How to Set Marketing KPIs That ActuallyDrive Retail Sales

‍

Here's a fun exercise. Pull up your last marketing report and ask one question about every metric on it: if this number dropped by half tomorrow, would anyone change what they're doing?

If the honest answer is "probably not" for most of the page, you don't have a KPI problem. You have a reporting habit.

That's the trap a lot of CPG and retail-distributed brands fall into. The dashboard grows every quarter. Impressions, reach, engagement rate, video completion, ROAS by network, share of voice. All real numbers. Very few of them tell you whether you sold more product, and even fewer tell you what to do next.

Good KPIs do two things. They connect to revenue, and they force a decision. Everything else is context.

‍

Start with the sale, then work backwards

‍

The most useful way to build a KPI set is to start at the register and walk backwards. What has to be true for a shopper to buy your product?

They have to find it. It has to be in stock. The price and the pack have to make sense next to the competition. And somewhere along the way, something has to tip them toward you instead of the brand next to you on the shelf.

Each of those steps has a metric attached. Distribution and in-stock rate. Search rank and digital shelf presence. Price gap versus your closest competitor. Conversion rate on the product page. Repeat rate after the first purchase.

Notice that only some of those belong to "marketing" in the traditional sense. That's the point. If your marketing KPIs ignore availability and shelf position, you can run a brilliant campaign that drives shoppers to an empty slot.

‍

Separate your KPIs into three tiers

‍

Not every number deserves the same airtime. We like a simple three-tier setup.

Tier 1: Business outcomes. These are the numbers your CFO and your retail buyers care about. Sales, velocity, share, household penetration, repeat rate. You probably don't control them on your own, but everything you do should ladder up to them.

Tier 2: Commerce drivers. These are the levers marketing can actually pull. New-to-brand buyers from retail media, conversion rate on key product pages, incremental sales from a campaign, share of search on your priority keywords, in-store activation compliance.

Tier 3: Diagnostics. Clicks, CTR, CPMs, engagement, video views. Useful for optimizing a campaign while it's running. Dangerous when they show up in a quarterly business review as proof that things are working.

The mistake most teams make is reporting Tier 3 to leadership because it's easy to pull and always looks busy. Keep Tier 3 inside the working team. Lead with Tier 1 and 2 everywhere else.

‍

Pick fewer KPIs than feels comfortable

‍

If you track 30 KPIs, you're really tracking zero. Nobody can hold 30 priorities in their head, and when everything is a KPI, the one that matters gets lost on slide 14.

A practical rule: no more than two or three Tier 1 metrics and five or six Tier 2 metrics per brand or campaign. Everything else goes into an appendix that the working team reads and leadership doesn't.

This forces a useful conversation. If you can only pick five drivers, which five actually move the business this year? For a new launch, it might be distribution, trial and first-to-second purchase. For a mature brand defending share, it might be price gap, search rank and repeat rate. Different problems, different scorecards.

‍

Tie every KPI to a retailer reality

‍

A national number can hide a lot. Your overall sales might be flat while you're growing fast at one retailer and quietly losing ground at another.

That's why the best commerce KPIs are set at the retailer level, not just the brand level. Your goals for a grocery partner should look different from your goals on a marketplace, because the shopper behaves differently, the data you get back is different and the levers you can pull are different.

A simple scenario: a beverage brand sets one national goal for retail media ROAS. Sounds tidy. But on one network, ROAS looks great because most of the spend is hitting loyal buyers who were going to purchase anyway. On another, ROAS looks weaker because the campaign is actually finding new households. Judged on one blended number, the team cuts the second network. They just cut the part that was growing the brand.

Set the KPI where the decision gets made.

‍

Build in a "so what" for every metric

‍

Here's the test we use internally. For each KPI, write down what you'll do if it goes up, and what you'll do if it goes down.

If new-to-brand rate drops, we shift budget toward upper-funnel placements on that network. If in-stock rate dips below an agreed threshold at a key retailer, we pause paid traffic to those product pages until supply catches up. If repeat rate stalls after a launch, we look at the post-purchase experience before we spend more on trial.

If you can't write a "so what" for a metric, it isn't a KPI. It's trivia.

‍

Agree on definitions before the campaign starts

‍

This one sounds boring. It's the reason half of all post-campaign meetings turn into arguments.

What counts as "incremental"? What attribution window are we using? Are we counting online and in-store sales, or just online? Does a new-to-brand buyer mean new in the last 12 months or ever?

Retail media networks don't all measure these things the same way, so if you don't pin definitions down upfront, you'll end up comparing numbers that were never comparable. Write the definitions on one page. Get your brand team, your sales team and your agency to sign off. Then measure against that page every time.

‍

Review on a rhythm that matches the business

‍

Diagnostics can move daily. Commerce drivers usually need a few weeks to show a real trend. Business outcomes often take a full quarter or more.

Match your review cadence to the metric. Weekly for in-flight optimizations. Monthly for drivers. Quarterly for outcomes and for deciding whether the KPI set itself still makes sense. Brands that judge long-term metrics on a weekly basis end up chasing noise and killing good ideas before they have time to work.

‍

The bottom line

‍

A KPI isn't a number you report. It's a number you're willing to make decisions on.

Start at the sale, keep the list short, set goals where the retailer reality lives, and attach an action to every metric. Do that, and your dashboard stops being a scrapbook and starts being a steering wheel.

At Five Eighty, we build commerce scorecards that connect media, shopper and retail data to the numbers that actually move the business. If your reporting is long on activity and short on answers, let's talk.