For agencies that grow D2C brands
What should a change alert tell your agency before the brand notices?
A useful change alert tells your agency four things the morning a number moves: what changed against the brand's own normal range, where it changed, the likely cause with a confidence level, and the record behind it. Anything inside the brand's normal range should stay silent, so every alert you get is worth opening.
2.13× Meta ROAS against a 4× benchmark · flagged every day on Moolwanai-dm · Paid ads · 30 days to 26 Sep 2026 · real data
What you'll see in 10 minutes
- Your first brand's report, built from its domain alone.
- Its website health, search visibility, AI assistant visibility and brand mentions, read for the first time.
- The channels to connect next, so ai-dm can measure each one against the brand's own baseline.
Why do fixed thresholds miss real drops and flag false ones?
Saturday, 11:40. A brand's cost per purchase has been climbing since Thursday. Your Google Ads rule says "email me above ₹600", and the brand normally runs near ₹370, so the rule stays quiet. The brand's founder spots it first and sends a screenshot.
One number can't fit every brand. Set the rule low and a brand that normally runs at ₹580 trips it every weekend, until the team stops reading the emails. Set it high and a brand that normally runs at ₹370 can rise by half before anyone hears. Here is how the common options decide when to speak.
| Option | How it decides | What it leaves to you |
|---|---|---|
| Google Ads automated rules | A fixed condition you set per account, such as cost per conversion above a number | Setting and updating a rule for every brand and metric; everything outside Google Ads |
| Meta automated rules | A fixed condition you set per ad account, campaign or ad set | The same upkeep, brand by brand; everything outside Meta |
| GA4 custom insights | Conditions or anomaly detection on GA4 metrics | Ad platform figures, search rankings, AI answers, reviews, Maps and rivals |
| ai-dm change alerts | Each number against that brand's own baseline, every morning, across its channels | The decision: nothing changes without your approval. No attribution models |
Is this change real? Test it on your own numbers
Paste one metric's recent daily values, today's value and your current rule. The checker draws the brand's normal range and shows whether your rule would have spoken.
Cost per purchase, last 14 days and today
A fixed rule at ₹600 stays silent while this brand runs 18% above its normal range.
Free, no login. Example values are an illustrative example. Normal range here is a simple version: the median of your values, ±2 × their typical day-to-day spread.
No plan to pick. Pay as you go, per brand.
What should every change alert include?
An alert that only says "CPA is up" starts a hunt, not a fix. Someone opens three platforms to find which campaign, since when and why, while the brand waits. Five parts turn an alert into something your team can act on in minutes.
The number against its normal range
Today's value beside the brand's own range, so the size of the change is clear at a glance.
Where it moved
Brand, channel and campaign, so the right person picks it up first.
Likely causes, graded by confidence
The two or three most probable reasons, each marked high, medium or low.
The evidence
A link to the raw records behind the number, so anyone can check it before acting.
A suggested next move, waiting for a person
What to try, and what it should change, held until your team approves it.
Leave out: alerts for movement inside the brand's normal range, and alerts with no owner on your team. Both teach people to ignore the channel.
How many alerts a day is too many?
Agencies rarely miss a problem because they had no alerts. They miss it because the inbox held dozens that morning and the real one looked like the rest. Run these four checks on the alerts you get today.
Count last month's alerts, and how many led to an action
If most changed nothing, the rules are tuned to noise, and the team has learned to skim.
Swap fixed thresholds for each brand's normal range
A rule that suits one brand misfires on the next. Compare each brand with its own history instead.
Give every alert an owner on your team
An alert that belongs to everyone gets read by no one. Route each channel to the person who runs it.
Alert on opportunities, not only drops
A page gaining impressions in Google's AI results is worth a message to the brand before it spots it.
(login required) · Walk into the brand's next review having already seen the change.
How do ai-dm change alerts work?
Each brand gets its own baselines. ai-dm fixes a reference point for every metric it watches, brand by brand, so a change is measured against that brand's normal, not a number someone typed into a rule.
Every morning, every connected channel. ai-dm reads the brand's ads, search, AI search visibility, reviews, Maps, rivals' ads and site health, and raises what moved outside its range, good or bad. Connect every channel a brand runs to widen what it reads. Each alert arrives with likely causes, their confidence and the records behind them.
normal range ₹240–₹305 · +30% on its median
Likely causes
- Impression share lost to budget, 8% → 31%high
- A rival began bidding on the brand's namemedium
- Landing page loading 1.2s slowerlow
Suggested next move: raise the daily budget on "Coffee subscriptions" by ₹1,500 for 7 days, then re-check cost per conversion.
In this illustrative example, Terra Brew Coffee's Google Ads cost per conversion reached ₹371 against a normal range of ₹240–₹305. The most likely cause, graded high, is impression share lost to budget rising from 8% to 31%. A rival bidding on the brand's name is graded medium and a slower landing page low. The suggested budget change waits for a person to approve it.
The alert starts the work; your team makes the call. Nothing changes in a brand's account until a person approves it, and ai-dm then tracks whether the change moved what it was meant to. Across your roster, every brand's alerts land in one morning view sorted by what changed, so the brand that needs you comes first. When the first likely cause isn't enough, find why the numbers moved with every candidate checked.
On a real brand. Moolwan's Meta ROAS sits at 2.13× against a 4× e-commerce benchmark, and ai-dm raises it every day until it moves (ai-dm · Paid ads · 30 days to 26 Sep 2026). It also shows Moolwan named in just 1 of 233 tracked AI assistant answers (ai-dm · Sep 2026). We show weak numbers as plainly as strong ones. You can watch ai-dm run Moolwan and check the source of each number.
Related pages
Questions agencies ask about change alerts
What is a change alert in marketing?
A change alert is a notice that one of a brand's numbers has moved outside its normal range: cost per purchase, ROAS, search clicks, reviews or Maps rank, for example. A useful one says what changed, where, the likely cause with a confidence level, and the record behind it. ai-dm raises these for every brand you add, each morning. Start with one brand's domain to see it.
Will we get an alert for every small dip?
No. ai-dm measures each number against that brand's own baseline, so ordinary day-to-day movement stays quiet and only changes outside the brand's normal range come through. That keeps every alert worth opening. Try the checker on this page with your own numbers to see where a fixed rule would have stayed silent or fired for nothing.
Does ai-dm alert on good news too?
Yes. A rise outside the normal range in a good direction is raised as an opportunity, such as a guide gaining impressions in Google's AI results or a search query climbing. Telling a brand about an upside before it spots it makes for a stronger review than explaining a drop after.
What do the confidence levels on an alert mean?
Each likely cause is graded high, medium or low by how strongly the evidence supports it, and each links to the raw records behind it. A low-confidence cause is a lead to check, not a conclusion. ai-dm does not build attribution models, so it never claims one cause for certain.
Can ai-dm pause a campaign when an alert fires?
No. ai-dm suggests the next move, such as pausing an ad set or changing a budget, and waits. Nothing changes in a brand's account until a person on your team approves it, and every approved decision is recorded with what it was expected to move. That is how Moolwan runs every day.
How do we explain a change alert to the brand?
Show the brand three things: the number, its normal range and the record it came from. ai-dm keeps all three with every alert, and the same evidence can go into the white-label reports and portal your brands see under your agency's name. The conversation moves from "why didn't you tell us" to "here is what we saw and what we did".
No plan to pick. Billed in USD and INR. A brand's data is deletable on request.