The Acquisition Map

The linear funnel (awareness, interest, desire, action) is a useful diagram and a false model. Google's own purchase-behaviour research replaced it with a "messy middle": a loop between exploration (expanding the option set) and evaluation (narrowing it) that a person runs until confidence crosses a threshold. People loop, double back, and leave for weeks. The job is not to push them down a tube. It is to be the easiest option to find and the most reassuring option to choose, every time they re-enter the loop.

The last stage is the cheapest version of the first. Increasing customer retention by 5% has been shown to raise profit by 25% to 95%, because a kept customer costs nothing to re-acquire and refers others for free. This is why the flywheel replaced the funnel: retention and advocacy are not the end of acquisition, they are its cheapest fuel.

Two jobs, not one. Demand creation makes a person want the thing before they're shopping (most social, video, content, PR, word of mouth). Demand capture catches a person already shopping (search, shopping ads, comparison). You need both. Capture without creation scales until the created demand runs out, then stalls.

People trust people. In Nielsen's global trust study, recommendations from people you know are the most trusted form of advertising by a wide margin (88% of respondents), ahead of every paid format. A referral arrives pre-qualified and pre-trusted, which is why referred customers convert better and churn less.

It is created upstream and spent downstream. The work that earned the recommendation (a good product, a remembered experience, a piece of content) happened weeks or months before the search that "captured" the sale. The channel that gets the credit is almost never the channel that did the work.

Word of mouth is not purely luck. Referral programs, post-purchase prompts, and a genuinely remarkable experience all raise the rate. The loop closes deliberately in Phase 05; the advocacy you build there re-enters the top here.

Dark social is any share that strips its origin: a link pasted into a chat, an email forward, a screenshot of a product. The browser that finally opens it sends no referrer, so the visit looks like someone typed your URL from memory. A large majority of all content sharing happens this way, not through public, trackable buttons.

As more conversation moves into private messaging and closed communities, the share of discovery that is structurally invisible grows. In considered and B2B purchases especially, much of the real evaluation happens in private channels and AI assistants long before anyone fills in a form.

You cannot track it directly. You can infer it: a self-reported "how did you hear about us?" field at conversion, and correlation between content launches and spikes in direct or branded traffic. Both covered in Phase 06.

This is mental availability being built: the brand becoming easy to recall at the moment a need finally appears. Ehrenberg-Bass research shows growth comes mainly from reaching all category buyers, including the light buyers who buy rarely, not from squeezing loyalty out of heavy users. Reach the 95% who aren't in-market so you are remembered when they are.

Most of this is consumed without a click. Someone watches, scrolls past, listens in the car. There is no link event. The response shows up later as a direct visit or a branded search, on a different day, credited to a different channel. Podcasts are the extreme case: pure audio, no clickable surface, and demonstrably undervalued by click-based measurement.

Before a considered purchase, buyers look for corroboration they didn't have to take your word for: a review, a Reddit thread, a "best in category" article, increasingly an AI assistant's summary. This is the evaluation half of the messy middle running on sources you don't own. Strong earned presence reduces the perceived risk of choosing you.

AI assistants now sit inside the research phase, answering "what's the best tool for X" with no click and no referrer to you at all. The query is invisible, the recommendation is invisible, and the visit (if it comes) lands as direct. Being citable by these systems is becoming its own discipline.

Reviews are not only social proof on your own page. They are an entry point: people search "[category] reviews" and "[competitor] alternative" and discover you inside a third-party list. Your review profile is a channel, not a vanity number.

When someone searches your brand name, the demand already exists. Something created it upstream: an ad they didn't click, a friend, a podcast, a post. Branded search is the most reliable signal that your demand creation is working. It is not the thing that created the demand. Rising branded search volume is one of the cleanest proxies you have for upper-funnel health.

