How to monitor competitor prices and discounts

Search this question and you get a page of subscription product-page crawlers. They do a real job, but they all share one flaw: they tell you a competitor's price changed after it changed. The discount almost always appears in the rival's ad creative first, because the offer has to be advertised to work, and ad libraries are free to read. This guide covers how to monitor both layers properly, which one leads, and what 50,671 analysed UK ads tell us about how deep competitors actually cut.

The two layers of competitor pricing, and why only one is a leading indicator

"Competitor price monitoring" collapses two different jobs into one phrase:

• Price state — what a specific SKU costs on a competitor's site right now. This is what crawlers like PriceShape and the scraping stacks sell. It is the right tool for repricing decisions, and nothing here replaces it.
• Price intent — what a competitor has decided to push, how hard, to whom, and with what framing. This lives in the ad creative, and a product-page crawler is structurally blind to it.

The gap matters because of sequencing. A promotion needs demand to exist. Media gets booked, creative gets built and the ads go live at or before the moment the price changes, so the ad library carries the offer at least as early as the product page, and often earlier. More usefully, the ads carry the parts of the offer that never touch a product page at all:

• Acquisition-only pricing. First-order discounts, new-customer codes and win-back vouchers are targeted at people who aren't existing customers. They're deliberately kept off the public product page so current customers don't claim them. A page monitor never fires; the ad library shows them plainly.
• Commitment, not just the number. Thirty fresh creatives behind a price cut is an offensive. The same cut with no media behind it is housekeeping. The crawler records an identical price in both cases.
• The reasoning. Ads state the audience, the deadline and the justification ("end of season", "while stocks last"). That's the context that tells you whether to respond at all.

Where to read competitor prices for free

Every major platform runs a public, no-login ad archive, and prices and codes appear in the creative and copy:

• Meta Ad Library — Facebook and Instagram. The richest source for retail and DTC offers. Search by brand, filter to the UK.
• Google Ads Transparency Center — Search, Shopping, Display and YouTube. Shopping and Search text ads often carry the price outright.
• TikTok Commercial Content Library — strongest on younger-skewing DTC.

If you haven't used these before, our guide to seeing competitors' ads for free walks through each interface. This guide is about pricing specifically; for the wider mechanics of spotting and logging a sale as it launches, see tracking competitor promotions, which covers offer cadence and the recording workflow in more depth.

What to record from each ad

A pricing log is only useful if the fields are consistent. We capture these on every ad, and they're the same five you should write down by hand:

How deep do competitors actually discount? Real numbers

This is where generic pricing advice runs out, because it has no data. Across the 50,671 UK ads we've captured and analysed so far (158 brands, Google, Meta and TikTok), 16,954 — 33.5% — carry an offer of some kind. Of those, 6,403 have a discount recorded that we can measure as a percentage. The distribution is not smooth:

Two practical reads. First, discounting clusters on round psychological anchors: 20, 25 and 50 alone account for 2,889 of the 6,403 measurable offers. If you're setting a discount, 22% buys you nothing that 20% doesn't, and reads as a worse deal than 25%. Second, the mean recorded depth is 32.5%, and 1,738 ads sit at 50% or deeper. If your category's rivals are routinely advertising half off, a 10% counter is not a response.

Read all of that as directional rather than exact. 84% of the discount percentages we hold appear verbatim in the creative's copy or the on-image text we transcribe; the other 16% are our model's reading of the brand's prevailing offer rather than a number printed on the ad. That's reliable enough to compare sectors and spot escalation, and not reliable enough to quote a single rival's exact discount without opening the creative yourself. Whether an ad carries an offer at all is read directly off the ad and is the firmer number.

The offer-type mix: most "discounting" isn't a discount

Cutting the headline price is only one lever, and competitors reach for others first. Across the same corpus:

Finance, free trials and free shipping together outnumber percentage-off ads. This is the single most actionable pattern in the dataset: the strongest advertisers protect the headline number and attack the perceived cost instead. A product-page crawler watching the price field sees none of it, because the price field doesn't move.

