Alternatives to enterprise media planning platforms: five classes, compared honestly

The alternatives to enterprise media planning platforms fall into five classes: full enterprise suites, spreadsheets with in-house macros, sales house tools, research vendor desktop software, and focused modern tools that do one layer to an inspectable standard. We sort by class, not brand. What decides the purchase is which calculations you can reproduce exactly, on demand, in front of a client.

Buyers land here from three directions. The contract covers far more than the part of the product they actually use. A renewal is coming and procurement wants options. Or one calculation disagrees with what their buyers do by hand. All three end at mechanics.

Price first. Enterprise media planning software contracts are negotiated, and vendors publish no rate cards. That "$100–500K per year" band circulating online is unverified market chatter — check it against your own contract and quotes. We will not pretend to know what you pay.

What are the five classes of alternative?

1. Full enterprise suites. One contract covering planning, buying, reconciliation and reporting on a shared reference data layer. The strength is real: a single dictionary of channels, dayparts and advertisers across every team. The cost is coupling. You buy the whole stack to use three modules, and the calculation logic sits in a box you cannot read line by line.

2. Spreadsheets plus in-house macros. Still widely used, often excellent. A senior buyer's workbook usually encodes market knowledge no vendor ever shipped. The weakness sits elsewhere: the logic lives in one file owned by one person, and six months later nobody can prove a number to a client. Worth the qualifier — that is a property of undocumented process, not of a region. Mature markets with a developed measurement ecosystem formalize, audit and tool their planning workflows. The open question there is reproducibility under scrutiny.

3. Sales house and broadcaster tools. Free or near-free, and authoritative on that seller's inventory and pricing structure. They are not neutral, and they were never built for cross-seller comparison, which is what a plan needs.

4. Research vendor desktop software. Built around audience measurement: strong on reach, frequency and demographic analysis. Usually desktop-bound, licensed per seat, and detached from the money layer, so the budget arithmetic happens somewhere else anyway.

5. Focused modern tools. Narrow scope, one job done to an inspectable standard, connected to the rest by exports and APIs. This class grew out of the coupling problem in class 1. It also fails hardest when the vendor is vague about mechanics.

Which four questions actually separate them?

Class-level descriptions do not close a decision. Ask any vendor, in any class, these four.

How do sales house discounts combine? They multiply; they do not add. A 20% discount and a 10% discount give a combined 28%, not 30%.

A tool that adds them overstates your buying power on every plan, and each extra discount line widens the gap. Then the follow-up: the budget sets the discount tier, the discount moves the budget — how does that loop resolve? We iterate to convergence, and when the value oscillates between two tiers we take the higher budget. That is deliberately conservative for the buyer: a plan that under-promises delivery is survivable, one that over-promises is not.

What does the prime-time uplift field apply to? Plenty of implementations silently reprice the whole deal, prime and off-prime alike. Defensible as a convention, but it has to be stated out loud, because a planner who assumes otherwise misprices systematically. Our full chain:

Every element in that chain should be visible to the person defending the plan.

Is off-prime a share of money or a share of ratings? Different input, different plan. We take off-prime as a share of money and convert it to a share of ratings using affinity and the off-prime discount, because the negotiation happens in currency. Two more bits of arithmetic worth checking in any media buying software: wGRP = wTRP ÷ (affinity ÷ 100), so affinity 200 means your target watches twice as densely as the general population; and spot length is weighted by coefficient, 30s = 1.0, shorter lengths worth less.

Where does exact age come from? This one quietly decides whether your demographic breaks are real. Text exports from measurement systems commonly carry an age band of 1–5. Industry breaks like 18–49 or 25–54 cannot be assembled from bands — the boundaries fall inside them. Blame the text export, not the measurement: the raw delivery files carry precise age, in our case 4–65. TV Planner reads the raw binary delivery files (EVS/RDS/RSP) instead of the text export, which is why the breaks hold.

How do you prove a tool is right, in any class?

Accuracy claims are cheap. Ask for the evidence format instead.

On the money layer, the test is parity against a reference. We rebuilt an agency's own reference workbook, drove the difference below 0.01%, then froze that behavior behind roughly 750 backend and 2,100 frontend tests, so changing one input cannot quietly move an unrelated number.

On the audience layer, the test is cell-by-cell reproduction. We rebuilt an official audience table: of 60,858 cells, all 21,548 non-zero cells matched within 0.5 of a person — the rounding tolerance, since the published table rounds to whole people — across 17 of 17 demographic cuts. The wording matters. Nielsen is our measurement partner, and what we state is exactly this: we reproduce Nielsen's numbers. That is the whole claim, and it is the one we can put on a screen next to the published table.

Three more properties worth demanding, because they are invisible in a demo and painful in production:

• Multi-day aggregation weighted by universe, not an arithmetic mean of daily percentages. Averaging percentages across days is a common and silent error. Reach and frequency belong at the individual level — ours runs on a universe of 2,295,613 people.
• Idempotent file intake. Deliveries get re-sent and corrected. Ours are keyed by SHA-256: a repeat delivery is skipped, a corrected one deletes and replaces the old rows. No silent duplicates inflating a week.
• No invented defaults. If an input is missing, the system fails with the name of the field instead of substituting a plausible value. What you see is what was computed. Same spirit: active months come from actual TRP, not from what the planner typed.

Two more things separate serious tools from demoware. Flight shaping: if a planner is typing 52 weeks by hand, the tool is not helping. We ship 17 flight templates — Flat, Front-Loaded, Mid-Peak and Crescendo among them — with 5 intensity levels from 0.5 to 1.7, and the shapes trace back to published work on advertising dynamics: Broadbent's adstock at roughly 2.5 weeks for FMCG, Jones's STAS, Binet and Field. Post-buy: "it aired" is not a result. Each aired spot should land in exactly one mutually exclusive status against the plan — we use seven (matched, out_of_flight, unplanned_channel, out_of_week, out_of_daypart, wrong_length, no_lines) — and the client should get a frozen snapshot of the report on a revocable link, never a spreadsheet that shifts under them.

Where we fit, and what we do not do

Tilsim sits in class 5. TV Budgeting handles the money layer — two-way TRP-to-budget and budget-to-TRP conversion on one frozen core, with the Shop List (the budget declared to the seller, which sets the discount tier) and Execution (what was actually run) kept as two separate money layers instead of one blended figure. TV Planner handles the audience layer on raw delivery files. They connect; neither pretends to be the whole stack.

The honest limits, stated up front because they belong in your comparison table: our CPP is a proxy — the data contains no real prices, so we make no claim about actual cost, ROI or money saved. Forecasting and autonomous planning are on the roadmap, not in the product. Audience data currently covers one market, Moldova. If your decision turns on any of those three, better to know it before a demo than during one.