How to read the six cards
Each card is one division. CTR is clicks divided by impressions, then multiplied by 100 so it reads as a percent. If you type 10,000 impressions and 150 clicks, CTR is 1.5 percent. That sentence is 150 divided by 10,000. It is not a result from a FashionUGC campaign, and it is not a benchmark. The same pair of numbers will produce 1.5 percent for a dress, a banner, or a blank cell in a spreadsheet.
CPC is spend divided by clicks. With a spend of 4,500 and 150 clicks, CPC is 30 in whatever currency those 4,500 were. CPM is spend divided by impressions, multiplied by 1,000. The same 4,500 across 10,000 impressions is a CPM of 450. The times 1,000 is the only reason CPM is not a tiny decimal. It means the cost of a thousand impressions, not the cost of one, and not a score out of 1,000.
Conversion rate here is orders divided by clicks, again as a percent. Nine orders from 150 clicks is 6 percent. CPA is spend divided by orders. 4,500 divided by 9 is 500. ROAS is revenue divided by spend. Revenue of 18,000 and spend of 4,500 is a ROAS of 4, which means four currency units came back for each unit spent, in the numbers you typed. It does not mean four times profit. It ignores the cost of the garment, returns, tax, and the fee the ad platform already may have folded into spend. If your spend figure is media only, say so in the brief. If it includes production, say that too. The calculator cannot tell.
A blank card is a missing denominator or a missing numerator you have not typed. Zero clicks with some impressions is a CTR of zero, and CPC stays blank because you cannot divide spend by zero. Zero impressions leaves CTR and CPM blank. Zero orders leaves conversion rate at zero if there were clicks, and leaves CPA blank. Zero spend leaves CPC and CPM at zero when their denominators exist, and leaves ROAS blank, because revenue divided by zero is not a return. Type a zero on purpose when the report showed zero. Leave the field empty when you do not have the report.
The note under the cards repeats that these are your figures. There is no sample button that loads a FashionUGC case study, because there is no case study. The worked numbers in the article below are the same kind of division, written out so you can check the page against a hand calculation. If the card and your hand disagree, trust the hand and reload. Rounding on the card is to two decimal places.
One worked row, labeled as arithmetic
Use this row only to check the tool. It is not a campaign. Impressions 10,000. Clicks 150. Spend 4,500. Orders 9. Revenue 18,000. CTR is 150 divided by 10,000, which is 0.015, which is 1.5 percent. CPC is 4,500 divided by 150, which is 30. CPM is 4,500 divided by 10,000, times 1,000, which is 450. Conversion rate is 9 divided by 150, which is 0.06, which is 6 percent. CPA is 4,500 divided by 9, which is 500. ROAS is 18,000 divided by 4,500, which is 4.
Change any one input and only the cards that use it should move. Drop orders to zero and CPA goes blank while CTR stays 1.5 percent. The click rate did not “cause” the orders. They were separate numbers you typed. Raise revenue to 22,500 and ROAS becomes 5, because 22,500 divided by 4,500 is 5. Nothing about the dress changed. The ratio changed because you changed an input.
A second imaginary row, still not a result: 8,000 impressions, 40 clicks, spend 2,000, orders 1, revenue 1,500. CTR is 0.5 percent. CPC is 50. CPM is 250. Conversion rate is 2.5 percent. CPA is 2,000. ROAS is 0.75. A ROAS under 1 means the revenue you typed is smaller than the spend you typed. It does not mean the video failed a FashionUGC test. It means, in that made-up row, less money was typed in the revenue box than in the spend box.
Do not average those two rows by eye and call it a blended truth. If you need a total, add the impressions, add the clicks, add the spend, add the orders, and add the revenue, then run the calculator once on the sums. Averaging a 1.5 percent CTR and a 0.5 percent CTR treats a large campaign and a small one as equals. Summing first gives the rate the combined delivery actually had, which in this arithmetic is 190 clicks over 18,000 impressions, or about 1.06 percent. That is still arithmetic on invented inputs. It is here so the method is visible.
The video does not contain the row
A FashionUGC file is an MP4 made from clothing photographs. It has a length and a frame. It does not have impressions. The bronze gown example is a 10-second 9:16 file from a mannequin still. The floral dress example is a 10-second 9:16 file from an on-body still. Neither file was run as an ad in a report this site can show you. Putting either clip in the calculator does nothing until you type numbers from somewhere else.
The stills are here so the creative and the spreadsheet stay separate objects. You can see the gown and the dress. You cannot see a CPM on them. If a deck puts a click rate on top of one of these frames, the click rate came from an ad account or it was invented. This page will not invent it. The examples are the place to watch the files. The ROI guide is the place to decide which store numbers belong in the row. The calculator only divides the row you bring.
A generated frame can differ from the source photograph. The gown’s mannequin becomes a person. That is a reason to accept or reject the file before you spend money behind it. It is not a conversion. A rejected file has no CTR because it was not the ad. Do not backfill a metric for a clip you did not run.
One job is one garment and one colour. If the ad showed the floral dress and the orders were for a black dress, the ROAS is not the floral video’s ROAS. Split the store report by the product that was actually bought. The calculator will not notice a mixed basket. You have to type the orders that belong to the piece in the file.
Where the numbers have to come from
Impressions and clicks come from the place that delivered the ad. Spend comes from that same place, for the same dates. If you pull clicks from one platform and impressions from another, CTR is nonsense. Write the platform name next to the row. The calculator has no platform menu, because a menu would look like a benchmark list. There is no list of good rates here.
Orders and revenue come from the store. Google’s note on the recommended purchase event is about a purchase analytics can receive when the store sends it. Shopify’s reports are the order book. The ROI guide says to match a week of store orders to a week of analytics purchases on that URL before you judge a video. Do that match before you type. A purchase event that double-fires will invent revenue. A purchase event that never fires will make a real week look empty. The calculator will faithfully divide a doubled revenue into a doubled ROAS.
Returns belong in the story even when the video looked fine. If you subtract returned revenue before you type, the ROAS is net of those returns. If you type gross revenue, the card is gross. Label which one you did. The calculator does not have a returns field, so it cannot do the subtraction in secret. A later return does not rewrite a card you already copied into a slide. Date the row.
Engagement on a page is not an order. Google’s engagement and bounce definitions describe how a session is counted. A long view of a product page can be a good sign and still not be the nine orders in the example above. Do not type engaged sessions into the orders field. Orders are completed purchases of the garment, from the store, for the dates of the ads.
Video plays are a third counter. Enhanced measurement in analytics can record video events for embedded YouTube videos that use the YouTube player API. A file in a shop gallery often will not. Do not type video views into the impressions field unless the ad platform literally called them impressions. A view and an impression are named by the platform. If you rename them, the CTR stops meaning what a media buyer thinks it means.
Keep two videos in two rows
A 10-second 9:16 file and a 20-second 16:9 file are different ads if you run them as different ads. They need different link parameters if they land on the same product URL. Google’s URL builder is how those parameters are added. Use a distinct utm_content for each file, and a distinct campaign name for the paid test, so the vertical cut and the wide cut do not share one bucket. FashionUGC does not append the parameters. You paste the finished URL into the ad.
Then run this calculator twice. Once for the vertical file’s impressions, clicks, spend, and the orders you can tie to that content parameter. Once for the wide file. Do not add the rows together and then talk about “the video.” The sum is a blend. The blend hides which file the clicks came from. If you want the blend as well, calculate it from the summed inputs, and label it as the blend.
Organic traffic to the product page should not wear the ad’s parameters. The ROI guide says that too. If a search visit is tagged as the campaign, the calculator will treat those clicks as paid. Your CPC will look cheaper than the invoice, because unpaid visits were typed into the click field. Pull the paid row from the ad platform, not from a mixed analytics channel, unless you have already filtered to the campaign.
