Laurence vs m19

m19 thinks about bidding a lot like we do. Its models forecast how likely each click is to turn into a sale and what that sale is worth, it builds its own campaigns and finds its own keywords, and it asks you for a goal rather than a list of bids. If you’re weighing the two of us, you’re already asking the right questions.

The differences come down to three things. First, who runs the account: m19 is software whose algorithms work toward the target your team sets for each group of products, while Laurence runs your Amazon ads for you. Second, how often bids move: m19 recalculates them once a day, and we re-price every search term’s bid every hour. Third, how each click is priced: we work out what a click is worth from your margin and its CVR, for each placement separately, and price how sure we are about each search term’s CVR into its bid.

m19, in a few paragraphs

m19 is a Paris company, founded in 2018, with a team in Japan as well. It’s an Amazon Ads Advanced Partner, describes its product as “Algorithms for Amazon Ads”, and says it runs ads in 19 Amazon marketplaces and has managed €500 million of ad spend. It sells two plans: Autopilot, a separate, simpler tool for small sellers, and Professional, for brands and agencies.

Its engine is built on a clear idea. As its co-founder and CTO Pascal Pons explains, the model estimates, for each keyword and product, the chance of a sale after a click and what that sale is worth, learning from your last 90 days. It’s retrained every day, weighting recent results more, and each client’s data is kept separate. You group products into strategies and give each one a mode: an ACOS target, which m19 treats as the main constraint while it recalculates the bid on every keyword and product target each day, a TACOS target, or a monthly budget. m19 builds its own campaign structure, with few keywords in each campaign so it can read results at the keyword and product level, and says to expect about four weeks to full performance.

Around the bidding sit tools for the people running the account: TACOS Management, a Top of Search Rankings Optimizer, a boost for promo days, hourly stats with dayparting you schedule, share of voice, Amazon Marketing Cloud reporting without writing SQL, and Amazon DSP with audiences built for you. It publishes its pricing, starting at $59 a month with a 30-day free trial and no annual contract, which makes it one of the lowest-cost places in the category to start.

Laurence, in a few paragraphs

We’re building autonomous advertising across the entire internet, starting with Amazon Ads for e-commerce brands. We’re a full-stack AI company: we aren’t a tool your team has to operate, and we aren’t an agency selling hours. Our models run every account, and our deployments team sets your strategy with you, reviews what the models did, and is who you talk to.

The idea underneath is that advertising is a math problem. Every click has an expected profit you can work out: how likely it is to convert, what the order is worth, what you keep after costs, and what the click cost. Once you can price a single click, the right bid, placement modifier and budget follow. Our methodology explains it in more detail.

Where we agree

We share a lot with m19. We both think models, not people adjusting bids by hand, should make the bidding calls. We both estimate how likely a click is to turn into a sale and what that sale is worth, rather than reading it off the last few clicks. We both build our own campaigns rather than taking over yours. And we both treat your target as a limit, not a number to spend up to.

So this isn’t a comparison between a modern approach and an old one. It’s a comparison between two modern approaches that make different choices about who does the work, how often a bid moves, and what goes into a click’s value.

Software your team sets up, or work that gets done for you

m19 takes the bidding, the keyword finding and the campaign building off your plate. What stays with your team is the setup and the judgement: grouping products into strategies, choosing each one’s mode and target, adding seed keywords and a blacklist, scheduling the hours ads pause, and deciding when a target should move. Its help center suggests starting an ACOS target at your last 30 days’ ACOS and waiting at least three days between changes. m19 says Professional is built for agencies and sellers who already understand Amazon Ads, and on that plan a Customer Success Manager runs your onboarding call and reviews your account after 2, 10 and 30 days.

With us, the running of the account is the product. You tell us the goal, whether that’s profit, growth, or a CAC, TACOS or ROAS number, and the limits you want. The models then make the calls a person would otherwise make: which searches to go after, what to bid on each one every hour, how to price each placement, and how to set budgets. Our team checks the work and is there when you want to talk it through.

How a click is priced

Here’s how we value a click:

m19 and Laurence agree on much of this. Its model forecasts the chance of a sale and the value of that sale for each keyword and product, much as we estimate CVR and order value. The difference is where the target comes from. With m19, you choose an ACOS or TACOS target for each strategy. We work out what each click can afford from your margin, which is what you keep after the cost of the product, Amazon’s fees and returns. A target ACOS lands on the right bid only when it happens to match what that product can afford, and the right number is different for almost every product:

There’s one thing we don’t build into the bid yet: the repeat orders a new customer goes on to place. We measure how often each product’s customers come back, and bringing that into the bid, along with what a better organic rank is worth, is what we’re building next. Until then, you can raise a target by hand for a product people buy again.

Bids, placements and budgets

m19 recalculates every bid once a day, and on Autopilot it adjusts bids once a day too. Budgets move daily as well: its algorithm resets each campaign’s daily budget every day from sales and costs, and can raise one that ran out to as much as three times what it was. It shows hourly stats, and you can schedule the hours when a strategy’s ads pause.

