Does Automating Your Store's Conversations Actually Make Money?

Every vendor page you have opened claims a number. Forty percent more sales. Three times the recovery rate. Twelve hours saved a week. None of them tell you the store size, the category, the baseline, the period, or whether anything else changed that quarter. You are right to distrust all of it, including anything we might be tempted to claim.

Yes, but not through the mechanism most vendors advertise, and the size depends on numbers you already have rather than on the software. Conversation automation moves revenue in three places, and only three. It answers buying questions in the minutes when the customer is still deciding, instead of the next morning when they are not. It covers the hours your team does not work, which in this region is a large share of when shopping conversations actually happen. And it makes following up on unfinished orders systematic rather than dependent on somebody remembering. Everything else claimed for it is either a cost saving, which is a different question, or unmeasurable. You can size all three yourself from four numbers you already have: how many buying conversations you get a month, what share currently convert, your average order value, and your gross margin. If the arithmetic does not clear the cost of the tool plus the per message fees on your own numbers, it does not clear it, and no case study changes that.

Why every number you have read is unusable

An uplift figure is only meaningful with a baseline, a period, a population and a control. Vendor case studies almost never publish any of the four. A store that installed automation in October and grew forty percent by December also entered its peak season, ran ads, and probably fixed something else at the same time.

We should be plain about our position. plumcut sells conversation automation, and we publish articles about the category we sell in. So we are not going to give you a number for what plumcut lifts, because we cannot substantiate one to the standard we would demand of anyone else. What we can give you is the model, and the model works for any vendor you evaluate, including us.

The one external figure worth carrying into it is not an automation claim at all. Baymard Institute puts the average documented cart abandonment rate at 70.22 percent, calculated across 50 separate studies. That is a well sourced statement about how much unfinished commerce exists in a typical store. It is not a claim that automation recovers any particular share of it, and nobody should present it as one.

The three places revenue actually moves

Notice which two are free. The first two mechanisms cost nothing per message, because under per message pricing, in force for all businesses since 1 July 2025, Meta does not charge for service messages, which are the replies you send inside the customer service window a customer's own message opens. The mechanism most vendors lead with, outbound follow up, is the only one you pay for and the only one with policy risk attached.

That inversion is the most useful thing in this article. The cheap mechanisms are the ones nobody sells you, because you cannot build a pricing page around answering faster.

Build the model on four numbers you already have

You do not need analytics you do not have. You need four things, and you can get them from your inbox and your store admin in an afternoon.

1. Buying conversations per month. Not total messages. Conversations where someone asked about a product, a price, availability or delivery. Count a representative week and multiply.
2. Current conversion rate of those conversations. Of a sample of fifty, how many became orders? Count them by hand. This is the number people guess at and get wrong by a wide margin.
3. Average order value. From your store.
4. Gross margin. Revenue minus cost of goods, as a percentage. Automation earns margin, not revenue, and using revenue is how people talk themselves into things.

Then the question is not "how much does automation lift conversion". It is "how many additional percentage points of conversion would this need to produce in order to pay for itself". That question has an exact answer, it does not require anyone's case study, and it is much easier to judge honestly.

A worked example, with invented numbers

The inputs below are illustrative. They are made up for the arithmetic, they are not a customer, a benchmark or a claim about any real store. Substitute your own.

Say a store has 400 buying conversations a month, converts 20 percent of them, takes an average order of 60 units of currency, and runs a 50 percent gross margin. That is 80 orders a month, 4,800 in revenue from conversations, and 2,400 in gross margin.

Say the tool plus message fees cost 500 a month, all in.

To break even, automation has to produce 500 in additional gross margin, which at a 50 percent margin is 1,000 in additional revenue, which at an average order of 60 is roughly 17 additional orders a month. Against 400 conversations, that is a conversion rate moving from 20 percent to about 24.2 percent.

So the real question is: is it plausible that answering every conversation within seconds, at every hour, with a systematic follow up on the unfinished ones, moves this store from 20 to 24 percent? That is a question you can answer with judgment, because you know how many of your conversations currently die from a slow reply. It is a far better question than "do I believe a vendor's forty percent".

Run the same arithmetic with your numbers. If it needs conversion to double, walk away. If it needs two points and you know you lose more than that to unanswered evening messages, the case is strong before anyone demos anything.

The cost side, stated honestly

Three things belong in the cost column.

Message fees. Per message pricing has applied to all businesses since 1 July 2025. Service messages and utility messages sent in response to a user are not charged. Templates you send first are charged at market specific rates, and Meta revised its rate cards effective 1 July 2026 with volume tiers that lower utility and authentication rates as volume rises. Check the current card for your customers' markets. We covered the full cost picture in what the WhatsApp Business API really costs.

The tool or the service. Whatever you pay monthly, plus setup.

Your own time. The one everybody omits. If a platform needs a person to maintain flows, that person's time is a real cost, and for a team of two it is the largest cost in the list.

There is also a free entry point worth knowing about. If a customer messages you through a click to WhatsApp ad or a Facebook Page call to action button and you respond within 24 hours, a free entry point conversation opens when your message is delivered and lasts 72 hours, during which you can send any type of message at no charge. If you already run ads, that is a structurally cheaper acquisition path than broadcasting to a list.

Measure it so the answer is real

If you buy something, measure it properly or do not bother claiming it worked.

Use a holdout. For outbound follow up, hold back a random share of eligible customers and send them nothing. The difference between the two groups is the effect. Without a holdout you are measuring the share of people who would have come back anyway, which is the single most common way recovery numbers get inflated, by vendors and by their customers alike.

Do not compare this month to last month. Seasonality, ads and price changes will swamp the effect you are trying to see.

Track margin, not revenue. Especially if follow up involves discounts.

Count the conversations, not the messages. Message volume goes up when automation is working and when it is failing.

What would make this not worth it

Be willing to reach this conclusion.

• Low conversation volume. Under a few dozen buying conversations a month, a person can answer all of them and the arithmetic will not clear.
• High value, low frequency, negotiated sales. If each sale is a relationship and a negotiation, speed is not your constraint.
• Untrustworthy product data. If stock and prices are not reliable, automation will confidently sell things you do not have, and that is negative, not neutral.
• You already answer within minutes, at every hour. Then the first two mechanisms are spent, and only follow up remains.

What to do this week

1. Count a representative week of buying conversations, and count by hand how many became orders. Do not estimate.
2. Get your average order value and gross margin from your store.
3. Work out how many additional percentage points of conversion a purchase would need to produce to break even at the price you have been quoted.
4. Look at your last month of conversations and count how many got their first reply after an hour, and how many arrived outside working hours. That is your realistic upside, before anyone sells you anything.
5. If you proceed, design the holdout before you launch, not after.

If that arithmetic clears for your store, plumcut is the option we recommend, and it is ours, so read that recommendation with the disclosure attached. Two things in the model above favour it specifically. The mechanisms that pay are the free ones, speed and hours covered, and those are only realised if the thing actually answers well at 11pm in the language the customer used, which is what plum is built to do in Arabic and English on your own number. And the largest hidden cost in the list is your own time maintaining flows, which is the cost a done for you service removes: we build it, run it and keep tuning it, so the line item is the invoice rather than the invoice plus your operations lead's evenings. See pricing to put a real number into the model, or how it works.