If your business has felt slower lately, you may be noticing something similar when you are on the other side of the register. Companies that rarely seemed to offer discounts are promoting more sales, restaurants are leaning harder on specials and loyalty offers, and customers appear increasingly willing to look around before deciding what feels worth the money.
There is evidence behind some of that behavior. Deloitte’s 2026 retail research found that four in ten Americans now show deal-driven or cost-conscious behavior, including people in higher-income households. The same research is a useful reminder that value is not just price: quality, service, convenience, loyalty, and the overall experience can materially affect what customers believe something is worth. (Deloitte)
That matters if your own sales have started to soften. The temptation is to respond quickly with a discount, more ads, a bundle, an email blast, or more social media, but each of those solves a different problem.
Before changing the marketing, figure out what changed in the business.
“Sales are down” is not a diagnosis
Suppose you own a neighborhood bakery that has done well for years. Saturday mornings still feel busy, but more pastries are left in the case at closing during the week, and the register does not seem to ring quite as often in the afternoon.
It would be easy to conclude that customers think the bakery has become too expensive. Maybe the answer is a 20 percent discount after 3 p.m., a two-for-one offer, or more advertising to get people through the door.
Those ideas might work. They also might address the wrong problem.
The bakery could be seeing fewer customers overall, but it could just as easily be seeing the same regulars visit less often. Customers might still come but buy one pastry instead of a box, weekday afternoons could be weakening while mornings remain healthy, or production could simply be based on an older demand pattern that no longer matches what people are buying.
Those situations all end with lower sales. They should not all lead to the same response.
A useful first pass looks more like this:
| What changed? | What it might point to |
|---|---|
| Fewer transactions | Fewer customers or fewer visits |
| Average purchase is lower | Customers are buying less or trading down |
| Regulars visit less often | Repeat frequency may be slipping |
| Only certain hours are weak | A daypart or occasion problem |
| Certain products sit longer | Product mix or value may have shifted |
| More questions about specials | Customers may be more value-conscious |
| More food is left at closing | Demand, product mix, or production may be out of sync |
| Plenty of interest but fewer purchases | Something may be getting in the way of the sale |
You do not need a formal marketing department to look at those questions. A point-of-sale system, appointment book, handful of customer conversations, recent reviews, staff observations, and your own knowledge of the business can provide a surprising amount of useful information.
Pay attention to how customers are changing, not just whether sales changed
Customers do not always respond to a tighter budget by disappearing. They may visit less often, order fewer items, switch to a lower-priced option, wait for a promotion, or save the more expensive purchase for a time when it feels more worthwhile.
That distinction matters because the response should match the behavior. A customer who still loves your business but visits once a month instead of twice is a different problem from someone who stopped considering you altogether.
Restaurants are a good example. Deloitte’s 2026 restaurant research notes that businesses are under pressure to drive traffic while protecting margins as some customers become more cost-conscious, but it also finds that price is only part of the value equation. Quality, service, convenience, and execution still influence whether people believe an experience is worth paying for. (Deloitte)
So if customers start asking about deals more often, do not automatically translate that into “our prices are too high.” They may be telling you that they need a clearer reason to choose you, a different size or package, more convenience, a better occasion to visit, or simply another price point to choose from.
Too Good To Go is an interesting example
Too Good To Go is a marketplace where restaurants, bakeries, cafés, and stores sell surplus food that might otherwise go to waste. Businesses offer unsold food in discounted Surprise Bags, and the company says more than 180,000 businesses participate. (Too Good To Go)
As a consumer, seeing a popular restaurant or bakery on the app can be surprising. A place that once seemed difficult to get into may now have a very inexpensive offer available at the end of the day.
That does not automatically mean the restaurant is struggling. The business may simply have unpredictable surplus and would rather recover some revenue than throw good food away, which is exactly the use case Too Good To Go promotes. (Too Good To Go)
For the business owner, though, it creates several useful questions. Are Surprise Bags mostly appearing on the same days? Are the same foods repeatedly left over? Are people discovering the restaurant through the app and later coming back at full price? Is the program recovering value from inventory that would otherwise be lost, or is it revealing that production needs to be adjusted?