Price positioning predicts how a rival will defend itself

We also classify where each ad claims to sit. Of 48,841 classified ads: value 22,551, mid 18,773, premium 4,031, budget 3,294 and luxury 192. Crossed against discount behaviour, the result is counter-intuitive:

Value-positioned brands discount most often and most deeply: they've trained their audience on price, so price is the only lever left. Premium and luxury brands discount rarely, and when they do they stay shallow, protecting the reference price. Note the luxury row rests on only 192 ads, so treat it as indicative rather than firm. The lesson for your own monitoring: a premium rival running 25% off is a far bigger signal than a value rival running 40% off, which is just Tuesday. Depth alone is meaningless without the positioning context.

Promo codes leak the campaign calendar

We currently track 871 qualified ads carrying a promo code across 38 brands. Codes are the most under-read pricing signal available, because the naming is rarely random. CBNEW (111 ads) is explicitly an acquisition code, new customers only, which is exactly the offer that never appears on a product page. FIRST15 (45 ads) states both the audience and the depth in six characters. FLASH (78 ads) and FLASH22 (25) mark short-burst discounting, and the numeric suffix implies a series you can expect to repeat. Codes with date stamps — we see things like SNWAUG26 and RACE0926 — hand you the rival's promotional calendar directly.

Worked examples from pages we publish

The method is easier to see on real brands. These are live teasers of what each is running:

• ASOS — 1,751 qualified ads, 377 with a measurable discount recorded, and value positioning throughout. A high-frequency discounter where the useful question is never "are they on sale" but "how deep, on what, this week".
• Charlotte Tilbury — 480 ads, premium positioning, but 171 with a measurable discount recorded and 76 stating a price outright. A textbook premium pattern: hold the reference price, concentrate the cutting into defined windows, lead with gifts and sets rather than percentages.
• Butternut Box — 519 ads with 417 discounted and 65 carrying a promo code: a subscription acquisition model where the first-order discount is the product pitch. Almost none of this is visible on a public product page.
• carwow — 1,734 ads, only 4 carrying a percentage discount, but 288 quoting an amount off and 103 advertising finance. A marketplace doesn't discount; it advertises the price it can get you. Price-led without being promotion-led, which a discount-only monitor would read as total inactivity.

Note how differently the four behave. Any monitoring setup that only watches for "% off" would see carwow as dormant and misread Charlotte Tilbury's restraint as weakness.

A monitoring routine that survives contact with a real week

1. Pick five rivals, not fifteen. Coverage you review beats coverage you collect and ignore.
2. Weekly, same slot. Open each brand in the Meta Ad Library and Google Ads Transparency Center, filtered to the UK.
3. Log the five fields. Price shown, depth, offer type, promo code, positioning. One row per new creative; skip anything you logged last week.
4. Record first-seen dates. Depth matters less than the date a new offer entered the market. That date is your lead time.
5. Review monthly, not weekly. Look for depth creeping up, offer type shifting from % off to finance or bundles (margin pressure), and codes repeating on a cycle.
6. Keep a page crawler for the SKUs that matter. The ads tell you what's coming; the crawler confirms the exact number on your twenty most contested products.

Budget roughly 45 minutes a week for five brands. The honest limit: manual review catches offers but not reach, and the libraries don't publish spend, so you can't tell a £500 test from a £50,000 push. Don't infer budget from ad count — our note on estimating competitor ad spend explains why that inference breaks.

The short version

Price-page crawlers answer "what does it cost now". Ad libraries answer "what is my competitor about to do, to whom, and how committed are they" — earlier, and for free. Watch depth against positioning rather than in isolation, treat finance and free-shipping offers as discounting by another name, and read promo codes as a published calendar. Then point a crawler at the handful of SKUs where the exact number actually changes your decision.

We maintain this as structured data because manual review stops scaling past about five brands. AdRival captures a brand's ads across Google, Meta, TikTok and LinkedIn, and qualifies every Google, Meta and TikTok creative we capture against 25 dimensions — price shown, discount depth, offer type, promo code and positioning among them — so a change in a rival's pricing posture shows up as a row in a report rather than something you hope you noticed. Brand reports are £99 one-off; a weekly tracker with promo alerts is £99 a month. The brand pages linked above are live samples, free to read.