Competitor Pricing Analysis: Turning Raw Moves Into Weekly Decisions

Most teams that run a competitor monitor end up drowning in scraped prices and starved of a verdict. The monitor returns a CSV every hour; the alert firehose pings Slack at midnight; someone on the team opens the dashboard on Monday morning and has no idea which of the 4,000 line items is the one that needs a decision before lunch. Monitoring collects data; alerts push events; what ships the verdict to the operator is the weekly report — the artifact that turns raw moves into a documented decision the rest of the company can read.

What competitor pricing analysis actually produces

Competitor pricing analysis is a dated, written deliverable — usually a weekly report — that turns a firehose of scraped prices into a small set of decisions with owners, deadlines, and dated sources. It is not a dashboard, not a notebook full of queries, and not an alert: it is a thing a person reads end-to-end and walks away from with an action. Three sections, in order:

1. The headline section. Three to five SKUs that moved materially this week, with the absolute and percentage move, the reference price used, and a one-line recommendation (hold, raise, drop, watch). The headline is what the buyer reads first and what the leadership team skims on Mondays; everything else in the report is evidence for it.
2. The category moves section. Median shifts per category, with a count of SKUs that crossed the band's ceiling, dropped below the floor, or moved outside the MAP threshold. This is where the trend shows up that the alerts hide — three modest drops on a single category can signal a competitor campaign that no single-event alert will ever surface.
3. The exception section. The long tail: SKU-pair outliers where the relationship to the band has flipped, where MAP drift was detected, or where the competitor's price moved without a plausible cause. The exception section is where new SKUs get added to the watchlist, where a competitor gets blacklisted from the alert surface, and where a category gets a deeper-dive on the next cycle.

A monitor tells you what happened. An alert tells you it just happened. The analysis tells you what to do about it — and is the only one of the three the leadership team will actually read on Monday morning.

Choosing the right slice for this week's report

The biggest mistake in a first analysis cycle is trying to cover the whole catalog. A 12,000-SKU report that nobody reads is worse than a 200-SKU report that the buyer finishes with a pen in hand. Two slice strategies work:

Slice by revenue band

The top decile of SKUs by revenue drives most of the margin. The analysis covers that set every week, with a deeper row per SKU; the next three deciles get a category roll-up; the long tail gets a single "no movement against band" line and is sampled only when an alert fires. The reader gets a report they can finish in fifteen minutes and the long tail still gets covered by the underlying alert surface.

Slice by trigger event

For the first six weeks of a new program, slice by what moved: any SKU that crossed the ceiling, dropped below the floor, or had a competitor MAP reset. The trend on a single campaign reveals itself two weeks earlier this way because the analyst is reading same-direction moves together, not as a flat ranked list.

Writing the headline section so someone reads it

The headline section is the only part of the report the leadership team reads. It is usually three to five rows, one row per SKU, in a fixed-shape format that the operator can scan in fifteen seconds. The format is the design — a freeform summary paragraph will be skim-read and forgotten. Each headline row carries five fields:

• The SKU and category. Plain text, never a SKU ID alone; the reader needs to recognize the product on sight.
• The absolute and percentage move. Both, always. A 6% drop on a $14 SKU is a different event from a 6% drop on a $640 SKU and the dollar figure is the only one that survives a quick read.
• The reference price. Your list, your last-shipped cart median, your price band's ceiling — whichever was used as the baseline for the verdict. Without the reference, the percentage move is meaningless.
• The verdict and the owner. Hold (with reason), raise (with target), drop (with floor), or watch (with the date that flips watch into action). Each row names the person accountable for the action by Friday.

A report with eleven rows in the headline section is a report nobody reads. If eleven SKUs moved materially this week, the analyst picks the five that matter and queues the other six for the next cycle.

Running the category and exception sections without losing the trend

The category moves section is where the campaign signals live. The trap is to render every category as a flat median shift and let the reader infer the signal; the median hides the bimodal distribution that a competitor campaign actually produces. Two habits keep this section useful:

1. Show the count, not only the median. A category where 4 of 38 SKUs dropped 8–12% on Tuesday and Wednesday is a campaign; a category where 38 of 38 SKUs dropped 2% over the week is a market-wide move. The same median can mask both, and the count is the only field that disambiguates them.
2. Annotate the outliers. Any category with a single SKU that moved more than three standard deviations from its category median gets a footnote in the exception section, with a one-line cause (MAP reset, stock-out, end-of-season promo, data error). The footnote is how an analyst catches a scraping bug without losing an afternoon to it.

Operating the weekly cadence without burning out the analyst

A weekly analysis cadence is only sustainable if the report ships in a fixed slot, has a fixed shape, and is read by a fixed group. The operating rules are the design:

• Ship on the same day every week. Monday is the conventional slot for B2C and marketplace teams; pick a day and never move it. A report that ships when the analyst finishes it is a report the leadership team stops relying on within a month.
• Use a template, not a fresh document. The headline, category, and exception sections have fixed column orders. Filling a template takes an hour; writing a new report takes a day.
• Track the realized vs recommended gap. A report whose recommendations are overridden 60% of the time is a workflow problem, not a model problem. Track which verdicts landed versus which verdicts the team actually shipped.
• Audit the source per quarter. A scraped competitor can disappear, can go behind a login wall, or can start returning the wrong SKU 8% of the time. The report's value is downstream of the data quality; the quarterly source audit is the cheapest insurance against a silent breakage.

Frequently asked questions

How is competitor pricing analysis different from competitor price monitoring?

Monitoring is the data collection layer — it scrapes competitor sites and accumulates rows on a schedule. Analysis is the downstream artifact that turns those rows into a verdict the team can act on. Monitoring is necessary for analysis, but analysis cannot be derived from monitoring alone: the analyst still has to choose the slice, write the headline, and decide which verdict ships.

How often should a competitor pricing analysis be published?

Weekly is the default for B2C and marketplace catalogs, because the leadership team's Monday-morning cadence matches it and most repricing cycles settle inside a week. Higher-velocity categories (consumer electronics, fashion, hardlines around a launch) sometimes ship twice a week; long-tail catalogs sometimes ship biweekly. The cadence matters less than the consistency: a report that ships on a fixed day is a report that gets read.

What does the analyst do when nothing moved this week?

Ship the report anyway, with "no movement against band" as the headline and the band's current state as the body. The shipped report proves the monitor is still working and gives the team a baseline to compare next week against. A skipped week is how a monitoring pipeline silently breaks for a month before anyone notices.

Analysis is the layer between the monitor and the alert-pipeline -> repricing decision loop. The other posts in this sequence cover the upstream inputs (what to scrape, how often) and the downstream real-time surface (what fires when a move matters).

For the inputs to the report — sources, scrape cadence, and the data layer beneath the analysis — see our competitor monitoring guide. For the real-time events that drive the exception section, see the competitor price alerts playbook. For the per-SKU ceilings and floors the headline verdicts reference, see competitive pricing bands. To see a weekly analysis compile against live competitor data, try the pricing report tool, or compare the tier with scheduled monitoring on the pricing page.