What Can You Honestly Measure in the First Ninety Days of a Marketing Programme?

A programme starts on the first of the month. Ninety days later somebody opens a spreadsheet and asks whether it worked. Nobody wrote down in week one what “worked” would look like in week thirteen, so the review turns into a negotiation about which number counts. That meeting is avoidable. The thing that avoids it takes about two hours, at the start, and almost nobody does it.

Most ninety day marketing measurement fails for an unglamorous reason: the numbers that move quickly are not the numbers anyone was promised, and the numbers that were promised run on a slower clock than a quarter. Sessions move. Qualified pipeline usually does not. If nobody says that out loud in the first week, it gets discovered in the thirteenth by the person who approved the budget.

So here is the position this piece defends. A first quarter should be judged on leading indicators and on work completed, not on revenue. That sounds like an escape hatch and it is the opposite of one: completed work is far easier to hold a team to than a revenue line with six other inputs feeding it. A measurement plan that names its own limits is much harder to argue with than one that promises everything and then reports selectively.

What follows is the split we use: what genuinely moves in a quarter, what cannot, the phased plan that gets you from one to the other, and the conversations that belong before day one rather than after day ninety.

What ninety day marketing measurement is really for

Ninety days is not a results window. It is a learning window with a results window bolted onto the end, and the two get confused because they share a review meeting.

In a first quarter you are trying to answer three questions. Does the demand you assumed exists actually exist? Do the channels you picked reach it at a cost that is not absurd? And does the site turn attention into something a salesperson can use? Each of those is answerable inside ninety days. None of them is the same question as “did we hit the number”.

Which is why the most useful sentence in a first-quarter review is usually here is what we now know that we did not know in January. It reads as softer than a revenue figure. It is not. It is the thing that decides whether the next two quarters are spent well or spent again.

What genuinely moves inside ninety days

Some things respond to work almost immediately, because they are not really market outcomes at all. They are outputs. You control them, so they change when you change them, and there is no honest reason for them to sit unmoved at day ninety.

Things you control directly

Page speed on the templates that carry traffic. The number of URLs that return something useful rather than a soft 404. Whether the pricing page says what the pricing is. Whether the tracking fires on the events you claim to be tracking. Whether a form asks for eleven fields when it needs four. These are build tasks with deadlines, and they belong early in a first-quarter plan precisely because nothing about them waits for the market to notice you.

The response path counts here too. If a visitor with a question has exactly one route, a form that gets answered in two days, then a share of genuine interest is being discarded for reasons that have nothing to do with marketing. Adding an on-site way to ask and be answered is a build task rather than a market outcome: a widget like Flidu puts website-informed answers alongside the contact and conversion actions in one place, and whether it earns its keep is measurable well inside the window. So is the smaller question of how many fields a contact form should have.

Signals that respond quickly

Paid channels report within days, which is why they are over-represented in first-quarter reviews. Click-through, cost per enquiry, and the shape of the search terms you are actually matching all give a usable read within a few weeks. Email behaves similarly: reply behaviour on a list you already own tells you something about message fit before any acquisition work has landed anywhere.

Branded search interest is the sleeper signal. It moves slowly in absolute terms and quickly in relative terms, and it is one of the few early numbers that indicates whether anything you did reached a human who then went looking for you by name. Watch it weekly. Do not build a forecast on it.

What cannot move in a quarter, and how to say so in week one

Now the harder half. Some outcomes run on a clock you do not control, and pretending otherwise is how a sound programme gets cancelled in month four.

• Competitive organic rankings. A new page on a site with little history does not out-rank an established one inside a quarter because you wrote it well. Discovery, indexing and evaluation each take their own time, and that is before anything competitive happens.
• Anything gated by a sales cycle. If a deal takes five months to close, ninety days of activity produces pipeline, not revenue. Revenue reported in that window mostly measures work done before the programme started.
• Retention and repeat purchase. A cohort has to live long enough to be a cohort. Ninety days gives you the first point on a curve that only means something once it has three or four.
• Demand for a category nobody searches. No amount of first-quarter effort creates a query that does not exist, which is a different problem with a different playbook: marketing a product nobody is searching for starts somewhere else entirely.
• Anything downstream of a change that shipped in week ten. A change needs a run-up before it has a read. Work that lands late in the window is work you are measuring in the next one, and saying so protects it from being judged early.

The organic case deserves an extra sentence, because it is where expectations break most often. Search work in a first quarter is mostly the removal of reasons not to rank. The crawl, render and index sequence has to complete before ranking is even a question, and a site carrying real technical debt can spend its whole first quarter on that sequence. That is progress. It simply does not look like progress on a traffic chart, which is why it needs describing in words.

A phased plan for the ninety days

The plan below is deliberately front-loaded. Anything that needs a run-up has to ship early, and anything that only reports has to exist before there is something worth reporting.

