Skip to content
Site
Book a meetingSign in

Iris predicts which campaigns deserve your money, and which don't.

Iris scores how ready each visitor to your site is to buy. Then it gives every Google Ads and Meta campaign one decision a day.

Iris doesn't make the changes unless you tell it to.

Book a meetingHow it works
An example plan
Raise by 14 423 krGeneric - BarbecuesStrong intent, room to groweROAS 4.8×
Stop−83 668 krMeta - Broad videoVery low intenteROAS 1.6×
Lower by 33 704 krGeneric - Garden furnitureOver your price limiteROAS 2.7×
Leave as isShopping - BestsellersAlready shows on almost every searcheROAS 4.3×
eROAS (expected return on ad spend): what a campaign brings back for each krona it costs.
The problem

Your ad platforms optimize for the wrong signal.

Clicks are not intent

Google and Meta see clicks, page views and the odd conversion. They cannot see who was about to buy.

Add-to-cart is not intent either

Visitors who add to cart and visitors who are ready to buy overlap by only 20 to 30%.

Existing customers look like buyers

Logged-in customers, support visits and people checking a bill look busy to the platforms. They are cost, not opportunity.

So the budget follows engagement, not intent.

The plan

One decision per campaign, every day.

Raise

Buyers keep coming, but the budget runs out before the day ends. Raise it.

Leave as is

Nothing has moved enough to act on. Leave it as is.

Lower

Intent is weak or falling, or each order costs too much. Spend less.

Stop

These visitors rarely buy. Stop paying for them.

When there is too little data, Iris says nothing instead of guessing.

The product

Three parts that feed each other.

Each round buys better traffic than the last: better data in, better traffic out.

  1. Iris Intent

    Scores every visit

    A model trained on your own site scores every visit for how ready the visitor is to buy.

  2. Iris Signal

    Tells the platforms who buys

    Sends your high intent visits to Google and Meta as a conversion they can bid on. Low intent and existing customers stay out.

  3. Iris Scale

    Moves the budget

    Reads your ad data and the intent data every night. Each campaign gets one decision, with the amount and the reason.

See the product
Who it is for

One account, three people who read it differently.

  • The plan in Iris: one row per campaign, with what to do, the reason and how sure Iris is.
    Performance marketers

    The plan

    What to raise, lower or stop today, campaign by campaign, and why.

  • Site health in Iris: the site's week against normal, and the pages where more visitors leave.
    Whoever owns the website

    Site health

    Your site's week against normal, and the pages where more visitors leave than before.

  • The CMO dashboard in Iris: spend, orders, visitor insights and how past advice held up.
    The CMO

    The CMO dashboard

    What the account spends, what it bought, and whether past advice held up.

No ad platform tells you to spend less on it.

Clicks and cost do not show whether a visitor was going to buy. Iris looks at what visitors do on your own site, so it can tell you to cut.

22×

the normal cost per order, in the two weeks after Iris said stop

4×

the normal cost per order, where Iris said lower for weak intent or a high cost

Measured on one live account, July to August 2026, in weeks Iris had not seen when it gave the advice.

See all results
Getting started

Value first. Then you decide.

You see what Iris is worth on your own numbers before you commit to anything.

  1. Add one tag and give access

    One JavaScript tag on your site, plus read access to Google Analytics, Google Ads and Meta. It takes days.

  2. We measure your potential

    We train the model on your site and show you, in a workshop, what Iris would change and what it is worth.

  3. Prove it with an AB test

    Run Iris against your current setup for 4 to 6 weeks in your own accounts. Then say yes or no.

Book a meetingWhat happens next
Industries

Built for companies that sell, sign or book online.

  • Energy

    Your site is a service portal with a shop attached. Most paid clicks are customers after a bill or a meter reading. The few who sign are the ones worth paying for.

  • Telecom

    Top-ups, bills and upgrade checks fill the site, and every phone launch adds a crowd of window-shoppers. New connections are a small, valuable minority.

  • Insurance

    Shoppers collect quotes from several insurers and buy one policy. Optimize to quotes and you buy quote collectors, at some of the highest click prices in search.

  • E-commerce

    Google and Meta both claim the order, and the basket doubles as a wish list. The campaigns that find real buyers run out of budget before the evening.

  • Booking services

    A campaign per location, a handful of bookings each, and most visitors only want your opening hours. Some who are ready to book find no free slot.

  • Travel and hospitality

    Weeks of research, minutes to book, and the last click takes the credit. Meanwhile booked customers come back to check in and manage their trip.

  • B2B lead generation

    Five forms a month, and two are a student and a supplier. Too little for the platforms to learn from, and too much of your sales team's time.

Customization

Set up by people, run by AI, checked by code.

Our customer success team turns your goals into instructions and limits. An AI council reads them every night. Code checks every number before it reaches you.

How we set it up
Rule settings in Iris: the AI instructions and the limits for one client.
Demo account with made-up numbers.

See Iris on your own numbers.

Book a meeting. We show you Iris on a demo account. Then we measure your potential and prove the value in your own accounts, before you commit.

Book a meeting

Already using Iris? Sign in