Whether to buy ads on your own brand name is a genuine debate. Sometimes a competitor bids on your name and you defend. Often you pay for a click you would have got for free, and the conversion was always going to happen. The honest test is incremental: hold out branded paid in a region and see if total branded conversions actually fall.

Non-branded SEO captures demand the searcher already feels, but it can also create preference: a problem-aware article meets someone before they know you exist and becomes their first impression. It is durable in a way paid is not. A page that ranks keeps earning after the work is done, where an ad stops the moment the budget does.

The mechanics of how a page goes from published to crawled to indexed to ranked to clicked, and every place it silently drops out, is a full system in itself. It is the planned sibling of The Signal Chain. Here it is one node; there it will be the whole map.

The search result is no longer ten blue links. AI overviews, featured snippets, and zero-click answers increasingly satisfy the query on the results page itself. Ranking is necessary and no longer sufficient; being the cited source inside the answer is the new front line.

Paid search and shopping put you in front of people already searching for what you sell. It is the purest demand-capture play: high intent, immediate, and measurable. It is also an auction, so its cost rises with competition and its volume is capped by how much in-market demand exists. You cannot buy more demand than has been created.

Performance Max spends across search, shopping, display, video, and Gmail from one campaign, optimising to a blended goal. That makes it convenient and opaque. It can quietly spend on branded search (cheap, high-credit) and broad display (cheap, low-quality), then report a flattering blended ROAS. Demand creation and demand capture get averaged into a single number that hides which is which.

The conversion event these campaigns optimise toward must be defined and clean, or the algorithm trains on noise. That plumbing is The Signal Chain. Whether a given ROAS is actually profitable depends entirely on margin, which is The Money You Actually Keep.

Marketplaces, maps, and comparison sites are where ready buyers shortlist. For local services, the Google Business Profile and the "near me" pack are often the single highest-intent surface that exists. For ecommerce, a marketplace listing is distribution you didn't have to build. The intent is high and the discovery is real.

You rent the demand and often forfeit the customer. The marketplace owns the email, the data, and the next purchase; the comparison site charges for the lead and ranks you against rivals on its terms. Strong here, weak in your own retention loop, and you are building someone else's flywheel.

The site is where all the upstream work is spent or wasted. It is also the only step in the entire journey you fully control: not an auction, not an algorithm, not a third-party shelf. That makes it the highest-leverage node on the map. A small lift in conversion rate multiplies the return on every channel feeding it at once.

The page is also a filter that fails silently. A slow page loses people who never see an error and never get counted as lost. Speed and clarity are not polish; they are conversion infrastructure, and they tax every acquisition channel equally.

The eleven-block anatomy of the page that turns a stranger into a customer, from above-the-fold to post-purchase confirmation, is its own field manual: The Conversion Journey. This block is the pointer; that playbook is the territory.

Retargeting catches the people the consideration loop already produced: they came, they didn't buy, you remind them. Click-through and conversion rates are several times higher than cold prospecting, because the audience is warm. Built well, it is genuinely the most efficient line in a paid account. Built lazily, it is a machine for paying to reach people who needed no reaching.

This is the trap that costs the most, so it gets its own paragraph. Retargeting shows your ad to people about to convert anyway, then claims the conversion. Holdout tests repeatedly find that a large share of retargeting-attributed sales would have happened with no ad at all. The attributed ROAS can overstate the true, incremental ROAS by a multiple. The number is real; the credit is borrowed.

Exclude recent converters (or you pay to sell to people who already bought). Cap frequency (or you irritate and waste). Set recency windows by considered purchase length (a 1-day window for impulse, 30+ for high-consideration). Use strict custom audiences, not broad auto-expansion, when the whole point is precision.

Every channel above this is rented: the platform owns the audience and charges you to reach it again. An email address or phone number flips that. It is the moment a stranger becomes someone you can reach for free, on your schedule, without an auction. For most considered purchases, capturing the contact is more important than the first-visit sale, because it lets you run the whole consideration loop on owned rails.

