Bottom of funnel
Converting the demand you already have.
Seven programs for converting demand that already exists, and the discipline that asks the one question every warm-audience channel would rather skip: would this booking have happened without it.
Bottom of funnel is every program working on demand that already exists — someone who already found the business, already interacted with it, or already has some relationship to it. Seven programs run here — conversion rate optimization, marketing operations, organic capture, email marketing, retargeting, creative production, and partnerships — none of them trying to reach someone new, and none of them splitting credit across a long, undecided consideration window.
What makes measurement hard here
Every program on this page starts with an advantage no top-of-funnel or middle-of-funnel program has: a real, warm population already in hand. That advantage is also the measurement trap. A channel pointed at people who already visited the site, already opened an email, or already know the business by name will produce bookings almost no matter what it does, because that population was already closer to converting than a cold one before the program ever touched them. The honest question at this stage is never whether a booking happened after the program ran — a booking was frequently going to happen regardless. It is whether that specific booking would still have happened without the program at all, which is an incrementality question, not a volume one.
That question only gets answered by deliberately withholding a program from a comparable slice of the same warm population and watching what changes. It is a different failure mode from earlier stages: top-of-funnel measurement worries about a touch's evidence surfacing somewhere else, later; middle-of-funnel measurement worries about splitting credit fairly across several touches; bottom-of-funnel measurement worries about a channel claiming credit for a conversion it never actually caused, simply because it was aimed at someone who was already going to convert.
Consider a retargeting campaign shown only to people who abandoned a cart or visited a pricing page twice. Almost everyone in that audience was already a strong candidate to buy, so the campaign's reported return will look excellent even in the version where the ads changed nothing at all — a purely mechanical consequence of who the audience already was. A branded search ad, an email to an existing customer list, and a partnership commission on a referred sale all share that identical structural risk, dressed in different mechanics. Two of the seven programs on this page, marketing operations and creative production, sidestep the distortion entirely: their conversion definition is the other five programs' measurability and supply, a job that never touches a customer directly.
How these programs combine
Marketing operations is the plumbing every other program on this page depends on to be measured accurately at all — a lead-routing or deduplication problem there quietly corrupts the numbers retargeting, email, and organic capture all report. Retargeting and email both re-engage the same already-interested population through different channels, and frequency between the two needs active coordination so the same person is not chased by both at once. Organic capture and conversion rate optimization both work on demand that has already reached the business's own systems — a profile view or a landing page visit — and hand off to each other at the point a lead becomes a real conversation. Creative production supplies the volume of fresh assets that retargeting in particular consumes to avoid audience fatigue, and partnerships depend on the same reconciliation discipline marketing operations already applies to every other lead source.
Where to start
An operator who has never run a holdout on any bottom-of-funnel program starts with whichever one currently reports the best-looking return, since that is precisely the program most likely to be coasting on demand it never actually created. An operator whose leads are getting lost, duplicated, or misrouted starts with marketing operations instead, since nothing else on this page can be measured accurately until that gets fixed first. And an operator running retargeting and email at the same time, with no coordination between them, starts by simply comparing the two campaign calendars side by side — a surprising number of accounts discover they have been sending the same offer through both channels on the same day for months.
Programs here report through the same attribution layer as every other stage, and the demand converted here originated in the programs covered at top of funnel.
Conversion rate optimization built for the real traffic volumes multi-location operators actually see. Covers sample size discipline before a test launches, post-form intake review since most loss happens after the form fires, and honest reporting when a test ends inconclusive rather than a rerun with a friendlier threshold. Where page-level traffic is too low to test formally, qualitative review replaces a split test that could never reach significance.
Lifecycle email and SMS programs measured on incremental revenue rather than open or click rate, since Apple's Mail Privacy Protection now pre-fetches images regardless of whether a human ever opens the message. Covers segmentation by lifecycle stage, deliverability management, and holdout testing — a comparable segment that receives nothing during a test window — as the only method that shows what a sequence actually produced beyond what would have arrived anyway.
Branded and local organic capture — Google Business Profile management, review response, map pack position, and branded queries — for the moment a searcher has already decided to consider a specific location. Distinct from non-brand SEO, which earns attention from people who have not decided yet. Covers what Google's own profile data can confirm directly and what has to be estimated, since most profile actions happen entirely on Google's surface and never touch a business's own analytics.
Retargeting built around the one question most programs never answer: whether the campaign changed any outcome at all, given that the audience it targets was already the most likely group on the internet to convert regardless. Covers audience definition, frequency capping, exclusion logic, and a holdout comparison as the only test that can tell a real effect apart from an audience that was going to buy anyway. States plainly that most retargeting reports overstate their own return.
Direct, negotiated partnerships between two businesses, built around a written attribution agreement signed before the relationship launches. Both sides of a partnership can plausibly claim the same conversion, and both are reporting it to their own leadership — without agreed rules in writing up front, the two reports disagree and the relationship absorbs the argument. Covers referral tracking that works by phone and in person, not just online, and reconciliation against the agreement rather than a fresh negotiation each time.
Producing the assets a testing plan actually calls for, at the volume, in the platform specs, and on the turnaround the plan requires. This page has no conversion metric of its own — its output is an input to the programs running on top of it, and its failure mode is a stalled testing schedule rather than a bad performance number. The hypotheses produced against come from creative ads strategy, a separate discipline with its own page.
Marketing operations consulting covering lead routing, deduplication, field mapping, and stack consolidation — the plumbing that determines whether any other program's numbers can be trusted. Framed plainly as foundational infrastructure rather than a headline metric of its own, since its success is measured by whether every other program's reporting holds up, not by a conversion count belonging to this page.
Find out how much of your bottom-of-funnel spend is doing real work.
A holdout test on one bottom-of-funnel program, showing the gap between what it reports and what it actually produced.