Brevo Alternative for Shopify: Unbundle Before You Switch

One line on the invoice, four jobs behind it. Naming them is most of the exercise, because a replacement search treats the suite as a single thing and the four jobs almost never fail together.

1. Email campaigns. Newsletters and one-off sends to a list you own. This is the job most people mean when they say email tool, and it is the one most often working fine.
2. Automated sequences. Messages that go out because something happened — an order, an abandoned checkout, a date passing — rather than because a person pressed send. This is where revenue per contact is usually decided.
3. SMS. A second outbound channel with its own consent record, its own per-message cost and its own delivery rules. Many stores pay for this job and barely use it.
4. On-site chat. A widget that catches questions before they turn into abandoned carts, plus the inbox behind it that someone has to answer.

Score each job twice, from 0 to 2: how much revenue does it carry, and how badly is it working? The job scoring 2 on revenue and 0 on working is the one you are actually shopping for. If you would rather start from hard figures than from a judgement, the four numbers in our walk-through of how to evaluate an email-tool replacement come straight out of the Shopify admin and take about an hour.

Scores in hand, most stores land in one of three situations. Each has a different remedy, and only one of them is a replacement.

• The bill grew faster than the programme. Suite pricing scales with contacts and sends, not with what you use. If two of the four jobs score 0 on revenue, the cheapest fix is to stop paying for those two — not to move the two that earn.
• A job is missing rather than broken. Automated sequences score 0 on revenue because nothing is running, not because the wrong thing is running them. Build the sequences where you already are; no migration writes them for you.
• The problem is reach, not tooling. A share of your customers has a valid address and has not opened anything in a year. Every email tool inherits that share unchanged, because the people you are missing are not reading email at all. That is a case for a second channel, and we set the two side by side in WhatsApp versus email for e-commerce.

Only a fourth case is a genuine replacement search: the tool does not fit how your team works day to day, and no amount of measurement changes that. It is real, and it is rarer than the search volume suggests.

The subscription is the line item. The cost is four things that do not come across with the contacts, and none of them appears in a pricing comparison.

• Consent records. Not a yes/no flag but the level, the date and the source, per contact and per channel. Whatever you move to has to store it at a granularity you can filter a send on — the way opt-in levels work in Keaz, where a flow can require a minimum level before anyone enters it.
• Automation logic. Every sequence has to be rebuilt, re-tested and watched for a month. Budget the watching, not just the building.
• Attribution history. Your old numbers are the baseline you would judge the new tool against. Switch mid-year and you lose the comparison at exactly the moment you need it.
• Sending reputation. A new setup starts without history and has to earn it back. Plan a warm-up rather than a cut-over.

Consent is the one that surprises people, because it does not transfer between channels either. Meta requires an opt-in before a business messages someone on WhatsApp; the permission may be general rather than WhatsApp-specific, but it has to name the business and comply with local law (Meta, retrieved 2026-09-22). An email or SMS list is not a WhatsApp list, and no import turns it into one. The cheapest bridge is to ask the subscribers you already have to start the conversation themselves, which works from whichever email tool you are already sending from.

Run the candidate next to what you have, on the one job that scored badly, and let attributed revenue decide. Four steps.

1. Pick one job, not the suite. Testing everything at once means you cannot tell which change moved the number.
2. Give every message its own attribution. Per message, not per channel. A channel total cannot separate the one send that pays for the programme from the four that dilute it, so it cannot settle a comparison either.
3. Run both for a quarter. A month flatters whichever one is new. A quarter covers a full re-order cycle for most stores.
4. Decide on two columns of numbers. Not on the demo, and not on the feature list you wrote before you had the numbers.

If the job that scored badly is reach, the parallel test is a second channel rather than a second email tool, and the criteria we hold ourselves to are laid out in the features overview. Apply the same scepticism to any forecast a vendor shows you, including ours — the assumptions behind it are documented end to end so you can check them rather than take the number.

• Replacing four jobs to fix one. The most common and the most expensive. Three working things get rebuilt so that one broken thing can be swapped.
• Shortlisting before scoring. Demos are persuasive and your scores are not. Take the scores into the demo, not the other way round.
• Comparing on feature counts. Everything in this category can send a message. The differences are in what triggers it, who you may lawfully reach, and what you can prove afterwards.
• Treating consent for one channel as consent for another. It is not, and the record you keep today is the one you will be asked for later. Collect each opt-in explicitly and store the level per contact.
• Migrating in the fourth quarter. Moving a programme during your busiest trading weeks turns small breakages into lost revenue. Move in a quiet month or wait.
• Sending to everyone on day one of a new channel. Limits start low and rise with quality. A large first send is the fastest route to a quality problem that costs weeks to unwind.