AI can help work through those questions because it can connect information that may normally live in separate places. A restaurant could combine a few weeks of surplus food, day of week, sales, product mix, promotions, and customer feedback and ask AI to look for patterns before assuming that the answer is simply “discount more.”
Give AI the clues you already have
You do not have to build a dashboard first. Start with what you can reasonably pull together in ten or fifteen minutes.
For a retail, food, service, or local business, that might include:
- a rough comparison between a good recent period and a slow one
- number of transactions, bookings, or customers
- approximate average purchase
- busiest and slowest days or times
- products or services that are moving differently
- things that are being left unsold or unused
- recent customer questions
- five to ten recent reviews
- whether regular customers seem to be coming back as often
- price, hours, staffing, inventory, or service changes
- new competitors or offers you are seeing nearby
- the marketing action you are considering next
This is not supposed to become a research project. You are trying to give AI enough context to help you form a better explanation than “business is slow.”
A prompt like this is a good place to start:
My business has been slower lately, and I want to understand what changed before I run a discount or spend more on marketing. I’m going to give you what I know about transactions, customer spending, repeat customers, slow days or times, products or services that are moving differently, customer questions, recent business changes, competitors, and anything else I have noticed.
Separate what the information tells us from what we are assuming. Give me the three most plausible explanations, tell me what evidence supports each one, identify the simplest thing I could check to distinguish among them, and suggest one small test before I make a larger marketing or pricing change.
The wording matters less than the assignment. You are not asking AI to invent marketing ideas yet; you are asking it to help you make sense of the business.
Make AI look beyond marketing
One reason this works well is that slow sales can come from places that have very little to do with advertising.
A restaurant can have strong demand but too much production at the end of certain nights. A consultant can have plenty of leads but an offer that customers do not understand, while a shop can have steady foot traffic and declining revenue because shoppers are choosing lower-priced items.
An appointment-based business can lose sales because popular times are no longer available. A retailer can believe customers have become price-sensitive when the real issue is that the product people came in for has been out of stock.
Give AI permission to consider those explanations too.
| Area to examine | Questions worth asking |
|---|---|
| Customers | Are they coming less often or spending differently? |
| Offer | Does what you sell still feel worth the price and effort? |
| Operations | Did hours, staffing, inventory, service, or capacity change? |
| Product mix | Are customers shifting toward different items or services? |
| Competition | Has another business changed its offer, price, hours, or positioning? |
| Timing | Is the slowdown concentrated around specific days, seasons, or occasions? |
| Repeat business | Are previous customers returning at the same pace? |
| Discovery | Are enough new people still finding you? |
| Buying experience | Is something making it harder to purchase, book, or understand the offer? |
| Broader market | Are customers becoming more selective about discretionary purchases? |
This is also a good way to keep the economy from becoming the explanation for everything. Customers may be more careful with spending, but that does not mean every decline in sales is caused by the economy.
If you suspect that is the problem, make AI challenge it:
I believe customers are spending less because they are becoming more cautious with money. Treat that as a hypothesis, not a fact. Tell me what I would expect to see if that explanation were true, give me two other explanations that could produce the same pattern, and tell me what I can check this week to separate them.
If customers are still buying premium products on Saturday but weekday traffic is weaker, that tells you something. If traffic is stable but the average purchase is lower, that tells you something different.
Before you discount, ask what the discount is supposed to fix
A sale can be a smart business tool. It can move perishable inventory, bring customers in during a quiet period, encourage trial, reward regular customers, or give someone a reason to buy sooner.
The risk comes from using a discount simply because revenue feels uncomfortable.
Suppose the bakery is selling out most mornings but has too much food left after 4 p.m. A late-day offer may make perfect sense because it is tied to a specific inventory problem.
If Saturday is already full price and busy, there is little reason to discount Saturday. If weekday customers are visiting less often but still speak highly of the bakery, a targeted offer might be worth testing without lowering prices for everyone.