Every hour we re-price the bid on every search term and check every budget. On the median account we run, clicks at the top of search convert at about twice the rate of clicks lower on the page. So we work out the CVR for each placement, set the base bid for the placement that converts worst, and raise the others with modifiers, reset every night. We also check budgets every hour: a campaign close to its daily budget that is still clearing your ROAS bar gets more for the rest of the day, so profitable ads don’t stop in the afternoon.

Search terms with only a few clicks

Most of what an Amazon account pays for barely has any data. On the median account we run, 95% of the search terms that get a click get fewer than ten clicks a month. A term with four clicks and no orders reads as a 0% CVR, but all it really tells you is that you don’t know yet.

m19 handles this with its modes. Its ACOS target mode asks for 40 to 50 ad sales across a strategy’s products in the last 30 days. Below that, it recommends its Monthly Budget mode, which gets the most it can from a budget you set, and for an account with under $200 of spend in 60 days, a constant bid you choose. Keywords in its exact campaigns that haven’t sold in two months go back to its auto campaigns. We borrow evidence from similar searches in your category, treat each term’s CVR as a range rather than a single forecast, and price that uncertainty into the bid, so a term we know little about is bid more carefully than one we know well, even when their averages match.

TACOS and organic rank

Ad sales help a product climb the organic results, and organic sales cost nothing per click. m19 goes further here today. Its TACOS Management lets you set one TACOS target across ad types, for each product or portfolio, and pulls spend back when ads start replacing sales you’d have made anyway. Its Top of Search Rankings Optimizer, which you switch on for up to 50 keywords in a strategy, lowers the top of search bid where your product already sits at the top of the organic results.

We track your organic rank on the terms we bid on every day and judge each product on TACOS, so the organic sales your ads help earn are counted. Pricing what each rank is worth into the bid itself is what we’re building next. If protecting your organic rank on specific terms is your main goal, that’s a real strength of m19’s today.

Channels, features and price

m19 works on Amazon, as we do, and covers the same ad types: Sponsored Products, Sponsored Brands and Sponsored Display, plus Amazon DSP, in 19 marketplaces. Its pricing is published. Autopilot is $59 a month (€49) for sellers spending under $5,000 a month on ads, Sponsored Products only. Professional is $479 a month (€400) plus 3% of managed ad spend, and agencies and larger companies get custom pricing. Every plan starts with a 30-day free trial without a card, there’s no annual contract, and you can cancel in one click. For a smaller seller who wants to try software before committing, that’s an easy, low-risk way to start.

We run Amazon Sponsored Products, Sponsored Brands and Sponsored Display, and Amazon DSP on our own seat. Meta and Google are in early beta, and we’re holding them to the same standard as Amazon before we call them ready. We don’t publish a price: we start by asking you to connect your Amazon Ads and Seller Central or Vendor Central accounts with read access, then build a pilot proposal from your own data, showing where your ad money goes today, the targets we’d set for each product line, and the bids our models would place. Pricing is agreed with the pilot.

Which one is right for you

Switching from m19 to Laurence

You don’t have to switch everything at once. We can run one marketplace or product family while m19 keeps the rest, so you can compare results on your own account before deciding anything.

1. 01
   Connect with read access.We read your Amazon Ads and Seller Central or Vendor Central data to build your pilot proposal. Nothing in your account changes at this point.
2. 02
   Agree on the pilot.You pick the products and the goals. Pause m19 on those products so two systems aren’t bidding on the same terms, and check your plan’s billing date so the handover lines up.
3. 03
   We build our own campaigns.Our campaigns carry a Laurence prefix, so results stay easy to tell apart. Your history stays in your account.
4. 04
   If you stop.If you stop, we pause and archive our campaigns and leave yours as they were.

Common questions

Is Laurence an alternative to m19?

For Amazon ads, yes. Both forecast how likely each click is to turn into a sale and build their own campaigns. m19 is software whose algorithms re-price bids once a day toward the target your team sets for each strategy. Laurence runs your Amazon ads for you, prices each click from your margin and its CVR, and re-prices every search term’s bid every hour.

How often does m19 change bids?

Once a day. m19 recalculates the bid on every keyword and product target each day, from a model retrained daily on your last 90 days. Laurence re-prices the bid on every search term every hour, resets placement modifiers every night and checks budgets every hour.

Does Laurence publish its price like m19?

No. We start by asking you to connect your Amazon Ads and Seller Central or Vendor Central accounts with read access, then build a pilot proposal from your own data, and pricing is agreed with the pilot. m19 publishes its prices, and every plan starts with a 30-day free trial.

Do I need someone on my team to run Laurence?

No. You set the goal and the limits, and Laurence does the day-to-day work. You can see every decision and the reason for it on the Laurence platform, and our team is there when you have questions.

How we wrote this. We described m19’s products, pricing and features from m19.com and its help center as of November 11, 2026, and things may have changed since. Laurence figures are medians across the 15 store profiles we run, covering all of their Sponsored Products spend, October 6 to November 4, 2026. The charts use made-up numbers to show the idea. If we’ve got anything about m19 wrong or out of date, tell us and we’ll fix it.