The AIDA model rewired for 2025: attention is not enough any more

AIDA is four questions, not four boxes

AIDA stands for Attention, Interest, Desire, Action. It is the oldest model in marketing that people still quote by name, and it says something very simple: before anyone buys from you, they have to notice you, want to know more, want the thing, and then actually do something about it. Four steps, in that order.

The mistake is to treat the four words as four boxes that a customer sits in, one after the other, like the queue in the post office on a Friday. Nobody experiences their own shopping that way. What the four words really are is four questions I can ask about any piece of marketing: does this get noticed, does it make anyone curious, does it make anyone want the product, and does it make the next step easy? A page or an advert that fails one of those questions fails, however good the other three are.

That is the whole model. Everything else in this article is about how to count it, how to use it for an online shop today, and where it runs out.

Worked example: a tea shop in Cork

Throughout this piece I will use an invented shop, Shandon Tea, a loose leaf tea business in Cork run by a woman called Maeve. Nobody real, no real numbers, every figure made up so that every step can be shown.

Maeve runs a short video on social media showing what happens when you pour hot water on a tea bag next to a glass of loose leaf tea. That is attention. The caption asks why a country that takes its tea this seriously still drinks most of it out of a paper bag. That is interest. The landing page shows three starter boxes at €12, €18 and €32, with a tasting note for each and a line from a customer in Galway who switched and did not go back. That is desire. A single "Add to basket" button, delivery cost shown before the basket, guest checkout, pay with the wallet already on the phone. That is action.

Four questions, each answered once. If the video is dull, the rest never happens. If the video is brilliant and the checkout asks for a password, the first three steps were wasted.

Where the model came from, and why the date on it is wobbly

Nearly every textbook says the same thing: AIDA was invented by an American advertising man called E. St. Elmo Lewis in 1898. I repeated that for years. It turns out to rest on thinner ground than its confidence suggests.

The 1898 story comes almost entirely from Edward K. Strong, an American psychologist who wrote about the theory of selling in 1925 and credited Lewis with a three step slogan, attract attention, maintain interest, create desire, from an advertising course in the 1890s.¹ Later writers copied Strong, and the date hardened into fact. In 2024 a marketing historian, Akinori Iwamoto, went back through the primary sources, including trade magazines that have since been digitised, and found no evidence that Lewis invented the model in or around 1898. He argues the credit belongs more to two other salesmanship writers of the time, Arthur Frederick Sheldon and Frank H. Dukesmith, who worked out the four step version and treated the fourth word, action, as the one that mattered.²

So the honest version is: AIDA grew out of American sales training between roughly 1900 and 1910, several people had a hand in it, and the neat attribution to one man and one year is a story that was repeated until it looked like a footnote. That matters to me for one reason. The model was built to script a sales call, face to face, one salesman to one prospect. It was never a theory of how people buy. It was a checklist for how to talk to them.

The academic family it belongs to is called the hierarchy of effects. In 1961 Robert Lavidge and Gary Steiner proposed a six step version, awareness, knowledge, liking, preference, conviction and purchase, and argued that advertising should be measured by how far it moves people up those steps rather than by sales alone.³ Thirty years later Thomas Barry and Daniel Howard reviewed almost a century of this literature and concluded that the evidence people actually move through the stages in a fixed order is weak.⁴ Buyers skip steps, loop back, and a good few of them buy first and develop the interest afterwards.

None of that makes AIDA useless. It makes it a checklist rather than a law. I find that more useful, not less, because a checklist does not pretend to know the order your customer will arrive in.

An example: a friend in Kilkenny bought a bread knife she had never heard of because it was the one on the counter when she paid for a loaf. Action, then desire (she liked it), then interest (she read about the maker), then attention (she now notices the brand everywhere). The four things all happened. The order was backwards. A fixed sequence would have called that impossible. A checklist shrugs and ticks four boxes.

The funnel is a bookkeeping device, so count at every stage

The funnel is the picture people draw of AIDA: wide at the top, narrow at the bottom, because more people notice you than buy from you. The shape is not a claim about psychology. It is bookkeeping. Each stage is a count of people, and each pair of stages gives a ratio.

The diagram runs from the widest count at the top to the narrowest at the bottom. Each box is a number of people in a given period. Each arrow between two boxes is a rate, and the rate is where the information lives. The counts tell you how big you are. The rates tell you where you are losing people.

