When a spreadsheet is still the right answer
A marketing platform is an operating cost with a person attached, and a meaningful share of teams buy one while their actual need is a list, a calendar reminder and someone writing better emails. The platform will not fail; it will simply sit at a low level of use while the licence renews, which is a worse outcome than not buying it, because it also absorbs the attention that the manual process needed.
What does a platform cost beyond the licence?
An owner, permanently. Somebody has to maintain the sequences, the field mapping, the consent state, the templates and the checks that mail is still arriving, and that work does not shrink for a small list. It is the same set of tasks whether you have two thousand contacts or two hundred thousand, which is why the cost per useful outcome is worst at the small end.
There is also an implementation cost that is mostly other people's time: agreeing field definitions, cleaning enough data to launch, connecting the CRM, and rebuilding the templates. This lands on the people who are already the constraint, usually the same person who writes the content.
And there is an opportunity cost that nobody records. Months spent configuring are months not spent talking to customers or rewriting the emails that were underperforming, and the emails were nearly always the actual problem.
What can you do manually that works surprisingly well?
A list with a status column, a saved reply set, and a recurring calendar block. Sent individually from a real mailbox, these messages arrive with better deliverability than any bulk send, get higher reply rates because they are genuinely personal, and cost nothing but the time of the person sending them.
The status column is doing the work the platform would do. It records where each person is and when they were last contacted, which is the bookkeeping function that fails first as volumes grow. A shared sheet with a defined set of statuses and a last-contacted date is a legitimate system, not a placeholder, for as long as one person can hold the whole picture.
The one thing to do properly from the start is consent. Record what each person agreed to, when, and how, because that record is what you migrate later, and reconstructing it retrospectively is impossible.
| Signal | Spreadsheet still fine | Time to automate |
|---|---|---|
| Repeated messages per month | A few dozen, sent by one person | Hundreds, or more than one person sending |
| Response time to an enquiry | Consistently within the hour | Varies from minutes to days by who is available |
| Memory of the pipeline | One person can recall the state of every active contact | Nobody can answer without checking two places |
| Consent and preferences | One list, one basis, opt-outs handled by hand | Multiple subscription types or regions with different rules |
| Sequence complexity | One or two steps with no branching | Several flows, running at once, with exit conditions |
| Failure consequences | A missed follow-up embarrasses one person | A mistake reaches the whole list at once |
Which signals mean you have genuinely outgrown it?
Four, and any one of them is sufficient. Volume, where the sending itself now takes hours a week that a person is visibly not doing. Latency, where responses depend on who happens to be at their desk. Memory, where nobody can say what a given contact has received. And consent complexity, where a single opt-out column no longer describes what people have agreed to.
Consent complexity is the one that arrives soonest for organisations selling into more than one region, and it is the one where the manual approach genuinely creates risk rather than just inefficiency. If people can subscribe to different things, or if you need a preference centre, that is a system requirement rather than a preference.
Notice that database size is not on the list. A large dormant list is not a reason to automate; it is a reason to work out how much of the list is real, which is a different project and usually a shorter one.
What is the middle option?
Sequencing inside the CRM you already have. Most CRMs now send templated sequences to modest lists, track opens and clicks, and handle the immediate response to an enquiry, which is the highest value automation available to most teams. It never costs a second identity model, though it often does cost more than the seat you already pay for: HubSpot gates Sequences behind Sales Hub Professional at roughly 100 USD per seat per month plus onboarding, Salesforce puts multi-step cadences in Sales Engagement as a paid add-on or a higher tier, and Pipedrive sells Campaigns as a separate add-on. Check your own tier before assuming it is included.
This works until sending volume starts affecting the deliverability of ordinary business mail, since CRM sequences typically send from your main domain. That is the specific point at which bulk marketing needs its own infrastructure, and it is a much clearer trigger than any feature comparison.
The other middle option is a simple sending tool with a list and no CRM integration at all. It is limited, it will not score anyone, and for a newsletter it is entirely sufficient. The temptation to buy something that could do more later is what produces the underused platform.
How do you keep a manual process from becoming a mess?
Constrain it. One sheet, not several copies emailed around. A fixed list of statuses that everyone uses, rather than free text. A last-contacted date that is filled in every time. An opt-out column that is checked before any send and never overwritten by an import.
Then set an explicit review date, perhaps twice a year, where you check the four signals above against reality. Manual processes fail by drifting past the point where they should have been replaced, and the drift is invisible from inside because each week is only slightly worse than the last.
Keep the sending out of a personal mailbox where possible, or at least out of a personal address that leaves when the person does. Continuity is the quiet weakness of manual processes, and it is the failure that costs the most when it happens.
What should you take with you when you do migrate?
The consent record, the status definitions and the message library. Those three are the actual value your manual process built, and all three are portable. What you should not carry over is the accumulated list of addresses nobody has contacted in years, because importing it is the fastest way to damage the sending reputation of a new platform on its first send.
Migrate the process you are already running rather than designing a new one. A team that automates its existing sequence knows what good looks like and can tell immediately if the automated version is worse. A team that designs an ambitious new programme during implementation has no baseline and usually cannot say afterwards whether the platform helped.
Then keep the sheet for a month and reconcile it against the platform. It is the only independent record you will ever have of what the system should be doing, and comparing them once catches the mapping errors that otherwise take a quarter to surface.
Common questions
- When do you actually need marketing automation software?
- When any one of four things is true: sending repeated messages takes hours a week that nobody is doing, response times vary with who is at their desk, nobody can say what a given contact has already received, or consent has become complex enough that a single opt-out column no longer describes what people agreed to. Database size alone is not a reason, since a large dormant list is a different problem.
- Can you run lead nurturing from a spreadsheet?
- Yes, up to the point where one person can hold the whole picture. A shared sheet with a fixed set of statuses, a last-contacted date and a consent column does the bookkeeping a platform would, and messages sent individually from a real mailbox get better deliverability and higher reply rates than any bulk send. Record the consent basis properly from the start, since it cannot be reconstructed later.
- What does marketing automation cost beyond the subscription?
- An owner, permanently, plus implementation time from the people who are already the constraint. Maintaining sequences, field mappings, consent state, templates and delivery checks is roughly the same workload whether you have two thousand contacts or two hundred thousand, so the cost per useful outcome is worst for small teams. The opportunity cost of months spent configuring instead of rewriting the emails is rarely recorded.
- Is CRM email good enough instead of a marketing platform?
- For many teams, yes. Most CRMs send templated sequences to modest lists and track opens and clicks, which covers the immediate response to an enquiry, the highest value automation available to most teams, and it needs no second identity model. It is not always free of an extra licence: sequencing is tier-gated in several major CRMs, including HubSpot Sales Hub Professional, the Salesforce Sales Engagement add-on and Pipedrive Campaigns. The other limit is deliverability: CRM sequences usually send from your main domain, so growing bulk volume starts to affect ordinary business mail.
- What should you migrate into a new marketing platform?
- The consent record, the status definitions and the message library, since those are what the manual process actually built. Leave behind addresses nobody has contacted in years, because importing them is the fastest way to damage a new platform's sending reputation on its first send. Automate the process you already run rather than designing a new one, so you have a baseline to compare against.
