Instagram ads have been running for three months, the budget keeps growing, and there is no more money in the till. The media buyer shows a report with cost per lead, the sales team complains about "empty" leads, and by the end of the month the owner is left with one feeling: the ads are not paying off. Sometimes that is true. More often the money leaks in two or three specific places, and you can find them in an evening if you put ad account numbers and CRM numbers into one table.
First, check that the ads really do not pay off
Ad payback is measured with ROI: (revenue from ads − spend) / spend × 100%. You spent $600 and got $2,400 in revenue: ROI is 300%.
Then comes an important correction, without which the formula lies. Revenue is not profit. With a 25% margin, $2,400 of revenue leaves $600, exactly what went into the ads. So at that margin a 300% ROI on revenue means zero. The break-even threshold on revenue is (1 / margin − 1) × 100%: 300% for a 25% margin, only 100% for 50%, and 900% for 10%.
The second correction is about time. A dental clinic or a gym often earns nothing on a client's first visit; the money comes from repeat visits. Measure ad payback over a single month and such a business is almost always in the red. A fairer approach is to count revenue per client over at least six months, even roughly.
If ROI is still below the threshold after both corrections, look for where the money leaks.
Ads that spend and sell nothing
The most common and simplest place. Average campaign ROI is made of ads that work great and ads that have not produced a single deal. The latter are invisible in the average.
Take the ad-level report for 30–60 days, sort by spend and look at the top ten rows. An ad that ate a noticeable share of the budget and brought zero deals goes first, even if it has the best cost per lead in the account. Cheap leads without sales usually come from broad audiences and loud creatives.
One caveat: an ad with no leads still has to be counted somewhere. If the pipeline report is built from leads only, the spend of such ads simply drops out, and ROI per pipeline looks better than it is. We tripped over this ourselves while building the report in shamCRM, so now each ad's pipeline is picked down a chain: where most of its leads went, then the pipeline of the page or Instagram account the ad runs from, then the ad account's pipeline.
Leads nobody answered in time
This place deserves a closer look, because it is completely invisible in an ad report and eats the most money.
Instagram ads work in the moment. A person scrolls the feed at 11:40 pm, sees a sofa and writes "how much?" in Direct. If the answer comes at 10 am, they have already written to three more shops, and one of them replied at night. For the ads, this lead is counted and paid for, it looks great in CPL, and it is missing from sales. The media buyer has nothing to do with it, and a new audience will not fix it.
The check is simple: open the last 30 leads from ads and see how long it took to answer them. If the median is over an hour, fix the replies first and touch the ads later. The options are known: a rep on duty, round-robin assignment so leads do not sit with someone who went to lunch, or an AI agent that answers routine questions at night and hands the client to a person in the morning.
Meta instant forms have a similar story. A lead lands in the Meta account, someone exports it to a spreadsheet once a day, and the call happens the next day. In shamCRM a lead from an instant form immediately becomes a CRM lead in the right status and goes to a sales rep in rotation, with no exports.
Leads but no sales
It also happens the other way round: leads arrive quickly, get answered on time, yet conversion to a deal is low across all ads at once. Then the ads are probably not the problem. Look at the script, the price and what the person sees after their first message. If conversion is equally low on expensive and cheap ads, the question is in sales.
Returning customers instead of new ones
Retargeting and broad ads to followers often collect leads from people who buy from you anyway. Those sales would have happened regardless, and the ads take credit for them. Separate new leads from returning customers at least to know whether the customer base is growing. In the shamCRM report these are two different columns, and cost per lead uses new leads only.
An ad report template: which columns to ask for
For the conversation with your media buyer to be about money rather than cost per click, the report should look roughly like this:
| Column | Source |
|---|---|
| Ad, ad set, campaign | Ads Manager |
| Spend | Ads Manager |
| New leads and returning customers | CRM |
| Cost per new lead | Spend / new leads |
| Deals and conversion | CRM |
| Revenue | CRM, in the ad account currency |
| ROI | (Revenue − spend) / spend |
Half of the columns live in the CRM, and the media buyer cannot see them. That is why ad reports usually stop at cost per lead. Filling in the other half by hand once a month is doable; once a week it gets heavy, and with currencies (account in dollars, deals in sum) it is easy to get wrong. In shamCRM this report builds itself: campaigns, ad sets and ads, a comparison with the previous period of the same length, an "active only" filter and Excel export. Revenue is converted at the rate of the deal date, and if there is no rate for that day, the row shows no ROI.
How long to wait before drawing conclusions
A week is almost always too short. Meta spends the first days training the ad, and the numbers jump. Deals from ads arrive with a delay, especially the expensive ones: a person left a lead and then thought for two weeks. On top of that Meta backfills statistics, so the last two or three days of any report are not final.
For working decisions, 30 days per ad is enough. Where the line is after which you should switch off advertising entirely rather than ad by ad, we have no ready answer. Too much depends on the industry, the season and how long you are willing to wait for repeat purchases.