A captured, hashed email is the bridge that survives cookie loss. It powers enhanced conversions and list-based retargeting when the browser signal is gone. The owned list is both an acquisition channel and a measurement backstop. The plumbing is in The Signal Chain.

For ecommerce, the checkout. For services, the booking or enquiry. For subscriptions, the signup or trial. For lead gen, the form. The architecture of that page, block by block, is The Conversion Journey. A custom booking flow that converts and survives no-shows is its own planned field manual.

A conversion is an event you have to declare. If the platform never receives a clean, deduplicated conversion event, the entire map above runs blind: the algorithms optimise toward nothing and you cannot tell which channel worked. That is The Signal Chain.

A conversion is not a win until the unit economics say so. A 4x ROAS is profit on a 60% margin and a loss after refunds on a thin one. Whether any of the acquisition on this map was worth doing is decided in The Money You Actually Keep.

Lifecycle flows run on owned channels, so the marginal cost of another touch is near zero and the margin on the revenue is high. Abandonment, welcome, and post-purchase sequences are the workhorses: an abandoned-cart sequence routinely recovers a meaningful slice of otherwise-lost orders, and a multi-message sequence sharply outperforms a single send. This is acquisition you already paid for, finished properly.

Rising acquisition costs make retention the lever that protects the unit economics. If you keep customers longer and they buy again, you can afford to pay more to acquire them in the first place, which lets you outbid competitors on every channel above. Retention is not a downstream nicety; it sets the budget ceiling for the entire acquisition map. Lifetime value is the planned sequel to The Money You Actually Keep.

This is where the loop closes. A retained, delighted customer leaves a review, refers a friend, posts the unboxing, recommends you in the group chat. That is demand creation again (the most trusted kind, from Block 02), except you earned it instead of buying it. Every turn of the flywheel makes the next turn cheaper.

Advocacy can be prompted without being faked: review requests timed to the moment of delight, referral incentives, making the product worth photographing, asking. The output feeds straight back into word of mouth, dark social, reviews, and earned media at the top of the map. Build this and a share of your "untracked top" becomes a system you operate, not weather you wait for.

Every conversion sits in one of three buckets. Tracked: a clean, deterministic click path. Modeled: the platform lost the signal (consent denied, cookie expired, iOS opt-out) and filled the gap with a statistical guess. Invisible: dark social, word of mouth, podcasts, AI assistants, offline, that leave no signal at all and surface, if anywhere, as "direct." The modeled and invisible layers have grown every year as privacy tightened.

Direct traffic is treated as "people who already knew us." In reality it is a catch-all for anything without a referrer: a link from a chat app, a stripped UTM, a visit from an app, a privacy-blocked source. A growing direct bucket is often evidence your untracked top is working, not evidence it isn't.

For considered and B2B purchases, the majority of the buying journey happens before any trackable contact: research in private communities, AI assistants, and internal conversations. By the time a trackable click appears, the decision is often most of the way made, by channels you never saw.

Last-click attribution gives all the credit to the final step before conversion, which is almost always a capture channel (branded search, direct, retargeting). It systematically overcredits the channels that harvest demand and starves the channels that create it. Run a business on last-click and you will defund your own top of funnel, then watch capture efficiency slowly fall as the demand it was harvesting dries up.

Multi-touch attribution (MTA) is better than last-click but still only sees tracked touches; it is blind to the invisible layer and degraded by signal loss. Data-driven attribution distributes credit with machine learning, but only across the touches it can see, and only with enough conversions to train on. Every click-based model inherits the same blind spot: it cannot credit what it cannot observe.

Add up the conversions each platform claims and the total routinely exceeds the orders in your bank. Different attribution windows, view-through counting, and independent modeling mean Meta, Google, and your analytics all claim the same sale. The backend is the only source of truth; the platforms are self-interested estimators. The reconciliation discipline is in The Signal Chain.