A product page that holds the video without a paid ad has no CPM. You can still count orders on that URL. You should not invent impressions so the CTR card lights up. Leave impressions blank. Conversion rate needs clicks, so leave that blank too if you do not have a click definition you trust. Revenue and orders can sit in a note beside the calculator. Not every useful count is one of the six cards.
Length and metrics are easy to tangle. A 30-second file costs more to generate than a 10-second file, at the published prices of ₹500 and ₹149. That production cost is not the ad spend unless you decide to include it. If you include it, add it into the spend you type and write “includes one generation” on the row. If you leave it out, the CPC is media only. Either choice is allowed. Mixing them across two rows, one with production and one without, makes the cheaper CPC a lie.
What the cards cannot say
They cannot say a rate is good. This site does not publish a target CTR, a target CPM, or a target ROAS for clothing video. A 1.5 percent CTR in the worked row is not a bar to clear. A live row of 0.4 percent can be a bargain if the CPA is healthy, and a live row of 3 percent can be expensive noise if the clicks do not buy. The cards do not know your margin.
They cannot see the audience. The same file shown to people who already bought from you and to people who have never seen the brand will not share a CTR, and they should not. If you compare two creatives, keep the audience and the dates as close as the platform allows, and say where they differed. The calculator will not ask. A briefing line under the numbers has to.
They cannot see the offer. A discount in the ad and a full price in another ad will move CTR and ROAS for reasons that are not the hem of the dress. Do not credit the video with a sale the price caused, and do not blame the video for a sale the price blocked. Type the results you have. Explain the offer in words next to them.
They cannot see returns that you did not subtract, fees you did not include, or tax you treated differently in revenue than in spend. ROAS of 4 can be a loss after the garment cost. ROAS of 1.2 can be fine on a high-margin piece. The ROI guide is where that fuller worksheet lives. This tool stops at the six divisions so it stays checkable.
They cannot repair a bad file. If the colour is wrong, pause the ad. A low CTR is not the diagnosis you need. The diagnosis is the cloth. Generating another length will not fix a wrong colour, and a prettier CPM will not make the colour right.
Mistakes this page is here to catch
The first mistake is pasting a sample from this article into a client report. The 1.5 percent, the CPC of 30, and the ROAS of 4 are division practice. They are not FashionUGC performance. A report that presents them as a test of the bronze gown is a false report.
The second mistake is dividing by zero in your head and writing a huge CPC. If a row has spend and no clicks, CPC does not exist. The card stays blank so you do not publish infinity. Say “no clicks” in the brief.
The third mistake is mixing currencies. Spend in rupees and revenue in dollars will produce a ROAS that looks precise and means nothing. Convert before you type, or do not type revenue.
The fourth mistake is one link for two files. Without a separate utm_content, or a separate ad name you can actually export, you do not have two rows. You have one blended row. Calculate the blend once, and do not narrate a winner.
The fifth mistake is calling video views impressions. If the platform’s column is named views, your CTR is views in the denominator only if you also change the name of the rate. Otherwise you will compare it with someone else’s impression CTR and think a file won or lost. Name the denominator in the same sentence as the percent.
The sixth mistake is a ROAS victory on a tiny spend. Four divided by one is a ROAS of 4, and it is also one order. The card will say 4. The brief has to say one order. Small counts move wildly. The calculator does not print a warning flag, because a warning flag would look like a statistical test this page does not perform. You add the count of orders next to the ratio.
The seventh mistake is waiting for the metric to forgive a file you have not watched. Play the MP4 next to the photograph. If the garment is wrong, stop. Metrics on a wrong garment are metrics on the wrong product, and the store orders, if any, may be returns waiting to happen.
Write the row so someone else can rerun it
A number without a date is not a result you can defend. Next to the six cards, write the first day and the last day the ads ran, in the timezone of the ad account. A store that closes its day in a different timezone will not match that window to the hour. Say which clock you used. If the test was three days, say three days. A ROAS copied from a dashboard that defaults to “last 7 days” while the brief says “launch weekend” is a different row than the one you think you exported.