Every rate in the funnel is the same formula:

$$r = \frac{\text{people who reached the next stage}}{\text{people who were at this stage}}$$

Here $r$ is the conversion rate between two neighbouring stages, the numerator is the count of people in the lower box, and the denominator is the count of people in the box above it. A rate only means something when you know both numbers, which is why I never accept a conversion rate without the counts behind it. The idea is covered from scratch in Foundations: ratios and rates.

Once you have the rates, the number of orders is just the top count multiplied down the chain:

$$\text{orders} = N \times r_1 \times r_2 \times r_3$$

Here $N$ is the number of people who saw the advert or post, $r_1$ is the visit rate (visits divided by people reached), $r_2$ is the basket rate (baskets divided by visits), and $r_3$ is the checkout rate (orders divided by baskets). Each is a plain rate, events divided by opportunities, as Foundations: ratios and rates explains. Multiply the four together and you have the orders. It is one line of arithmetic and it is the most useful line in this article.

Worked example: Shandon Tea in October

Maeve counts a month. I will use round numbers so every step is visible.

Attention: her video and her search adverts were shown to 50,000 different people. Interest: 5,000 of them visited the site. Desire: 600 of those added a starter box to the basket. Action: 180 placed an order.

The three rates are:

$$r_1 = \frac{5{,}000}{50{,}000} = 0.10 \qquad r_2 = \frac{600}{5{,}000} = 0.12 \qquad r_3 = \frac{180}{600} = 0.30$$

So $r_1$ is a visit rate of 10%, $r_2$ is a basket rate of 12%, and $r_3$ is a checkout rate of 30%, which is to say that seven in ten baskets were abandoned. Checked against the formula: $50{,}000 \times 0.10 \times 0.12 \times 0.30 = 180$. Grand.

The overall rate from seeing the advert to ordering is 180 divided by 50,000, which is 0.36%. Put another way, 3.6 orders for every 1,000 people reached. With a mean order of €32, October brought in 180 times €32, or €5,760.

That 70% abandonment figure is not me being gloomy about Maeve. Baymard Institute keep a running average of published cart abandonment studies, and across 50 studies from 2006 to 2025 the unweighted average sits at 70.22%.⁵ Maeve is, by that measure, perfectly normal. Normal is still expensive.

One better stage multiplies through the whole funnel

Because the orders are a product of the rates, improving any one rate by a given fraction improves the orders by the same fraction. The maths does not care which stage you pick. Your budget does.

Worked example: fixing the checkout instead of buying more attention

Suppose Maeve moves the checkout rate from 30% to 36%. That is six percentage points, which is a 20% improvement (the difference between points and a percentage is in Foundations: percentages). Nothing else changes.

$$50{,}000 \times 0.10 \times 0.12 \times 0.36 = 216$$

Here $N$ is still 50,000, $r_1$ and $r_2$ are unchanged at 0.10 and 0.12, and only $r_3$ has moved from 0.30 to 0.36. Orders go from 180 to 216, an extra 36 a month, or about €1,152 at €32 each. The visitors were already there. She paid nothing more for attention.

Now the alternative. To get the same 216 orders by buying attention, she would need 60,000 people reached instead of 50,000, 20% more, at whatever the adverts cost per person. In the shops I have worked on, that is nearly always the more expensive route, because the top of the funnel is where the auction is and the bottom is where only you are.

The improvements also multiply with each other. If Maeve lifts the checkout rate to 36% and the basket rate from 12% to 14.4%, each a 20% improvement, the orders become $50{,}000 \times 0.10 \times 0.144 \times 0.36 = 259$ (rounded). Two 20% improvements give a 44% improvement, because $1.2 \times 1.2 = 1.44$. That is the quiet argument for working on two or three stages at once rather than hunting for one heroic fix.

Along the bottom is the checkout rate, the share of the 600 baskets that turn into orders. Up the side is the number of orders in the month, with the 50,000 reached, the 10% visit rate and the 12% basket rate held steady. The bars rise in a straight line because orders are directly proportional to the rate: every five percentage points on the checkout is 30 more orders for Maeve, roughly €960, without a cent more spent on being noticed. The takeaway is not that checkout is always the stage to fix. It is that you should know, in orders and euro, what each stage is worth before you decide where to spend your month.

One caution on small numbers. Maeve's 180 orders are a month. A week is about 45. If she changes the checkout on Monday and sees 11 orders on Tuesday instead of six, that tells her almost nothing, because a handful of orders bounce around on their own. How many you need before a change is real is the subject of Foundations: sampling, and I go deeper on it in how much data you actually need.

