Email pre-testing
Send the winning campaign,
not the experiment.
Pre-test every version of a campaign on a panel of simulated consumers, built from the purchase and online behavior data of real people.
95% agreement with real campaign outcomes. How we measure that
Testing on your list
Pre-testing on Foresight
The problem
Your list is your test bench. That is expensive.
Most sends ship untested
Live A/B testing needs volume. Lifecycle flows, small segments, and mid-size campaigns never reach significance, so the copy ships on instinct.
Every test burns the list
Half your test audience gets the losing email. Fatigue, unsubscribes, and spam complaints are the quiet price of learning on your own subscribers.
You get a which, never a why
The dashboard says variant B won. It cannot say what to write next week, because clicks do not explain themselves.
How it works
Three steps, twenty minutes
Upload your variants
As many versions as you want. Subject lines, body copy, offers, full templates. Choose the audience they are meant for.
A thousand simulated consumers respond
Each profile receives one variant, opens or ignores it, reads as far as it reads, clicks or does not, and explains itself in its own words.
Get the ranking and the reasons
Simulated open and click rates per variant and per segment, and the diagnosis behind them: what carried the winner, where the others lost the reader.
Who answers your test
Every simulated consumer is built from the purchase and online behavior data of real people, then calibrated on your own send results, so it answers the way your actual subscribers would.
Example: Dana, 42
Owns her home, two kids, suburban North Carolina. Shops online weekly, opens about a third of retail emails, clicks few.
Nothing changes in your stack. Pre-testing happens before the send, so your ESP, your templates, your segments, and your calendar stay exactly as they are. The only difference is that the version you send is already the winner.
Sample report
What you actually receive
An excerpt from a Foresight test report: five framings of one promotional send, identical offer and button, tested on 1,000 profiles across two audiences.
"One order, everything handled"
"Save 25%, same everything"
"The one everyone is ordering"
"Ends Sunday: 25% off"
"The one thing most people forget"
The best copy on open rate loses on click rate
Curiosity takes the best open rate of the five and collapses at the click: the subject makes a promise the body never pays off, and 11% of its cohort calls the email clickbait. An open-rate dashboard would crown the wrong version. The reasons catch it.
“I opened it to find out the answer. Felt tricked when it was just an ad.”
Woman, 31, on the curiosity email
“One order and it is handled. That is the entire pitch for me.”
Mother of three, 38, on the winning email
Demo study on a fictional retailer’s send, shown in the exact format of a client report. In client work, rates are calibrated to your channel baseline first. Full sample report available on request.
Proof
Tested with the world’s largest retailers
Does the panel reach the same conclusions as real consumers? We test that head-to-head against real campaigns, and we publish the results.
The same study, run twice
A European grocery retail leader ran one study on Foresight and, in parallel, in the field with real consumers. We answered without seeing their results.
21 / 24
questions where both studies reached the same conclusion
0
questions where the two studies contradicted each other
3
questions that differed in amplitude, none in direction
How we measure accuracy
Accuracy is the share of calls where a Foresight study and a matched real-world study reach the same conclusion (which variant wins the opens and the clicks for each segment), compared one by one. Across validation studies, that agreement averages 95 percent.
Calibration on your audience
Calibrated to your audience before anything runs
Calibration first
Before your first test, the simulation is aligned to your historic data, and you review the calibration before anything else runs.
For your technical team
Responses are available per profile, with stable profile identities across test cycles, as files or over an API. Teams that run their own optimization algorithms use the panel as the audience and the reward signal, cycle after cycle.
Questions
Asked by every team, answered upfront
They are simulations, built from the purchase and online behavior data of real people, and they answer the way the people behind the data would. Your subscribers only ever see the winner.
Two mechanisms. Validation: we test the panel head to head against real field studies and publish the agreement, misses included. Calibration: before your first test, the panel is aligned with the actual open and click rates of your list, and you review it first.
No. Your stack stays exactly as it is. Pre-testing happens before the send, so you keep your ESP, your templates, your segments, and your calendar; the only thing that changes is that the version you send has already won.
Yes. The same panel runs subject lines, landing pages, offers, concepts, claims, packaging, and pricing. Email is one door into it, usually the one with the most sends per year behind it.
The send used to be the experiment. Now it is the conclusion.
Your next campaign can go out already tested. Every send on the calendar, pre-tested in a few minutes.
Start pre-testing your sends