Discounts and promotions are at an all-time high, often comprising the single biggest cost within many retailers’ earnings statements. Yet despite these high stakes, and the growing adoption of sophisticated analytics, many retailers continue to take a broad-brush approach to running promotions that results in missed sales and lost profits.

The most common explanation for this lack of precision is that retailers tend to evaluate promotions at a high level, without drilling down to understand the impact of individual tactics within each promotion. By this we mean they compare the overall event to last year’s promotion in aggregate without understanding how different parts of the promotion might be more or less successful — getting the mix right is the key to creating more impact on sales and profits.

The good news is that many large retail organizations already have the tools and data they need to craft more effective promotions. What they lack, more often than not, is a logical way to structure their promotional effectiveness analysis. The solution can be found by asking a series of strategic questions, then carefully parsing the data for the answers.

When are discounts most likely to stimulate a sufficient sales response?

Promotions vary in effectiveness depending on the time of year, key holidays and even day of the week. Retailers that have a firm grasp of how their customers shop during different periods can use this information to formulate more precise promotional strategies. For example, during certain holidays — Christmas, Valentine’s Day, Mother’s Day, etc. — consumers tend to be in active shopping mode, and retailers should structure their promotions accordingly. Recent analysis for a leading apparel retailer showed elasticity of customer response to promotional discounts was up to three times higher during holiday periods than during nonholiday periods and two times higher during weekends than on weekdays. Given this, retailers are almost certainly throwing away sales and margin dollars by not tailoring their promotions to certain time periods.

Furthermore, retailers should use this insight to educate themselves about when aggressive promotions are a waste of effort and valuable margin dollars. If a retailer’s typical customers have a traditional workweek, it’s very difficult to entice them into the store for an unplanned purchase during the week, regardless of the type of promotion. Therefore, rather than trying to use steep promotions to drive traffic on weekdays, retailers should use more strategic promotions to encourage consumers who are actually in the store during the week to fill their baskets.

Does the promotion work best in-store or online?

It’s important to structure promotions based on what works best for a particular channel. Our recent work in women’s fashion suggests more complicated promotional structures aimed at basket-building — such as buy one item and get a second at a discount — work far better in-store. Simpler promotional structures — such as an off-the-top 20% discount — tend to yield better results online.

What products are most likely to garner more response when promoted? 

Not all items a retailer sells are created equal. Some have higher elasticity or stronger pricing power. Certain items that are on promotion will be especially effective in driving traffic, while others are better used as basket- or margin-builders. For example, our analysis in the men’s apparel category shows that compared with everyday basics, promotions for in-season fashion items bring significantly more people into stores. The same is true for lower-ticket sub-brands and product lines at aggressive price points.

At the same time, retailers need to be aware of the signals they send to consumers when deciding which items to promote and how often. Understanding which are the strongest or marquee items, brands and product lines is essential because their value could easily be eroded by over-promoting them. Promoting too frequently may cause consumers to equate a brand with low prices, which is probably not the profile a retailer wants to communicate unless it’s a low-cost retailer.

How does response vary across tactics (discount depths and promotion structure)?

Promotions comprise a rich set of tools — it’s no longer enough to just pull out a hammer every time there’s something that needs doing. Instead, retailers should match the tool to the job, making sure their promotional structure is aligned to the desired outcome and the manner in which their customers shop. For instance, if the aim is to build basket size, more complex, tiered promotions — such as giving increasing discounts for each subsequent item purchased — may be a better way to go. Copromoting the right items across the store, and knowing which promotional levels are too deep or not deep enough, will help drive sales further and improve profits.

Who is most likely to respond (e.g., new versus existing customer groups)?

Most retailers have a vision that a customized promotions strategy — one that optimizes promotions for every consumer — is the future. But for major marketed promotions, that’s not always achievable or practical. Nevertheless, retailers need to make sure that their efforts target their most important customers who can “move the needle” — either those who drive the most sales or the ones who are the most profitable. Companies now have the data and analytic power to create specific customer profiles at relevant group levels — for example, loyalty members or top customers, or those who tend to purchase on certain occasions or within certain categories — and craft promotional offers tailored specifically to those groups.

Getting promotions right

Savvy retailers now have the tools to build their promotional calendars from the bottom up, eliminating wasted promotional dollars and moving from scattershot decision-making to effective actions tied to strategic objectives (e.g., driving traffic, building baskets, driving sales and profits). Our experience suggests that retailers can find between 400 and 700 basis points of incremental margin dollars by deconstructing and deploying promotions that are strategically designed, targeted and timed.

Of course, the most effective promotional strategy will differ for each retailer and must account for many things, including the intended value proposition, competitive position and pricing power, and the specific consumer segments served. But through data science and a careful, strategic approach, retailers can solve the promotional Rubik’s Cube and unlock substantial value.

Editor’s note: This article originally appeared on the Harvard Business Review website (HBR.org).