Attention is earned again and again, not captured once

In the sales call AIDA was written for, you had the prospect's attention from the moment they let you through the door. On the web you have a thumb hovering over a screen, and you get attention in small, repeated doses or not at all.

This is where the old version of this article, and most articles on this subject, reached for a statistic: the human attention span has fallen to eight seconds, less than a goldfish. I have cut it, because it is not true. The BBC traced the figure to a 2015 Microsoft Canada report, which itself cited a website called Statistic Brain, whose listed sources turned out to hold no such research at all. The psychologists the BBC spoke to said the idea of an average attention span is close to meaningless, because attention depends entirely on the task, and there is no evidence that it is shrinking.⁶ The goldfish, for what it is worth, has a perfectly good memory and is used in laboratories to study learning precisely for that reason.

What has changed is not the human. It is the number of things competing for the same moment. So the practical rule is not "say it in eight seconds". It is "expect to say it several times, in several places, and make each time count on its own".

In practice that means I treat attention as something that accumulates. A person sees the video, scrolls on, sees a search advert a week later when they type "loose leaf tea", recognises the name, and clicks. Two exposures, one visit. Neither exposure "captured" attention. Together they earned it. GDPR and consent banners mean I usually cannot stitch those two moments to one person in the analytics, so the top of the funnel is always counted with some blur. I accept the blur, use reach and visits as the counts I can trust, and stop pretending to a precision the data does not have.

An example: Maeve's first version of the video opened with the Shandon Tea logo for two seconds. Nobody is interested in a logo they have never seen. The second version opened on the tea bag going grey in the glass, and the logo came at the end, once there was a reason to remember it. Same product, same budget, the second one earned the visits. Attention is not won by announcing yourself. It is won by showing the viewer something about their own life.

Interest and desire are about answering the question the visitor arrived with

Interest is the moment someone gives you a second look. Desire is when they start imagining the product in their own kitchen. The two are different, and online they tend to fail for the same reason: the page does not answer the question the advert raised.

Every visitor arrives with a question. The person who clicked the tea bag video wants to know whether loose leaf tea is genuinely better or just more bother. The person who searched for "best Assam tea Ireland" wants to know whether this shop has good Assam. The person coming back for the third time wants to know whether the €18 box is enough for two people for a month. If the page opens with a slogan about passion for tea, all three questions go unanswered and all three visitors are one tap from gone.

So I build interest by matching the page to the route in. A cold visitor from a video gets the argument, briefly: why the leaf matters, what it costs per cup. A visitor from a product search gets the product, the price and the delivery date above everything else. A returning visitor gets the thing they looked at last time, not the homepage again. This is the practical face of Jobs To Be Done (JTBD): work out what the person is trying to get done, then make the page about that.

Desire is built from evidence rather than adjectives. A tasting note is evidence. "Delivered to Ennis in two days, and the box was still sealed" from a customer is evidence. A price per cup next to the price of a coffee shop latte is evidence. "Premium artisan blends crafted with passion" is noise, and your reader knows it is noise, because they have read it on every other site this week. The desire stage is where I look hardest at what a page claims and ask whether a sceptical person in Limerick would believe a word of it.

Worked example: the €18 box

Maeve's data shows that visitors from the video look at the €12 box, and visitors from search look at the €32 box, and that the €18 box in the middle is rarely clicked at all. The reason turns out to be the copy. The €12 box says "try three teas". The €32 box says "a month of tea for two". The €18 box says "our most popular selection", which answers no question anyone has. She changes it to "two weeks for one, four teas", adds a line that it works out at about 26c a cup, and within the month it is the box most often added to the basket. Nothing about the tea changed. The page simply started answering the question.

Action is a chain of small actions, and the chain breaks at the basket

In the salesman's version, action was the signature. On the web it is a chain: tap the product, pick a size, add to basket, open the basket, start the checkout, type an address, choose a payment, confirm. Every link in that chain is a place to leave, and the data says most people do. Seven in ten baskets, remember, never become orders.⁵

The usual reasons are dull, which is good news, because dull problems have dull fixes. Delivery cost that appears for the first time at the checkout. A forced account before you can pay. A form that rejects a perfectly good Eircode. A payment method that is not the one already in the phone. None of these are marketing problems. They are plumbing. But the AIDA checklist puts them where they belong, at the last step, carrying all the cost of the first three.

