Late winter. The owner of an appliance store in Almaty, the one where order T-310 for a vacuum cleaner was placed, writes to you: "Our competitors are running videos on YouTube, and the only people who know us found us through search, Instagram or Kaspi. Let's invest in awareness too. But I need orders every week, not in six months." The numbers in this lesson are made up.
There are two jobs in that one message, and both are reasonable. Trouble starts when you measure them with the same ruler: the video by the cost of a search order, or search by reach. A month later the owner will decide the video flopped, when it was never meant to do what he is judging it on.
Video and search on the path to an order
Saturday
A person saw the appliance store's video
Tuesday
Searched Google for the store by name
Purchase
Last click credits the order to search
The video's contribution is assessed separately from last-click cost per order.
Two jobs, two rulers
A performance campaign works with people who already want to buy: someone types "buy a fridge almaty", someone left a cart on the website, someone is comparing prices on Kaspi. It is judged by orders and by how much ad money went into each order, and the result shows within a few days.
An awareness campaign works with people who are not looking for anything today. The bet is this: six months from now, when their washing machine breaks, they will remember the store, or pick it out of a list of unfamiliar names because they have heard of it. It is judged in stages: did the ads reach the right people, did they remember them, and only then, did orders go up. That takes weeks and months.
| Performance | Awareness | |
|---|---|---|
| Who it works with | People already searching or choosing | People who will need appliances later |
| What you count | Orders and ad spend per order | Reach and frequency in Almaty, then searches for the store by name, then orders from every storefront |
| When the result shows | Within days | Over weeks and months |
| At the store | Google search: about 40 orders a week on last click, roughly 5,000 ₸ of ads per order | A YouTube video and banners: 600,000 ₸ over four weeks |
The platform does not set the job. You can run a campaign for orders on YouTube, and you can buy ads on your own name in search so that a competitor does not sit above you. First decide what the campaign is for, then pick the metric and the time frame to fit that purpose.
There is another reason you cannot fold the video's reach and search CPA into one measure of success: the channels depend on each other. Someone sees the video on Saturday, types the store's name into Google on Tuesday and buys. On last click the order goes to search, even though search did not create the need: it collected demand that other touchpoints prepared. Switch off the video and some of the "search" orders may go with it.
Does the business need awareness
Before talking about budget, answer five questions. A "no" to any of them is a reason to put awareness off and do what the right-hand column says first.
| Question | The store's answer | If the answer is "no" |
|---|---|---|
| Is the ready demand already captured? | Yes: impressions lost to budget in search are about 2%, so new money there would go into pricier clicks | Put money where people are searching: those are the cheapest extra orders |
| Does the choice depend on a familiar name? | Yes: appliances are bought rarely and cost a lot, and fewer people know the store than know the chain competitors | Invest in price, delivery times and search |
| Is the path to an order fixed? | Yes: the promises in the ads and on the site match, and leads go into the CRM | Fix it first: ads will bring more people to the same holes |
| Can the business wait? | Yes: the video money is separate, and the owner is prepared not to see it back for several months | Put it off until a quarter when you can wait |
| Is there anything that will show a trace? | Yes: about 3,000 searches by name a week in Wordstat | It is a blind bet: keep the amount small and tell the owner so |
The first question gets confused with a different one: whether people search for the product at all. In spring an air conditioner repair shop shows an Instagram ad saying "get it cleaned before the heat" and judges it by orders that same month. Hardly anyone searches for cleaning, but this is performance: the offer creates demand and sells straight away. You need awareness when the purchase will happen later but the person has to remember you now.
The last question decides whether you get any feedback at all. A Brand Lift survey asks people about ad recall and brand awareness, but for Kazakhstan it starts at 5,000 dollars for 10 days per question, and reading its result takes skill too. The store is nowhere near that bar, so its trace is searches by name. At 3,000 such searches a week, a rise of a fifth is visible. At 40 a week, you cannot tell it from the usual swings.
Awareness money does not come out of search
The answer to "I need orders every week" is simple. This week's orders come from search, Instagram ads, Kaspi and retargeting, and their budgets stay as they are. The video goes on a separate line, as a test on top of the base. If you cut search to pay for the video, orders drop straight away and you never find out what the video did: two things changed at once.