1. Days 1 to 15: instrument and baseline. Fix tracking, agree definitions, and record where every headline number stands before anything changes. A baseline captured in week six is not a baseline, it is a comparison against work already underway.
2. Days 10 to 30: ship what you control. Speed, broken paths, thin pages that carry real demand, the form, the response route. These changes have the longest tail, so they go first rather than last.
3. Days 20 to 45: put spend where it reports fastest. Paid search and paid social buy you a read on message and offer in weeks rather than months, and that read then informs everything slower.
4. Days 30 to 75: build the assets that pay back later. Content, structured data, internal linking, the pages that answer questions people actually ask. These will not repay inside the window, and they are the reason the second quarter looks different from the first.
5. Days 75 to 90: read it, write it down, decide. Not a dashboard screenshot. A written read on the three questions from the top of this piece, with a recommendation attached to each one.

Four conversations that belong before day one

These take an afternoon. Skipping them costs a quarter.

• What does a good day ninety look like, in written numbers? Write the target and the tolerance around it. “More enquiries” is not a target, it is a mood, and a mood cannot be reviewed.
• Which number are we not going to move, and does everyone accept that? Name it early. The point is not to lower expectations, it is to move them onto a line that can actually respond.
• Who owns each number? Marketing owns enquiry volume and the quality signals attached to it. Sales owns what happens next. A shared number with no owner becomes nobody’s problem at exactly the point it needs to be somebody’s.
• What would make us stop? Agree the condition under which a channel gets cut, in advance. That decision is far easier to make before anyone in the room has a stake in defending their own idea.

There is a fifth conversation that only matters once the first four have gone well: how will you credit a result to a channel when several channels touched it? Last-click hands the credit to whichever channel happened to be standing nearest the door. We have written about why your best performing channel is probably being miscredited, and the first quarter is the cheapest possible moment to settle the rule, because nobody has anything invested in the answer yet.

Instrumentation is the first deliverable, not the last

Measurement built after the work is measurement built to flatter the work. It also misses the first six weeks, which are the weeks you will most want to look at when something surprising shows up later.

Name things once

Agree the event names, the campaign parameters and the definition of an enquiry before anything launches. One person writes the convention down, everybody follows it, and nobody spends day eighty-eight reconciling three spellings of the same campaign. This is dull work. It is also the difference between a report and an archaeology project.

Definitions that survive contact with sales

A “lead” means one thing in a marketing tool and something else in a sales conversation. Pick the definition the business already uses rather than the one the tool defaults to. If sales only calls something qualified after a discovery call, then your ninety-day number is discovery calls booked, and everyone can live with that because everyone agreed to it in advance.

The same discipline applies on-site. A quarter is rarely enough traffic for a clean split test on a low-volume business site, which is why conversion work without the traffic to test leans on qualitative evidence and known failure patterns instead of statistical significance. Treat that as legitimate conversion optimisation, report it honestly as such, and never dress a considered judgement up as a test result.

What the day ninety review should actually contain

Four things, in this order, and preferably written in prose rather than assembled as slides.

First, what was delivered, with dates. It is the least arguable part of the review and it should be read out even when it feels obvious, because it is the record that the quarter had substance in it.

Second, what the numbers did against the baseline, including the ones that went the wrong way. A review that reports only favourable movement teaches its reader to distrust the favourable movement as well.

Third, what you now know. The three questions again: is the demand there, do the channels reach it, does the site convert it. Answer each with evidence and with a confidence level you are willing to defend in front of someone sceptical.

Fourth, the recommendation: continue, adjust or stop, and why. If the answer is that the organic work compounds from here, say exactly what compounds and when you will check it. Search work has a payback curve that begins after most first quarters end, and the honest version of that sentence persuades better than a hopeful chart does.

A first quarter can only be judged fairly against the question it was able to answer. Choose that question on day one, in writing, with the person who will read the review.

Our parent company has written a companion piece on the same window from a slightly different angle, what ninety days can honestly show, and the two agree on the part that matters most: the expectation is either set at the start or argued about at the end.

When ninety days is the wrong window entirely

Here is the concession. There are programmes where a ninety-day review is not merely early, it is actively harmful, and running one anyway does damage that takes longer than a quarter to undo.

A new category, a long enterprise sale, a brand rebuild, a technical rescue on a site with years of accumulated debt: in each of those, the first quarter produces very little a chart can hold. Reviewing it as though it should produces pressure to do visible things instead of useful ones. That is how a rescue quietly turns into a campaign, and how the underlying problem survives into year two.

The fix is not to skip the review. It is to change what the review is for. Run it on delivery and diagnosis, state plainly that outcome measurement begins at month four, and put the outcome review in the calendar immediately so that nobody has to ask for it later.

The rule we would take into a planning meeting is one line: measure the fastest thing that would genuinely change your mind. Not the fastest thing available, and not the most flattering one. If a number moving would not change a single decision, it is decoration, and reporting it crowds out the number that would. If you are setting up a first quarter and cannot yet name the numbers you will be able to defend at the end of it, tell us what you are planning and what has already been promised to whom.