact the editor.