Objective-first marketing

Turn your growth goals into audiences built from real purchases.

Geminus helps CPG brands decide whom to reach using purchase behavior across retailers. We build audiences around goals such as acquiring new buyers or bringing back lapsed customers, then measure the campaign’s effect on purchases.

Start a conversation

The problem

A brand’s growth goal spans every retailer. The data behind its audience usually covers only part of that.

Brands set goals in business terms: new households, lapsed buyers coming back, more frequent purchases, a move to a higher-priced product. A campaign then needs an audience someone can activate, such as a retailer’s definition of a new customer, a CRM list or a platform interest segment.

Each of those sources gives an incomplete view of what a consumer buys. A retailer’s customer history records purchases at that retailer, so a shopper who looks new there may buy the brand somewhere else. A platform’s interest signals describe activity on that platform, and they can’t show what someone bought in a store.

The audience and the reported results can then drift away from the goal that justified the budget. Geminus starts with the goal and uses purchase behavior across retailers to decide whom to reach and what to measure.

How it works

Four decisions take a campaign from a growth goal to a read on purchases.

Brand X is a made-up CPG brand, and its numbers are illustrative. They show how the steps connect and do not describe a client result.

  1. 01
    Choose the goal
    You decide what should change and over what period. Geminus writes that down as a specific behavior, a time window and the budget in scope, and you can edit any of it.
    Brand XBrand X wants more households buying it for the first time over the next 12 weeks.
  2. 02
    Agree on the audience and success measure
    Geminus studies households that made the target purchase in the past, looks at what they bought beforehand and ranks the wider population on how likely each household is to do the same. You review the definition, the audience size and the success measure before anything runs, and you can accept, reject or change them.
    Brand XThe target purchase is a first Brand X purchase. Current buyers are excluded. Eligible households are randomly split into a treatment group and a holdout group, and success means more first-time buyers in treatment than in holdout.
  3. 03
    Activate the campaign
    Your team runs the campaign through the media platforms it uses now. Geminus provides the audience and does not buy media. You choose the destinations and the timing.
    Brand XThe agency activates the treatment group through The Trade Desk and keeps the holdout group out of the campaign.
  4. 04
    Review the results
    Geminus compares purchases between the two groups and reports the estimated effect, the uncertainty around it and what the comparison can’t tell you. Your team decides what to do next.
    Brand XIn this example, 3.2% of treatment households buy Brand X for the first time, compared with 2.5% of holdout households. The team reads the 0.7 point difference with its confidence interval before deciding whether to expand.

See the technical architecture

Two separate questions

Finding likely buyers and measuring the effect of advertising take different methods.

Who is likely to buy?
A predictive model answers this. It learns which behavior came before the target purchase and ranks households on how likely they are to make it, which helps select the audience. A high-scoring household may buy without seeing an ad, so the score says nothing about what advertising caused.
Did the advertising change purchases?
A treatment and control comparison answers this. Households are assigned to each group before launch, so the difference in purchases between the groups estimates the incremental effect of the campaign.

What is available today

Some of this runs in production now, and some of it is still being built.

Running today
Audiences built from purchase histories, checked for identity quality and available for activation through The Trade Desk.
Being put into operation
Goal-specific scoring, richer purchase features from multiple data sources, and treatment and holdout designs in partner programs.
In development
Formal goal definitions, connections to campaign reporting, and purchase-outcome reads that feed budget recommendations.

People make the decisions at each stage. Your team approves the goal, the audience, the success measure and the activation, and it decides what to do with the results.

Why Geminus

The audience starts from your goal
You don’t need a starting segment. Geminus looks at households that reached the goal in the past and studies their behavior before they got there.
A separate model for each strategy
Winning new buyers and bringing back lapsed buyers depend on different behavior, so each strategy gets its own target purchase, lookback period and model.
Purchase behavior across retailers
Models learn from what people bought at many retailers. That shows behavior one retailer’s data can’t, such as a shopper buying a competing brand at another chain.
A control group set before launch
Holdout households are assigned before the campaign starts. An analysis run afterward on who happened to see the ads can’t recover that comparison.
Your media teams keep their tools
Campaigns run through the DSPs and other platforms your team uses now. Geminus provides the audience and reads the results.
Results measured in purchases
Results are read in the same purchase data used to build the audience, so the campaign is judged against the goal it started with. The data comes from sampled panels, so each result carries its coverage and uncertainty.

Who it’s for

Find the right consumers for your business goal. Measure the effect on purchases.

CPG brands
Whether you want to attract new buyers, bring back lapsed customers, or increase purchase frequency, Geminus builds audiences from purchase behavior across retailers and measures campaign impact against your goal.
Agencies
Geminus helps turn a client’s business goal into audiences your team can activate through its existing media platforms. Purchase data across retailers provides a common basis for planning the campaign and evaluating its results.
Retail media networks
A shopper who is new to a brand at one retailer may already buy it elsewhere. Geminus uses purchase behavior across retailers to help distinguish customer acquisition from a change in where someone shops.
Demand-side platforms
Geminus builds audiences for a specific business goal and makes them available for activation through DSPs. Media teams manage campaigns in their existing tools, while Geminus evaluates purchase outcomes across retailers.
Identity and data onboarding platforms
Geminus defines the audience using purchase behavior and the campaign’s business goal. Identity and onboarding platforms match the audience to supported advertising destinations so media teams can reach those consumers.

From the blog

Latest thinking

View all posts
July 2026 · Geminus
The audience you're buying is not the audience you need
When a brand buys an audience, what they're actually buying is usually a stand-in for the audience they need. Both sides know it, at some level. The transaction happens anyway, because there hasn't been an alternative.
Read the post
July 2026 · Geminus
A model that's only seen your own transaction data can't connect a gym membership to a sports drink
A person connects a gym membership, deodorant, and a sports drink instantly. A model trained only on your own transaction history can't, and that gap is the whole argument for using a language model.
Read the post
July 2026 · Geminus
Search ads systems don’t understand what they’re doing
Ask a person to judge whether a search result is good and they do it instantly. Ranking systems can't, because they don't know what things are. They know what users clicked before.
Read the post

Tell us about a growth goal you’re working on.

We’ll go through it with you and assess whether the available purchase data can support building the audience and measuring the result. The team behind Geminus has spent twenty years working in advertising execution.

Discuss a growth goal

Or write directly: