Competitor timeline: how to read 90 days in 5 minutes

Nobody reads a bank statement line by line: you glance at the rhythm of the deposits, recognize the recurring debits, and only really stop at the entry that broke the pattern. A competitor timeline reads the same way. Each detection is a line on the statement: it has a date, type, channel and value. And your competitor's campaign history, which looks too long to analyze, becomes a 5-minute read once you know what order to look in. This guide shows the method: three passes (rhythm, pattern, exception) over the 16 detection types the AI classifies automatically.

Every detection is a line on the statement

In Batedor, the timeline lives on the Detected campaigns page (the path is /campanhas in the dashboard). Each detection carries the fields you would use to check a bank entry:

• Date: when the signal was first and last seen (the card shows “Since Mar 12” or the full range).
• Type: one of the 16 classifications the AI assigns, from “Discount coupon” to “Product launch”.
• Channel: which platforms the signal appeared on (Instagram, Facebook, YouTube or the competitor's own website).
• Value: when present, the discount percentage or amount and the coupon code appear highlighted on the card.
• Confidence: how sure the AI is of the classification, as a percentage.

One detail that changes the reading: detections from the same competitor with the same type are grouped into a single campaign. A feed post, a story and a banner on the site announcing the same free shipping are not three campaigns: they are three pieces of evidence for the same decision, and the dashboard shows the group with a “3 detections” counter. This avoids the classic mistake of overestimating a rival's activity by counting the same move several times.

The statement also records closings. A campaign becomes “Ended” when its end date is in the past or when the signal disappears from the sources for 14 days. The duration of each entry is information: a coupon that stayed up 3 days tells a different story from a coupon that stayed up 3 weeks.

The 16 detection types (and what each family reveals)

The 16 types exist so you don't have to interpret post by post. In practice, they organize into six families, and each family points to a different reason behind the move:

The family matters more than the isolated type. Three “Direct price” detections in the same week tell a conversion story. Three “Inventory turnover” detections in the same period tell another one, far more interesting for anyone competing head to head.

How to read a competitor timeline in 3 passes

The method is the same as the bank statement: you don't analyze entry by entry, you make three sweeps with different questions. With the history open on screen, each pass takes a minute or two.

Pass 1: rhythm (the cadence of new campaigns)

First question: how many campaigns a week does this competitor run? Ignore types and values, count frequency. A rival that always ran two promotions a month and started running two a week changed something: a tight target, margin under pressure or new management. The opposite informs too: a prolonged silence from someone who used to promote every week may signal a repositioning or an inventory problem. In the page filter, use “All” for this pass: if you look only at the active ones, the ended campaigns disappear and the rhythm gets truncated.

Pass 2: pattern (what repeats)

Second question: what always comes back in the same place? A coupon that reappears in the first week of every month (aimed at payday), free shipping that switches on every Friday, a launch every 45 days. A pattern is not news: it is commercial policy, something decided once and executed on a loop. The value of the pattern is different: it is the baseline that lets you judge everything else. Without knowing that CUPOM10 comes back every month, you treat its fourth appearance as news and react for nothing.

Pass 3: exception (the recent decision)

Third question: what broke the pattern? A type that had never appeared on the timeline, a depth above the historical norm, a new channel for the same competitor. The pattern tells the policy; the exception tells a decision someone made last week, and a recent decision is the most valuable information monitoring delivers. This is the pass where rigor pays off: before reacting, check whether the exception really adds up to an aggressive campaign or just a routine variation. And when exceptions pile up in a short window, the reading moves to another level: the case of the clearance sale detected in 9 days is exactly that, five cumulative exceptions that, on their own, looked innocent.

Filters: isolate the signal before reading

A good read starts with the right cut. The campaigns page offers three filter axes, and each pass calls for one:

• Status (“Active”, “Ended”, “All”): rhythm and pattern call for “All”; the “Active” view serves the operational question of the day, “what am I competing against right now?”.
• Type: filtering to only “Discount coupon” builds your coupon dossier; only “Product launch” becomes a map of the rival's portfolio expansion.
• Competitor: read one timeline per rival. Two competitors mixed together scramble the cadence and hide the individual patterns.

Pay attention to the confidence of each detection too. Above 85% the classification rarely disappoints; below around 60%, treat it as a hypothesis and open the captured evidence (the post, the story or the page that generated the detection) before taking it to a meeting. And for daily follow-up, the Dossier summary at the top of the page condenses the last 24 hours: reading 90 days is a different exercise, worth doing once a month or before big dates like Black Friday and Mother's Day.

What the timeline doesn't show (and how to compensate)

No bank statement tells the whole story, and here it pays to be direct about the limits:

• Public content only. The collection covers Instagram, Facebook, YouTube and the competitor's website. A promotion that ran only in a private email or WhatsApp list never becomes a detection.
• Behavior, not motive. The timeline records what the competitor did; the why (a target, cash, a collection swap) is your hypothesis, and a hypothesis gets tested before it becomes a reaction.
• Movement, not result. You see the rival's campaign, not how much they sold with it. A frequent campaign is not the same as a campaign that works.
• Classification has a margin of error. The confidence score exists precisely for that, and anything that fits none of the 16 types falls under “Other”.

The practical consequence: start before you need to. The 14-day trial, no card required, already records the first two weeks of the statement, and that is history you can't recover later.