Beyond the checkout itself, I have come to treat small actions as real actions. Saving a product. Entering an email for a delivery reminder. Sending the basket to yourself to look at on a laptop later. Each of these is a person telling you they are not ready to buy today but have not said no. In a world where I cannot follow them from phone to laptop with a cookie, these small actions are the only thread I have, and they are given freely, which is rather the point. I write more about where that money goes in cart abandonment as a five figure problem.

Worked example: delivery shown early

Maeve's checkout shows delivery at €3.95 on the payment page. Of 600 baskets, 180 complete. She moves the delivery cost to the product page and the basket, adds free delivery over €30, and lets people pay with the wallet on their phone without an account. The next month, with the same 600 baskets, 222 complete, a checkout rate of 37%. Orders go from 180 to 222. At €32 a go, that is €1,344 more a month, from the visitors she already had. Fair play to her, and no great mystery either. In my experience that is the shape of nearly every first win on a shop: the best value attention you will ever get is the attention you have already paid for and are currently wasting at the basket.

What AIDA cannot see, and where to read next

AIDA stops at the purchase. It was written for a salesman who got paid on the sale and did not much care what happened afterwards. For an online shop that is exactly the wrong place to stop, because the second order is where the margin lives, and the model has no word for it.

That is not a flaw to fix inside AIDA. It is the edge of the tool. Three other articles pick up where it ends, and I would rather point you at them than cram a fifth letter onto a model that has done its job.

The full funnel growth framework extends the four steps through retention and advocacy and shows how to find the leaks across the whole system rather than one stage. Pirate metrics (AARRR) is the version built for software and subscription businesses, with activation and referral as stages in their own right, and it is the one I reach for when the product is used rather than unboxed. And the Hook Model deals with what happens after action: how a product becomes a habit without dark patterns, which is the question AIDA never thought to ask.

If you want the underlying picture of how people actually decide, before any framework, start with why consumer behaviour is the basis of marketing strategy.

How I actually use it

Here is the whole method, as plainly as I can put it.

I take any campaign, page or shop and ask the four questions in order. Is it being seen by enough of the right people, and can I count them? Does the page answer the question the visitor arrived with? Is there evidence on the page that would convince a sceptic, rather than adjectives? Is the next step small, visible and possible without an account?

Then I count. People reached, visits, baskets, orders, for one period, and the three rates between them. I write down what a 20% improvement at each stage would be worth in orders and euro, exactly as in the Shandon Tea example, and I fix the stage where the gap between cost and value is widest. It is nearly always further down the funnel than the marketing budget is currently pointed.

Then I accept that the top of the funnel is blurry, that people arrive out of order, and that the model ends at the first purchase. AIDA was a checklist for a sales call in about 1905. Used as a checklist, with counts attached, it still earns its place on a wall in Cork in 2025. Used as a theory of the customer, it never did, and the historians have now caught up with that too.

Attention is not enough. It never was. It is the first of four questions, and the other three are where the orders are.

Footnotes

1. Edward K. Strong Jr., Theories of selling, Journal of Applied Psychology, volume 9, issue 1, pages 75 to 86, American Psychological Association, 1925, https://doi.org/10.1037/h0070123

2. Akinori Iwamoto, The origin of AIDA: who invented and formulated the AIDA model?, Japan Marketing History Review, volume 3, issue 2, pages 150 to 166, 2024, https://www.jstage.jst.go.jp/article/jmhr/3/2/3_150/_article/-char/en

3. Robert J. Lavidge and Gary A. Steiner, A model for predictive measurements of advertising effectiveness, Journal of Marketing, volume 25, issue 6, pages 59 to 62, American Marketing Association, 1961, https://www.jstor.org/stable/1248516

4. Thomas E. Barry and Daniel J. Howard, A review and critique of the hierarchy of effects in advertising, International Journal of Advertising, volume 9, issue 2, 1990, https://www.warc.com/en/article/a-review-and-critique-of-the-hierarchy-of-effects-in-advertising-08fa9db6887e41f6bc5594bf6567777f

5. Baymard Institute, 50 cart abandonment rate statistics 2025, average documented online shopping cart abandonment rate 70.22%, Baymard Institute, 2025, https://baymard.com/blog/cart-abandonment-rate

6. Simon Maybin, Busting the attention span myth, BBC World Service, 10 March 2017, https://www.bbc.com/news/health-38896790