There is another issue worth watching. Deloitte notes that increased value-seeking can push brands toward more price-based competition, which raises the risk of training customers to expect discounts rather than strengthening the broader reason they choose the business. (Deloitte)
Before launching a promotion, ask AI:
I am considering this promotion: [describe it]. Based on what I know about sales, customer behavior, slow periods, product mix, and recent changes, tell me what problem this promotion is supposed to solve. Explain what it could help with, what it would not fix, what margin or customer-behavior risk I should watch, and whether there is a smaller test I could run first.
That question can be especially useful when a business feels pressure to copy the discounts it sees competitors offering. Their problem may not be your problem.
Try one thing that teaches you something
Once you have a reasonable explanation, resist the urge to change five things at once. A small test gives you a chance to improve the business while learning whether the diagnosis was right.
For the bakery, perhaps the strongest clue is that weekday afternoons are weak and surplus rises after 4 p.m. The owner might test a limited late-day bundle on Tuesday and Wednesday instead of cutting prices across the menu.
The test should answer a question. Did the offer bring in customers who would not otherwise have purchased, move inventory that would have gone unsold, increase the number of items per transaction, or simply give a discount to people who were already planning to buy?
AI can help review the result afterward:
We believed that weak weekday afternoon demand was causing more unsold inventory. We tested [offer] on Tuesday and Wednesday for two weeks, and here is what happened before and during the test: [results]. Tell me whether the evidence makes that explanation stronger, weaker, or still unclear. Also check whether we created additional business, shifted purchases from another time, or reduced margin without enough benefit.
That is a much more useful AI loop than generating a new promotion every time sales soften.
The questions customers ask are part of the diagnosis
Do not overlook the things customers are already telling you. Repeated questions about price, sizes, scheduling, availability, shipping, ingredients, what is included, or how something works often contain useful clues about where buying becomes difficult.
The same goes for reviews. Positive reviews can reveal the reason customers believe you are worth choosing, while negative or mixed reviews can show friction that may never appear in a sales report.
You can give AI anonymized customer comments and ask it to sort them without overinterpreting them:
Group these customer comments into recurring themes such as price, value, convenience, quality, service, trust, timing, availability, and confusion. Show me what customers directly said separately from what you infer it may mean, and identify the one or two patterns that are strong enough to investigate.
That separation matters because AI can create a very convincing explanation from a handful of comments. Customer language is evidence; the reason behind the language may still need to be tested.
A simple slow-sales check
Before putting money behind a solution, answer these eight questions.
| Question | What it may reveal |
|---|---|
| Are there fewer transactions? | Whether customer volume changed |
| Are people spending less each time? | Changes in basket or service mix |
| Are regular customers visiting less often? | A repeat-frequency issue |
| Is the decline concentrated somewhere? | A specific day, time, product, or service problem |
| What are customers asking about more often? | Price, value, convenience, trust, or clarity |
| What is being left unsold or unused? | Inventory or capacity mismatch |
| What changed inside the business? | Operational causes |
| What changed around the business? | Competition, local conditions, or consumer behavior |
You do not need perfect answers to every question. Even rough observations can help AI distinguish between several plausible explanations and identify what is worth checking next.
The useful output should sound something like: “Here is what we know, here are the two or three explanations that fit, here is what is still uncertain, and here is the smallest thing worth testing.” If AI jumps from a few numbers straight to “increase your ad budget” or “offer 20 percent off,” ask it to show you what evidence makes that the right response.
Slow sales are telling you something
A slow month does deserve attention. When money is tighter and customers are more selective, waiting indefinitely and hoping things improve is not much of a strategy.
But reacting quickly and diagnosing carefully are not opposites. A few customer comments, sales observations, leftover products, booking patterns, or changes in purchase behavior can help narrow the problem before you give away margin or spend money attracting more people.
AI is useful here because it can help connect those clues and challenge the first explanation that comes to mind. Once you have a better idea of what changed, you can decide whether the right response is a promotion, a different offer, a reason to bring customers back, a change in inventory, better communication, more marketing, or something else entirely.
That is a better place to start than assuming every slow period needs another sale.