The size of the test is set by how much the owner is willing to lose if no effect shows. Say the store keeps 7,000 ₸ per order. Then 600,000 ₸ comes back with 86 extra orders, about 22 a week over four weeks. At 115 orders a week across all storefronts, that is a rise of almost a fifth. If you do not expect that much in the first month, say so before launch. Then the 600,000 ₸ is a bet on orders months from now, not a purchase of orders this week, and the owner decides whether he is willing to make it.
Horizon: what to ask of the campaign, and when
The horizon is the date when you will ask the campaign for a result, and the number you will judge it on. An awareness campaign has three such dates, and you write them down before launch:
| When | What we look at | What we decide |
|---|---|---|
| 3-5 days after launch | Impressions in Almaty, unique users, frequency, share of viewable impressions | Fix the settings: location, frequency cap, placements. These numbers are not used to judge the result |
| End of the campaign, week 4 | Searches for the store by name against the season-adjusted expectation | Whether the ads left a trace in memory |
| End of a pause of the same length, week 8 | Orders from every storefront against expectation, and searches by name after the pause | Whether the wave paid back, and whether the trace holds |
The expected level without ads comes from a category search index: searches by name grow on their own when the season comes.
Purchases on the day of the first impression do not measure awareness. Someone who saw the video in April might buy a fridge in July after moving, and a short window only counts people who were going to buy that week anyway. But you cannot wait forever either. If orders have not gone up by week 8, "let's wait another month" without a new rule turns into "let's wait until something lines up".
The store wrote its rule as one sentence per date. There is a trace if every campaign week's searches by name are above the strongest baseline week adjusted for the season. The wave paid back if 86 extra orders accumulate over eight weeks. If there is a trace but no payback, the next wave is a bet on later orders, and the owner names the amount for it.
The same rule guards against the opposite swap: a campaign for orders that brought no orders does not become an "image" campaign after the fact.
Reporting to an owner who needs orders this week
The weekly report has two lines, each with its own numbers. At the end of the campaign the message to the owner looks like this:
Four weeks in total. Orders: an average of 128 a week across all storefronts against an expected 120, which is within the normal spread. About 40 from search, at roughly 5,000 ₸ of ads per order, so search is on plan. Awareness: we spent 600,000 ₸. 150,000 people saw the YouTube video, 4 times each on average, and 100,000 saw the banners. On Yandex, people searched for the store by name 3,800 times a week on average against an expected 3,120. We look at orders from the video after the four-week pause, as agreed.
What is not in the report matters just as much. There is no last-click cost per order for the video: 600,000 ₸ over 9 orders is 66,667 ₸ an order, and that number judges the video against a job it was not given. There are no view-through conversions added to orders: they say someone saw a banner and bought, not that they bought because of the banner. And there is no line saying "a million impressions, awareness went up". Impressions say the ads were delivered, not that anyone remembered them.
By the end of the pause the store had its answer from the table of branded searches and orders. Searches by name were 22% above expectation during the campaign and went back to it during the pause. Orders over the eight weeks cannot be told apart from the normal spread, and they are far short of 86 extra. By the rule, there was a trace but no short-term payback. Whether to drop awareness altogether or repeat it with a smaller amount is the owner's call, but he makes it by the rule he approved before launch, not on what GA4 showed on Monday.
Exercise: an awareness plan for your own project
Take a business where someone is asking for "awareness": your own, a client's or one you know. The result is one page the owner signs off before launch.
A good plan answers the owner's question before he asks it. Take a car wash in a residential neighbourhood whose owner also wants "awareness". Its name gets searched about ten times a week in Wordstat, and its Google ad for "car wash near me" is still losing impressions to budget. The plan stops at the very first question: the cheap extra customers are still in search, and the owner can see that in one Google Ads column.
If all five answers were "yes", the next check would be the bar. A test that pays back with a rise of a few percent, when ordinary weeks swing by a third, cannot be made out in orders. That goes into the plan up front, not into a discovery two months later: we will judge by searches for the name or by a survey, and the owner is paying for a bet, not for a measurable result.
The last thing to rewrite before sending is the line "a million impressions, awareness went up". In the report it reads: "the video was shown so many times to so many people in the city; the name was searched so many times against an expected so many". And without a written "yes" to the plan, the conversation in week two goes back to "where are the orders?", and you will have nothing to answer with.