Name the platform in the same note. Impressions in one ads manager are not interchangeable with impressions in another, even when both columns use the word. A placement that counts a view after two seconds and a placement that counts an impression when the ad paints a pixel are different denominators. The calculator will still divide. Your sentence has to say which definition you exported. If you cannot find the definition in that platform’s own help, do not compare the percent with a percent from a second platform.
Name the garment the way the store names it: product title, colour, and the URL that was in the ad. One job on FashionUGC is one garment and one colour, and the row should be that narrow too. If the landing page sells three colourways and the file shows one, the orders field should be the colour in the file, not the page total, unless you are honest that you cannot split them. “Could not split” is a valid label. Pretending the page total belongs to the bronze gown is not.
Name the file by its length and frame, not by a mood. “10-second 9:16, bronze gown, job from 11 November” can be found again. “The hero cut” cannot, once a second hero exists. The length calculator only tells you which sold duration is nearest to a number of seconds. It does not score that duration. If you ran the 10-second file and not the 20-second file, the cards belong to the 10-second file. Do not borrow them for a length you have not paid to generate and have not put in an ad.
Keep the export. A screenshot of the six cards is not the source. Save the ad-platform export and the store export for those dates, with the column names visible. When someone asks why CPC is 30, you should be able to point at 4,500 and 150 in a file you still have, and then immediately say whether those two numbers were a real week or the practice row on this page. The practice row must never be the file you save into a client folder.
Paid delivery, a listing, and a page view are three counters
An impression in a paid campaign is a delivery event from that campaign. A view on a marketplace listing is whatever that marketplace counts when a shopper opens the listing. A page view in analytics is a hit on a URL. Adding them into one impressions field produces a CTR that no platform would recognise. If you are reading the marketplace guide and the ad generator page in the same week, keep two rows. The listing row may have no spend. Leave spend blank there. CPM stays blank, which is the correct card for a place you did not pay per thousand impressions.
A click has to be the same kind of click as the impression above it. Paid clicks over paid impressions. Listing clicks over listing views, and only if you are willing to call that rate something other than the paid CTR in the next paragraph. Do not put marketplace listing views under a paid click count. The percent will look decided. It will be a collage.
Revenue has the same split. An order from a paid click and an order from a search visit can both be real, and they should not be added into the ROAS of the ad unless you are calculating a blended store week on purpose. Label that blend. The ad’s ROAS uses the orders you can tie to the ad. The store’s week uses every order. Both sentences can be true. One card cannot say both at once, so run the calculator twice and keep the labels on the two notes.
Production cost sits in a third place. The prices to generate a clip are published on the site: ₹149, ₹199, ₹279, ₹349, and ₹500 for the lengths that are actually sold. That invoice is not an impression and it is not a click. Fold it into spend only when the question you are asking is “what did this test cost us in total,” and then the CPC is no longer a media CPC. Leave it out when the question is “what did the platform charge.” Write the question in the note so the next person does not “fix” your spend by adding the generation fee a second time.
None of this requires FashionUGC to connect to an ads manager. It does not. You export, you type, you keep the note. The file you generated is still only the garment in motion. If the note and the file disagree about which dress was advertised, believe the file you can watch, and correct the note before you correct the math.
A short way to brief a test
Name the file, the frame, the length, and the product URL with its parameters. Name the dates. Pull impressions, clicks, and spend from the ad platform for those dates and that file only. Pull orders and revenue from the store for the same piece. Subtract returns if net revenue is what you mean. Type the five numbers. Copy the six cards and the sentence that says they are division of that row.
Under the cards, write what was held still: audience, offer, placement, and how long the test ran. Write what was not held still. If you cannot fill that note, you do not have a comparison yet. You have a description of one row. A description is allowed. A winner is not, until a second row exists under the same rules.
Keep the row with the job. The next colourway does not inherit the CTR. It inherits the habit of counting the same way. The dress in the file is still the thing you have to be able to name with the sound off, whether the card says 0.4 percent or 2 percent.
